Diabetic Retinopathy Screening: Current Screening Practices and … · 1" " Diabetic Retinopathy Screening: Current Screening Practices and Novel Retinal Imaging Modalities By Daniel

1

Diabetic Retinopathy Screening: Current

Screening Practices and Novel Retinal Imaging

Modalities

By Daniel Shu Wei TING MBBS (1st Class Honours) MMed (Ophth)

20799514

A collection of papers presented for the Degree of Doctor of Philosophy to the Centre

of Ophthalmology and Visual Science, University of Western Australia

June 2015

Coordinating Supervisor: Associate Professor Mei-Ling Tay Kearney

Centre of Ophthalmology and Visual Science, Lions Eye Institute, Perth

External Supervisor: Professor Yogesan Kanagasingam

Commonwealth Scientific Industrial and Research Organisation

Research Director of Australian e-Health Research Centre

2

PREFACE

All presented manuscripts relate to the current practices and novel imaging

technologies for diabetic retinopathy screening. The thesis is divided into four

sections. Section 1 consists of the introduction and literature review. Section 2

evaluates the current diabetic retinopathy screening practices and attitudes of

Australian optometrists (Chapter 1) and general practitioners (Chapter 2). Section 3

validates a novel diagnostic device (Chapter 3) and optimal reading screen size

(Chapter 4) for retinal still photography. Section 4 proposes and validates a novel

video-based screening technology using retinal video recording (Chapter 5) and

compression (Chapter 6) for diabetic retinopathy screening. Each paper is presented

with the original internal headings, figures and tables; however, for ease of reading

and flow, the thesis has been formatted uniformly.

For all chapters, I was responsible for the conceptualization, design, development,

data collection, analysis, writing and presentation. Other co-authors have provided

guidance, supervision, assistance and proofreading of the thesis. For Chapters 1 and 2,

the Eye and Vision Epidemiology Research Group (EVER) of the University of

Western Australia helped with the design and conceptualization of the research

project. The contribution of others is presented below and at the end of each chapter.

3

DECLARATION

I, Daniel Shu Wei Ting, submitted in fulfilment of the requirements for the award of

Doctor of Philosophy in the Centre of Ophthalmology and Visual Science of the

University of Western Australia. To the best of my knowledge and belief, the

publications and the work arising from this thesis represent the original work of the

author. This thesis contains no material which has been accepted for the award of any

other degree of diploma in any university.

4

ACKNOWLEDGEMENTS

I would like to express my sincere thanks to my supervisors, Associate Professor Mei

Ling Tay-Kearney and Professor Yogesan Kanagasingam. Without their ongoing

support, this thesis would not have been possible.

Associate Professor Tay-Kearney kindly dedicated her precious time for most of my

studies over the past two years. Her tireless guidance and ongoing encouragement has

further enhanced my clinical and research interest in ophthalmology.

Professor Yogesan Kanagasingam has always been an extremely supportive and

approachable supervisor. Despite having various titles, awards and publications under

his belt, he is always humble, down to earth and readily available for questions.

Without his ongoing encouragement and prompt replies to all of my enquiries

throughout my research period, I would not have been able to publish and present my

research findings at different conferences, both locally and internationally.

I would also like to thank the members of the Eye and Vision Epidemiology Research

Group (EVER)—Associate Professor Nigel Morlet, Associate Professor David Preen,

Dr Jonathon Ng, Dr Antony Clark and Dr Joshua Yuen—for assisting me with the

design and conceptualization of Chapters 1 and 2. They constantly gave me statistical

advice and comments on the articles. The support from Professor Jill Keefe and

Professor Hugh Taylor (Centre for Eye Research Australia) is also very much

appreciated.

My sincere gratitude goes to the Australian E-Health Research

Centre/Commonwealth Scientific Industrial and Research Organisation (CSIRO) for

granting me scholarships and funding for the OIS EyeScan device. I would also like

5

to thank the Royal Perth Hospital Medical Research Foundation for additional

funding over the first 12-month period.

I am indebted to Janardhan Vignarajan, the software engineer from CSIRO who

constantly and patiently gave me technical support for the EyeScan device, retinal

images/videos processing and storage. He also enlightened me with his computing

knowledge throughout my entire research period.

Last but not least, I would like to thank my family members (dad, mum and my two

brothers, Darren Ting and Derrick Ting), who have given me endless moral support.

My wife, Celene, is currently an obstetric and gynecology resident, and she has

played a big part in my life during this period, despite her busy routine. She has

accompanied me through all of the ups and downs and never failed to offer me

invaluable advice. Without her ongoing support, it would not have been possible for

me to successfully complete my research on time while also completing my clinical

training.

6

PUBLICATIONS

1. Ting DSW, Ng J, Morlet N, Yuen J, Clark A, Taylor H, et al. Diabetic

retinopathy management by Australian general practitioners. Aust Fam Physician

2011;40(4):233–8.

2. Ting DSW, Ng J, Morlet N, Yuen J, Clark A, Taylor H, et al. Diabetic

retinopathy management by Australian optometrists. Clin Experiment

Ophthalmol 2011;39:230–5.

3. Ting DSW, Ng J, Morlet N, Yuen J, Clark A, Preen DB. Differences in diabetic

retinopathy management by primary eye care providers in Australia. Clin

Experiment Ophthalmol 2011 Aug;39(6):585–6. doi: 10.1111/j.1442-

9071.2010.02489.x.

4. Ting DSW, Tay-Kearney ML, Lim L, Constable I, Preen DB, Kanagasingam Y.

Retinal video recording: A new way to diagnose and image diabetic retinopathy

screening. Ophthalmology 2011;118:1588–93.

5. Ting DSW, Tay-Kearney ML, Constable I, Vignarajan J, Kanagasingam Y.

Retinal video recordings at different compression levels: A novel video-based

imaging technology for diabetic retinopathy screening. Eye (Lond) 2013 Jul;27(7):

848–53. doi: 10.1038/eye.2013.53. Epub 2013 May 10.

6. Ting DSW, Tay-Kearney ML, Lim L, Constable I, Preen D, Kanagasingam Y.

Light and portable novel device for diabetic retinopathy screening. Clin

Experiment Ophthalmol 2012 Jan–Feb;40(1):e40–6. doi: 10.1111/j.1442-

9071.2011.02732.x. Epub 2011 Dec 23.

7. Ting DSW, Tay-Kearney ML, Vignarajan J, Kanagasingam Y. Diabetic

retinopathy screening: Can the viewing monitor influence the reading and grading

outcomes. Eye (Lond) 2012 Dec;26(12):1511–6. doi: 10.1038/eye.2012.180.

Epub 2012 Oct 12.

7

INVITED BOOK CHAPTERS

Ting DSW, Kanagasingam Y, Constable I, Tay-Kearney ML. Video imaging

technology: A novel method for diabetic retinopathy screening.

CONFERENCE PRESENTATIONS

1. A novel video-based imaging technology for diabetic retinopathy screening.

Singapore Malaysia Ophthalmology Scientific Congress, Singapore (2012).

2. Validation of a new alternative diabetic retinopathy screening method: Retinal

video recording. 26th Asia Pacific Academy of Ophthalmology Conference,

Sydney, Australia (20–24 March 2011). Free paper session, primary presenter.

3. Evaluation of the optimal compression level for retinal video recording in the

setting of diabetic retinopathy screening. 26th Asia Pacific Academy of

Ophthalmology Conference, Sydney, Australia (20–24 March 2011). Free paper

session, primary presenter.

4. Video-based imaging technology: A novel method for diabetic retinopathy

screening. 45th Singapore Malaysia Congress of Medicine, Singapore (22–23

July 2011). Free paper presentation, primary presenter.

5. The optimal screen sizes of reading devices for diabetic retinopathy screening.

45th Singapore Malaysia Congress of Medicine, Singapore (22–23 July 2011).

Poster presentation, primary presenter.

6. Retinal digital videos at different compression levels for diabetic retinopathy. 45th

Singapore Malaysia Congress of Medicine, Singapore (22–23 July 2011). Poster

presentation, primary presenter.

8

7. A portable multipurpose ophthalmic imaging device for diabetic retinopathy. 45th



8. Australian national survey: Diabetic retinopathy screening by community. 45th



9. Australian national survey: Diabetic retinopathy screening by general practitioners.

45th Singapore Malaysia Congress of Medicine, Singapore (22–23 July 2011).

Poster presentation, primary presenter.

10. An economical portable device for diabetic retinopathy screening. American

Academy of Ophthalmology Conference, Chicago, IL, United States (16–19

October 2010). Poster presentation, primary presenter.

11. Australian national survey of diabetic retinopathy management among general

practitioners and optometrists. World Ophthalmology Congress, Berlin, Germany

(5–9 June 2010). Poster presentation, primary presenter.

12. Are we neglecting visual neglect? World Ophthalmology Congress, Berlin,

Germany (5–9 June 2010). Oral presentation, co-author.

13. Screening for diabetic retinopathy by Australian optometrists. Annual Scientific

Congress of Royal Australian New Zealand College of Ophthalmologists,

Brisbane, Australia (14–18 November 2009). E-poster presentation, primary

presenter.

9

PUBLIC MEDIA RELEASES

Retinal Video Recording: A New Way to Image and Diagnose Diabetic Retinopathy 1. Web Release:

• Review of Ophthalmology: Video Screening for Diabetic Retinopathy: A new

approach to detection appears to be just as effective as traditional methods, but

easier to use—even by non-medical staff (Appendix 1)

o URL: http://www.revophth.com/content/d/technology_update/c/44733/

• Advanced Ocular Care: Retinal Videos May Be New Option for DR Screening

(Appendix 2)

o URL:

http://bmctoday.net/advancedocularcare/2011/09/article.asp?f=retinal-

videos-may-be-new-option-for-dr-screening (September 2011)

• Retina Today: Retina Videos May be An Option for Diabetic Retinopathy

Screening

o URL: http://bmctoday.net/retinatoday/2011/08/article.asp?f=retina-videos-

may-be-an-option-for-diabetic-retinopathy-screening (August 2011)

• Reuters Health (The Doctor’s Channel Daily Newscast, New York): Retinal

Video Recording: A Novel Option for Diabetic Retinopathy Screening

o URL: http://www.youtube.com/watch?v=5VD5zKfDYfQ (July 1, 2011)

• Top story in Medscape Ophthalmology: A New Screening Tool for Diabetic

Retinopathy (July 13, 2011)

o URL: http://www.medscape.com/viewarticle/745710?src=mp&spon=36

(Accessed July 28, 2011)

• Asia Pacific Academy of Ophthalmology Congress (via Virtual Medical Centre):

Retinal Video Recording Could Be Used in Diabetic Retinopathy Screening

o URL: http://www.diabetic-retinopathy.org/2011/04/retinal-video-

recording-could-be-used.html (April 4, 2011)

• ScienceNetwork Western Australia: New Imaging Technique to Boost Diabetic

Retinopathy Detection

o URL: http://www.sciencewa.net.au/health-and-medicine/new-imaging-

technique-to-boost-diabetic-retinopathy-detection.html (Accessed April 16,

2011)

10

2. Press Release

o West Australian: Video Spots Eye Disease in Diabetics (March 23, 2011)

3. Television Release

o Channel 10 in Perth, Brisbane, Sydney, Adelaide and Melbourne: Video

Helps Find Sick Eyes

o URL: http://ten.com.au/ten-news-perth.htm?movideo_m=98103 (Accessed

March 28, 2011)

11

NOVEL CONTRIBUTIONS

The novel contributions of this thesis are outlined below.

1. Two large-scale national surveys on current diabetic retinopathy (DR) screening

practices and management by primary eye care providers—Australian GPs and

optometrists.

2. Validation of a novel, portable and economical fundus camera that can be

utilized to screen for DR screening in the primary health care setting.

3. Validation of portable, small reading devices to interpret color retinal still images

for DR screening.

4. Validation of a novel diagnostic modality for DR screening—Video-based

Imaging Technology (retinal video recording and compression) for DR screening:

a. This is the first study worldwide to evaluate the use of retinal video

recording in screening for DR.

b. This study has had significant media coverage nationally and

internationally:

1) Reuters’ Health, The Doctor’s Channel Daily Newscast, New

York (URL: http://www.youtube.com/watch?v=5VD5zKfDYfQ).

2) Top story in Medscape Ophthalmology, July 13, 2011.

3) Published in the top Ophthalmology journal, Ophthalmology, in

2011.

4) Televised on Channel 10 in Perth, Brisbane, Sydney, Adelaide

and Melbourne.

5) Reported in multiple e-magazines and newspapers, including

Retina Today, Review of Ophthalmology and ScienceNetwork

Western Australia and West Australian).

12

ABSTRACT

Diabetes is a metabolic disease that is rapidly increasing in prevalence. One in four

people with diabetes will develop diabetic retinopathy (DR). Therefore, it is important

for primary eye care providers, such as optometrists and general practitioners (GPs),

to actively participate in screening services by performing dilated fundoscopy or

retinal imaging. Retinal cameras are generally expensive and technically challenging

to operate; thus, they are often not readily available in the primary health care setting,

especially in rural areas. The aim of this thesis is to evaluate the current DR screening

practices and attitudes of primary eye care providers and validate some novel, cost-

effective and easy-to-operate imaging technologies as alternative screening tools that

can potentially increase their interest and desire in DR screening.

Section 1 evaluates the current management practices of Australian optometrists

(Chapter 1) and GPs (Chapter 2). A total of 3,000 self-administered questionnaires

consisting of questions related to screening practices/attitudes and hypothetical

clinical scenarios were mailed out to optometrists (n=1,000) and GPs (n=2,000)

across Australia. The results showed that nearly 78% of optometrists reported having

a strong desire to screen for DR, compared to only 40% of GPs. The leading

screening barriers for GPs and optometrists were: 1) poor confidence in performing

direct ophthalmoscopy; 2) time limitations; 3) patients’ unpreparedness to drive; and

4) the fear of inducing angle-closure glaucoma. The use of a retinal camera was

shown to significantly increase optometrists’ confidence to detect DR changes.

Further research should focus on the identified barriers in order to allow early

detection of sight-threatening DR that requires prompt laser treatment to prevent

severe visual impairment.

13

In light of the barriers to screening for DR (poor confidence, time limitation and lack

of desire to screen for DR), the use of an affordable and user-friendly retinal camera

would be significant in improving GPs’ and optometrists’ interest in participating in

community screening services. Section 2 investigates the use of inexpensive and

portable screening and reading devices for retinal still photography. Chapter 3

validates a novel, economical (US$30,000) and easy-to-operate multipurpose

ophthalmic imaging device called EyeScan (Ophthalmic Imaging System (OIS),

Sacramento, US), which screens for DR, whereas Chapter 4 validates the use of the

MacBook Pro and iPad for interpreting retinal color still images. The results showed

that three-field 30° mydriatic retinal still photography captured by the EyeScan had

comparable sensitivity and specificity to FF450 plus (EyeScan—sensitivity: 92.1%,

specificity: 98.4%, FF450—sensitivity: 94.4%, specificity: 98.9%) in detecting any

grade of DR. The technical failure rate for EyeScan was not statistically different

from FF450 plus (8.8% vs. 7%, p>0.05). Compared to the 15-inch MacBook Pro and

the 27-inch iMac, the 9.7-inch iPad was found to have similar sensitivity and

specificity of more than 90% in detecting any grade of DR, increasing up to 100% for

sight-threatening DR grading. Hence, the EyeScan and the 9.7-inch iPad are

economical and effective devices that can potentially be utilized by primary eye care

providers for DR screening in the community.

Retinal still photography is by far the most common DR screening tool used

worldwide. Nevertheless, a good retinal image is always highly dependent on a

photographer’s skills and the patient’s compliance; thus, retinal still photography

often requires the expertise of experienced personnel. The aim of Section 3 was to

purpose a novel and easy-to-operate video-based imaging technology for DR

screening. This is one of the first prospective studies to validate the use of retinal

14

video recording as an alternative DR screening technique. This technique simulates

what is seen from the slit lamp examination, and it offers a greater field of retinal

view within a shorter timeframe compared to retinal still photography. To evaluate the

ease of use of this technique, it was performed by a medical officer with no previous

ocular imaging experience (after receiving a three-hour training session). The retinal

still photography was performed by an experienced retinal photographer. The results

showed that the sensitivity and specificity of retinal video recording were comparable

to traditional three-field (optic, macula and temporal view) 35° mydriatic retinal still

photography (retinal video recording—sensitivity: 93.9%, specificity: 98.5%; retinal

still photography—sensitivity: 92.4% and 98.5%) in detecting any grade of DR. The

technical failure rates for retinal video recording and retinal still photography were 7%

and 5.5% respectively, and both were not statistically significant.

In addition, Chapter 6 shows that retinal videos could be rapidly (25 seconds) and

significantly compressed from 500 MB to 20 MB with excellent sensitivity and

specificity (more than 90%) in detecting DR changes by both the ophthalmologist and

professional grader. Given that retinal videos are readily compressible and provide a

greater field of view within a shorter timeframe compared with three-field retinal

photography, the technique of image acquisition offers an easier alternative to

maximize the participation rate of primary eye care providers, such as optometrists

and GPs, in DR screening services.

In conclusion, this thesis reveals that the desire of optometrists and GPs to screen for

DR needs significant improvement, and more efforts should be directed towards the

identified barriers to DR screening. To improve the cost effectiveness and

convenience of screening services, primary eye care providers should consider

15

utilizing an economical retinal camera for retinal still photography with a portable

reading device (e.g., 9.7-inch iPad) to interpret retinal color still images. This thesis

also reports using retinal video recording as a novel, video-based imaging technology

that could potentially be utilized as an alternative DR screening modality among

primary eye care providers, as it possesses high diagnostic accuracy (sensitivity and

specificity) in detecting DR changes, and it can readily be compressed to a smaller

file size for storage and transmission. Future studies will be significant in exploring

this technique’s user-friendliness among experienced and non-experienced personnel,

as well as the cost-effectiveness, clinical-effectiveness and use of this technique in a

routine, mobile or tele-ophthalmology setting.

16

ABBREVIATIONS

AACG Acute angle-closure glaucoma

ACE Angiotensin-converting enzyme

ADA American Diabetes Association

AGE Advanced glycation end products

AVI Audio Video Interleave

BMES Blue Mountain Eye Study

CERA Centre for Eye Research Australia

CI Confidence interval

CSME Clinically significant macular edema

CWS Cotton wool spot

DCTT Diabetes Control and Complications Trial

DKA Diabetic ketoacidosis

DM Diabetes mellitus

DME Diabetic macular edema

DR Diabetic retinopathy

ETDRS Early Treatment Diabetic Retinopathy Screening

FDA Food and Drug Administration

FFA Fundus fluorescein angiogram

GAD Glutamic acid decarboxylase

GP General practitioner

Hex Hard exudates

HHS Hyperosmolar hyperglycemic stat

ICD Islet cell autoantibodies

IFG Impaired fasting glucose

IGF Insulin-like growth factor

IGT Impaired glucose tolerance

IRMA Intra-retinal microvascular abnormalities

LDL Low-density-lipoprotein

LEI Lions Eye Institute

MA Microaneurysm

17

MB Megabytes

MVIP Melbourne Visual Impairment Project

NEI National Eye Institute

NHMRC National Health and Medical Research Council

NICE National Institute for Clinical Excellence

NIDDM Non-insulin-dependent diabetes mellitus

NPDR Non-proliferative diabetic retinopathy

NVD New vessels on the disc

NVE New vessels elsewhere

OCT Optical coherence tomography

OGTT Oral glucose tolerance test

OIS Ophthalmic Imaging System

PEDF Platelet-derived growth factors

PDR Proliferative diabetic retinopathy

SD Standard deviation

STDR Sight-threatening diabetic retinopathy

T1DM Type 1 diabetes mellitus

T2DM Type 2 diabetes mellitus

UKPDS United Kingdom Prospective Diabetes Study

US United States

UWA University of Western Australia

UWF Ultra wide-field fundus imaging

VB Venous beading

VEGF Vascular endothelial growth factor

WA Western Australia

WESDR Wisconsin Epidemiologic Study of Diabetic Retinopathy

WHO World Health Organization

18

LIST OF TABLES

Table 1.1: International Clinical Diabetic Retinopathy Severity Scales and International Clinical Diabetic Macular Edema Disease Severity Scales8 ......................................................................................................... 37

Table 1.2: Classification of diabetic retinopathy into retinopathy stages (Wisconsin level)9 ....................................................................................... 38

Table 2.1: The current diagnostic criteria for diabetes, impaired fasting glucose and impaired glucose tolerance published by the American Diabetes Association11,12 ............................................................................................ 45

Table 2.2: Diagnostic criteria for DKA and HHS by the ADA39 ................................ 48 Table 3.1: Demographics of the optometrists responding to a survey on diabetic

retinopathy screening .................................................................................. 71 Table 3.2: Barriers to optometrists performing dilated fundoscopy ............................ 72 Table 3.3: Current optometrists’ management and attitudes to diabetes and

diabetic retinopathy ..................................................................................... 73 Table 3.4: Optometrists’ management of hypothetical clinical scenarios and

specific signs of diabetic retinopathy .......................................................... 74 Table 4.1: Demographics of general practitioners responding to a survey on

diabetic retinopathy screening .................................................................... 83 Table 4.2: Current general practitioners management and attitudes to diabetes and

diabetic retinopathy ..................................................................................... 84 Table 4.3: Barriers to general practitioners performing dilated fundoscopy ............... 84 Table 4.4: General practitioners management of hypothetical clinical scenarios

and specific signs of diabetic retinopathy ................................................... 85 Table 5.1: International Clinical Diabetic Retinopathy Severity Scales8 .................... 96 Table 5.2: Diabetic retinopathy grading of the study patients based on slit lamp

biomicroscopy examination ........................................................................ 96 Table 5.3: The self reported diabetes micro- and macrovascular complications of

the enrolled study population ...................................................................... 97 Table 5.4: Sensitivity, specificity and Kappa correlations of overall diabetic

retinopathy grading by a consultant ophthalmologist and a medical officer from color fundus photographs of EyeScan (Ophthalmic Imaging System, CA) and FF450 (Carl Zeiss, North America), with reference to slit lamp biomicroscopy examination by a consultant ophthalmologist .......................................................................................... 97

Table 5.5: Kappa statistics for retinal photography using EyeScan and FF450 plus in comparison with the gold standard slit-lamp biomicroscopy examination by an ophthalmologist and a medical officer ......................... 98

Table 5.6: Classification of diabetic retinopathy into retinopathy stages (Wisconsin level)9 ....................................................................................... 99

Table 6.1: Specifications and prices of 27-inch iMac, 15-inch MacBook Pro and 9.7-inch iPad (The indicated prices are obtained in United States Dollars) ..................................................................................................... 109

Table 6.2: International Clinical Diabetic Retinopathy Severity Scales9 .................. 110 Table 6.3: The sensitivity, specificity and Kappa coefficient of 15-inch MacBook

Pro and 9.7-inch iPad in detecting diabetic retinopathy grading by a retinal specialist and a medical officer with reference to 27-inch iMac ... 111

19

Table 6.4: The sensitivity, specificity and Kappa coefficient of 15-inch MacBook Pro and 9.7-inch iPad in detecting microaneurysms and retinal hemorrhages by a retinal specialist and a medical officer with reference to 27-inch iMac ......................................................................................... 112

Table 6.5: The Kappa correlation of diabetic retinopathy changes interpreted by a retinal specialist and a medical officer on 15-inch MacBook Pro and 9.7-inch iPad with reference to the retinal findings detected on 27-inch iMac .......................................................................................................... 113

Table 7.1: International Clinical Diabetic Retinopathy Severity Scales and International Clinical Diabetic Macular Edema Disease Severity Scales8 ....................................................................................................... 126

Table 7.2: Patient characteristics and their diabetes history ...................................... 127 Table 7.3: The sensitivity, specificity and Kappa correlation coefficient for retinal

photography (FF450 plus, Carl Zeiss Inc., North America) and retinal video recording (EyeScan, Ophthalmic Imaging System, CA, US) with reference to slit lamp biomicroscopy examination ................................... 127

Table 7.4: The Kappa statistics for retinal videos (EyeScan) and retinal photography (FF450 plus) by both consultant ophthalmologists compared to the gold standard slit lamp biomicroscopy examination ...... 128

Table 8.1: The file size of a retinal video with different compression levels by reducing its bit rate while keeping other parameters constant (frame rate, frame size, zoom) .............................................................................. 139

Table 8.2: The quality of retinal videos (with/without diabetic retinopathy changes) of different compression levels rated by an ophthalmologist and a medical officer ................................................................................. 140

Table 8.3: The average conversion timing of an uncompressed raw retinal video (1 Gigabytes) to different compression levels .......................................... 140

Table 8.4: The sensitivity, specificity and Kappa correlation of different compression levels for retinal videos in detecting diabetic retinopathy grading by an ophthalmologist and a medical officer with reference to uncompressed raw retinal videos (1GB) ................................................... 141

Table 8.5: The Kappa correlation of the microaneurysms and retinal hemorrhages detected by both ophthalmologist and medical officer at different compression levels with reference to the uncompressed raw video files (1GB) ........................................................................................................ 141

20

LIST OF FIGURES

Figure 1.1: World’s first multipurpose portable imaging device (EyeScan) with a retinal video recording function .................................................................. 26

Figure 2.1: Microaneurysms and retinal hemorrhages ................................................ 51 Figure 2.2: New vessels formation at the disc (NVD) and a cotton wool spot

(CWS) ......................................................................................................... 52 Figure 2.3: Hard exudates, microaneurysms and retinal hemorrhages ........................ 52 Figure 2.4: Intraretinal microvascular abnormalities (IRMA) and venous beading

(VB) ............................................................................................................ 53 Figure 2.5: Area of non-perfusion accompanying venous beading (white arrow) ...... 53 Figure 2.6: Airlie House seven standard 30° stereoscopic fields on right eye ............ 59 Figure 5.1: Images captured by EyeScan and FF450 plus ......................................... 100

21

CONTENTS

PREFACE .................................................................................................................... 2 DECLARATION ........................................................................................................ 3 ACKNOWLEDGEMENTS ....................................................................................... 4 PUBLICATIONS ........................................................................................................ 6 INVITED BOOK CHAPTERS .................................................................................. 7 CONFERENCE PRESENTATIONS ........................................................................ 7 PUBLIC MEDIA RELEASES ................................................................................... 9 NOVEL CONTRIBUTIONS ................................................................................... 11 ABSTRACT ............................................................................................................... 12 ABBREVIATIONS ................................................................................................... 16 LIST OF TABLES .................................................................................................... 18 LIST OF FIGURES .................................................................................................. 20 CONTENTS .............................................................................................................. 21 SECTION 1: INTRODUCTION AND LITERATURE REVIEW ...................... 24 1. Introduction ........................................................................................................... 24

1.1 Research Background ....................................................................................... 24 1.2 Aims and Hypotheses of the Study ................................................................... 26 1.3 Structure of Thesis ............................................................................................ 28 1.4 General Methods ............................................................................................... 29

1.4.1 Section 2 (Chapters 1 and 2): Evaluation of Diabetic Retinopathy Screening Practices in the Primary Eye Care Setting in Australia ............ 29

1.1.1.1 Overall Study Design .......................................................................... 29 1.1.1.2 Sample Size Estimation ...................................................................... 29 1.1.1.3 Sample Population .............................................................................. 30 1.1.1.4 Mailing Coordinator ............................................................................ 30 1.1.1.5 Survey Package ................................................................................... 30 1.1.1.6 Survey Questionnaire (Appendices 1 and 2) ...................................... 31 1.1.1.7 Instrument Validity and Reliability .................................................... 32 1.1.1.8 Data Entry and Analyses ..................................................................... 32

1.4.2 Section 3 and 4 (Chapters 3 to 6): Validation of Easy-to-operate and Economical Novel Imaging Devices for Diabetic Retinopathy Screening—Retinal Still Photography and Retinal Digital Video Recording .................................................................................................. 33

1.4.2.1 Overall Design .................................................................................... 33 1.4.2.2 Sample Size Estimation ...................................................................... 33 1.4.2.3 Sample Population .............................................................................. 35 1.4.2.5 Diabetic Retinopathy Grading System ................................................ 36 1.4.2.6 Flow of the Screening Process ............................................................ 39 1.4.2.7 Reference Standard ............................................................................. 40 1.4.2.8 Diagnostic Modalities and Operators .................................................. 40 1.4.2.9 Retinal Color Digital Video Recording .............................................. 41

22

1.4.2.10 Data Processing and Interpretation ................................................... 41 1.4.2.11 Statistical Analyses ........................................................................... 42

2. Literature Review ................................................................................................. 44 2.1 Diabetes Mellitus .............................................................................................. 44

2.1.1 Definition and Diagnostic Criteria ............................................................. 44 2.1.2 Classification and Pathogenesis ................................................................. 46 2.1.3 Epidemiology ............................................................................................. 47 2.1.4 Clinical Features ........................................................................................ 47

2.2 Diabetic Retinopathy ........................................................................................ 49 2.2.1 Types and Clinical Features ....................................................................... 49 2.2.2 Pathogenesis ............................................................................................... 54 2.2.3 Epidemiology ............................................................................................. 56 2.2.4 Diabetic Retinopathy Screening ................................................................ 56

2.2.4.1 Screening Practices among Primary Eye Care Providers in Australia ............................................................................................. 56

2.2.4.2 Diabetic Retinopathy Screening Classification System ...................... 58 2.2.4.3 Diabetic Retinopathy Screening Modalities ....................................... 59

SECTION 2: EVALUATION OF DIABETIC RETINOPATHY SCREENING PRACTICES IN THE PRIMARY EYE CARE SETTING IN AUSTRALIA ............................................................................................................. 63 3. CHAPTER 1: DIABETIC RETINOPATHY MANAGEMENT BY AUSTRALIAN OPTOMETRISTS ......................................................................... 63

3.1 Summary ........................................................................................................... 63 3.2 Introduction ....................................................................................................... 64 3.3 Methods ............................................................................................................. 65 3.4 Results ............................................................................................................... 66 3.5 Discussions ....................................................................................................... 75

4. CHAPTER 2: DIABETIC RETINOPATHY MANAGEMENT BY AUSTRALIAN GENERAL PRACTITIONERS ................................................... 78

4.1 Summary ........................................................................................................... 78 4.2 Introduction ....................................................................................................... 78 4.3 Methods ............................................................................................................. 79 4.4 Results ............................................................................................................... 80 4.5 Discussion ......................................................................................................... 86

SECTION 3: RETINAL STILL PHOTOGRAPHY: NOVEL, EASY-TO-OPERATE AND EFFECTIVE DIAGNOSTIC DEVICES FOR DIABETIC RETINOPATHY SCREENING IN THE PRIMARY EYE CARE SETTING IN AUSTRALIA ....................................................................................................... 89 5. CHAPTER 3: LIGHT AND PORTABLE DEVICE FOR DIABETIC RETINOPATHY SCREENING .............................................................................. 89

5.1 Summary ........................................................................................................... 89 5.2 Introduction ....................................................................................................... 90 5.3 Methods ............................................................................................................. 91 5.4 Results ............................................................................................................... 94 5.5 Discussion ....................................................................................................... 100

23

6. CHAPTER 4: DIABETIC RETINOPATHY SCREENING: CAN THE VIEWING MONITOR INFLUENCE THE READING AND GRADING OUTCOMES ........................................................................................................... 104

6.1 Summary ......................................................................................................... 104 6.2 Introduction ..................................................................................................... 105 6.3 Methods ........................................................................................................... 106 6.4 Results ............................................................................................................. 108 6.5 Discussion ....................................................................................................... 113

SECTION 4: NOVEL VIDEO-BASED IMAGING TECHNOLOGY FOR DIABETIC RETINOPATHY SCREENING ....................................................... 118 7. CHAPTER 5: RETINAL VIDEO RECORDING: A NEW WAY TO IMAGE AND DIAGNOSE DIABETIC RETINOPATHY ................................. 118


8. CHAPTER 6: RETINAL VIDEO RECORDINGS AT DIFFERENT COMPRESSION LEVELS: A NOVEL, VIDEO-BASED IMAGING TECHNOLOGY FOR DIABETIC RETINOPATHY SCREENING ................ 133


SECTION 5: DISCUSSION, CONCLUSIONS AND FUTURE DIRECTIONS ......................................................................................................... 146 9. DISCUSSION ...................................................................................................... 146 10. CONCLUSIONS ............................................................................................... 156 11. FUTURE DIRECTIONS .................................................................................. 157 12. REFERENCES .................................................................................................. 159

24

SECTION 1: INTRODUCTION AND LITERATURE REVIEW

1. Introduction

1.1 Research Background

Diabetes mellitus is a metabolic disease that is rapidly increasing in prevalence

globally.1 In Australia, it is estimated that 941,000 Australians are currently living

with diabetes, and that this number will rise to 1.6 million by 2030.1 Diabetic

retinopathy (DR) occurs in 25% of patients with diabetes in Australia.2 Given the

increasing prevalence of diabetes worldwide, it is important for primary eye care

providers, such as optometrists and general practitioners (GPs), to actively participate

in screening services for DR. After the release of the Australian National Health and

Medical Research Council’s (NHMRC) guidelines on the ‘Management of Diabetic

Retinopathy’ in 1997, two national surveys on DR management by optometrists were

conducted in 1999 and 2001.3,4 However, none of the which was published on the GPs’

DR screening practices and management, except for a local Victorian GP survey.5

The published surveys3-5 showed that it is more desirable for optometrists to screen for

DR compared to GPs (84% vs. 50%). Nevertheless, primary eye care providers’

interest in DR screening needs to be improved because of the rising prevalence of

diabetes in Australia.2 Screening services in the community are conventionally

performed using dilated fundoscopy; however, retinal still photography is becoming a

more popular screening technique because it has been shown to be the most effective

DR screening tool, with sensitivity of at least 80%.6 However, retinal cameras are

generally expensive and technically challenging to operate; thus, they are often not

readily available in the primary health care setting, especially in rural areas. For

25

Chapters 1 and 2, we conducted national surveys to evaluate the current screening

practices and management of optometrists and GPs in Australia.

The cost and user-friendliness of screening devices, reading devices and imaging

techniques can play a major role in the participation rate of primary eye care

providers in a screening program. A reduction in screening costs and technical

difficulties in the primary health care setting will not only help to increase patients’

accessibility to diabetes eye care services in the community, but it will also redirect

finite specialist services, such as the provision of lasers, intravitreal injections and

surgical interventions for patients with sight-threatening DR changes, to more

appropriate areas.

In 2007, the Lions Eye Institute (LEI) in Western Australia (WA) developed and

patented an economical (US$30,000) and easy-to-operate multipurpose ophthalmic

imaging device called EyeScan (Ophthalmic Imaging System (OIS), Sacramento, US)

(see Figure 1.1). It was approved by the US Food and Drug Administration (FDA) in

October 2009. It is a light (1 kg) portable camera that has a 5.3 MP sensor; hence, it

may be a suitable device for primary eye care providers to use in the routine, mobile

and tele-retinal screening setting. Chapter 3 evaluates the efficacy of EyeScan in

detecting DR changes in patients with diabetes, whereas Chapter 4 explores the

possibility of using smaller reading devices (e.g., laptop and iPad) to read the retinal

images for DR screening.

In addition, this device can record videos. To date, no studies have reported on the use

of retinal video recording in the DR screening setting. This is a novel technique that

offers a panoramic view of the retina within a short timeframe, and it simulates what

is seen in the slit lamp. Hence, this study aims to investigate the clinical effectiveness

26

and user-friendliness of this technique in screening for DR in the primary health care

setting.

Figure 1.1: World’s first multipurpose portable imaging device (EyeScan) with a

retinal video recording function

1.2 Aims and Hypotheses of the Study

The aims of this thesis are to evaluate the current DR practices and management of

primary eye care providers and validate novel imaging technologies for DR screening.

The hypotheses of this study are:

1. The use of a retinal camera can help to increase primary eye care providers’

confidence and desire to screen for DR.

2. The use of an easy-to-operate, economical and effective multipurpose,

portable OIS (EyeScan, Sacramento, US) for retinal still photography is as

effective as conventional, expensive retinal cameras (FF450 plus, Carl Zeiss,

Inc.) in screening for DR in the primary health care setting.

3. The iPad is an effective reading device for specialist and non-specialist

personnel when reviewing retinal color still images in screening for DR.

27

4. Retinal video recording is a novel, easy-to-operate and effective alternative to

screening for DR in the community.

5. Retinal still photography and retinal video recording captured using EyeScan

can be efficiently operated by an inexperienced medical officer with minimal

training.

6. Retinal videos can easily be compressed with excellent diagnostic accuracy in

detecting DR lesions and grading.

28

1.3 Structure of Thesis

Section 1: Introduction and Literature Review

Section 2: Evaluation of Diabetic Retinopathy Screening Practices in Primary

Eye Care Setting in Australia

Section 3: Retinal Still Photography: Novel, Easy-to-operate and Effective Diagnostic Devices for Diabetic Retinopathy Screening

Section 4: A Novel Video-based Imaging Technology for Diabetic Retinopathy

Screening

Discussion: Overview, Conclusions and Future Directions

Introduction Literature Review

Chapter 1: Diabetic Retinopathy Management by Australian

Optometrists

Chapter 2: Diabetic Retinopathy Management by Australian GPs

Chapter 3: A Light and Portable Device for Diabetic Retinopathy

Screening

Chapter 4: The Use of Portable Devices for Retinal Color Still Images Interpretation in Diabetic Retinopathy

Screening

Chapter 6: The Use of Compressed Digital Retinal Videos for Diabetic

Retinopathy Screening

Chapter 5: Retinal Video Recording: A New Way to Image and Diagnose

Diabetic Retinopathy

29

1.4 General Methods

1.4.1 Section 2 (Chapters 1 and 2): Evaluation of Diabetic Retinopathy Screening

Practices in the Primary Eye Care Setting in Australia

1.1.1.1 Overall Study Design

This section consists of two cross-sectional descriptive surveys that were completed

by optometrists and GPs. The surveys provided an overview of the self-reported

current DR management practices by primary eye care providers in Australia. This

study was approved by the University of Western Australia’s (UWA) Human

Research Ethics Committee prior to the commencement of the study.

1.1.1.2 Sample Size Estimation

For both surveys, a minimum sample size of 400 participants was required to allow a

power of 80% with a=0.05 to detect a significant difference in change from a 40%

response to a given question to 50%.7 A change from 40% to 50% was selected

because it is the change that would require the highest number of respondents to attain

statistical significance.

30

1.1.1.3 Sample Population

The survey packages were mailed to 1,000 optometrists and 2,000 GPs throughout all

states in Australia. The details of the optometrists and GPs were obtained from the

Australian Optometric Association membership database (4,414 members) and the

Royal Australian College of General Practitioners membership database (12,938

members).

1.1.1.4 Mailing Coordinator

A print-and-mail service (UniPrint) within UWA was appointed as the mailing

coordinator of these surveys. Being independent of the research group, UniPrint kept

a list of currently practicing optometrists and GPs in Australia and assigned a specific

digit to each survey package sent to the survey participants. This was done in order to

keep a record of the non-respondents so that a reminder could be sent out later in the

study period in an effort to maximize the participation rate of primary eye care

providers. In this project, two repeat mail-outs were sent during the fourth and eighth

months of the study period. For the respondents to remain anonymous in this project,

the completed questionnaires were received directly by the researcher. Therefore, the

researcher was blinded from the respondents’ details, while UniPrint was blinded

from the questionnaires completed by the respondents.

1.1.1.5 Survey Package

Each survey package consisted of: 1) an information pamphlet detailing the objectives

of the study; 2) a three-page number-labeled questionnaire; and 3) a postage-paid

return envelope. On the information sheet, the recipient could choose to opt-out of the

study and return the survey package if he or she did not want to take part in the survey.

In addition, the information sheet stated that consent to participate in the study was

31

assumed to be given by the participants on the basis of the returned and completed

questionnaire.

1.1.1.6 Survey Questionnaire (Appendices 1 and 2)

The design and development of the survey questionnaire was performed in

collaboration with the Centre for Eye Research Australia (CERA), which previously

conducted two similar surveys. With the permission and support of CERA, the

majority of the questions adopted from the previous surveys were purposely kept

unchanged, with only slight modifications to sentence structures and the order of

questions to allow a direct comparison and analysis of the survey findings.

Each self-administered survey questionnaire consisted of questions relating to:

1. Demographics of the respondents, including:

a. age

b. duration of practice

c. location of practice (urban or rural)

d. state of practice

e. previous training location.

2. Frequency of risk factors assessment, such as glycosylated hemoglobin

(HbA1c), blood pressure, lipid profile and smoking.

3. DR screening frequency and practices for patients with diabetes.

4. Frequency of referral of patients with diabetes to ophthalmologist.

5. Perceived barriers to DR screening.

6. Desire to screen for DR in the community.

7. Twelve hypothetical clinical scenarios for patients with diabetes.

32

8. Frequency of use of a retinal camera in the practice. (This question was only

included in the survey for optometrists, as most GPs in Australia do not have a

retinal camera in their practice.)

1.1.1.7 Instrument Validity and Reliability

To increase the generalizability of the survey, the sample population was stratified to

give a proportional representation according to the state and location (urban versus

rural) of the optometrists and GPs. With regards to this survey, the honesty and

accuracy of the participants’ responses was a potential reporting bias. In light of this,

we ensured the protection of the respondents’ anonymity by employing UniPrint as

the mailing coordinator in order to encourage more truthful and honest responses

from the optometrists and GPs. Given that the questions listed in the questionnaire

were related to routine day-to-day practices, recall bias was not an issue in this survey.

1.1.1.8 Data Entry and Analyses

Upon receipt of the survey questionnaires, data entry was performed by a researcher

using Microsoft Access. To maximize the response rate for each question, no

respondents were excluded from the survey. Data analyses were performed using

SPSS version 17 (SPSS, Chicago, IL, US) and Stata 10.0 (StataCorp, College Station,

TX, US). Descriptive statistics were calculated for all continuous variables.

Relationships between categorical variables were explored using Pearson chi-square

tests. Multivariate logistic regression models were used to explore outcomes of

interest, such as the use of a retinal camera and desire and confidence to detect DR

signs while controlling for possible confounding factors such as training duration,

state and location of practice.

33

1.4.2 Section 3 and 4 (Chapters 3 to 6): Validation of Easy-to-operate and

Economical Novel Imaging Devices for Diabetic Retinopathy Screening—

Retinal Still Photography and Retinal Digital Video Recording

1.4.2.1 Overall Design

These sections evaluate the use of:

a. an economical, novel diagnostic screening device using EyeScan

b. an economical and portable reading device, such as an iPad, for the

interpretation of retinal color still images

c. a novel imaging technique using retinal digital video recording for DR

screening

d. retinal digital video recording at different compression levels for DR

screening.

These studies were approved by Royal Perth Hospital and the UWA Human Research

Ethics Committee.

1.4.2.2 Sample Size Estimation

• Chapter 3: An Economical, Portable Device for Diabetic Retinopathy Screening

o To allow a power of 95%, desired precision of 0.10 and expected

sensitivity and specificity of 95%, the total number of eyes required for

each diagnostic modality was 77 (prevalence was set at 0.25, as 25% of

people with diabetes develop DR).

o A total of 136 consecutive patients (272 eyes) were recruited from the

diabetic retinopathy screening clinic of Royal Perth Hospital between

March 2010 and September 2010.

34

• Chapter 4: The Use of a Portable Device for Retinal Color Still Images

Interpretation in Screening Diabetic Retinopathy

o To allow for a power of 95%, desired precision of 0.10 and expected


each device was 71 (prevalence was set at 0.50, as the selected samples

consisted of 50% normal and 50% abnormal retinal color still images).

o Of the 136 patients (272 eyes) recruited from the diabetic retinopathy

screening clinic of Royal Perth Hospital, a random sample of 50 normal

eyes and 50 eyes with diabetic retinopathy were selected from the

analyzed images from Chapter 3. These images, consisting right and left

eyes, were randomized and analyzed using 15-inch MacBook Pro and 9.7-

inch iPad, with reference to 27-inch iMac monitor.

• Chapter 5: Retinal Video Recording: A New Way to Image and Diagnose

Diabetic Retinopathy

o To allow a power of 95%, desired precision of 0.10 and expected


each diagnostic modality was 77 (prevalence was set at 0.25, as 25% of

people with diabetes develop DR).

o A total of 100 consecutive patients (100 eyes) were recruited from the

diabetic retinopathy screening clinic of Royal Perth Hospital between

March 2010 and September 2010.

• Chapter 6: Validation of the Use of Compressed Digital Retinal Videos for

Diabetic Retinopathy Screening

o To allow for a power of 95%, desired precision of 0.10 and expected


35

each compression level was 31 (prevalence was set at 0.50, as selected

samples consisted of 50% normal and 50% abnormal retinal digital videos).

o Of the 100 patients (200 eyes) recruited from the diabetic retinopathy

screening clinic of Royal Perth Hospital between March till September

2010, a random sample of 18 normal retinal videos and 18 retinal videos

with diabetic retinopathy were compressed from 500 MB (raw video file)

to 100 MB, 30 MB, 20 MB and 5 MB. These images, consisting right and

left eye retinal videos, were randomized and analyzed using 27-inch iMac

monitor.

1.4.2.3 Sample Population

The sample population was recruited from the Diabetic Retinopathy Screening Clinic

at the Royal Perth Hospital’s Ophthalmology Department over a six-month period.

Prior to signing the consent form, all patients were informed of the objectives of the

study and given an information sheet. In addition, they were informed of the possible

side effects and risks, possible benefits, privacy and confidentiality, and contact

details of the relevant authorities. Patients who agreed to participate in the study were

asked to sign a consent form. If they wanted to ‘opt-out’ of the study, they were

reassured that they would still receive the usual standard of care for DR screening.

1.4.2.4 Standard Care for Diabetic Retinopathy Screening at the Royal Perth Hospital

Three-field (optic disc, macula and temporal view) mydriatic non-stereo color retinal

still photography was performed using a conventional retinal camera, FF450 plus

(Carl Zeiss, Meditec, US). The images were interpreted by a senior medical officer

who subsequently determined the next appointment date for the patients based on

their DR severity using the International Clinical DR Severity Scales (see Table 1.1).8

36

1.4.2.5 Diabetic Retinopathy Grading System

International Clinical DR and Macular Edema Disease Severity Scales (see Table 1.1)

was utilized in this study for the DR grading,8 as it is a more simplified grading

system compared to the Early Treatment DR Screening (ETDRS) (see Table 1.2).9 It

has fewer severity levels and diagnostic criteria; hence, it is much easier to use. It was

published in the Global DR Project in 2003 by the American Academy of

Ophthalmology with the aim of developing a common clinical severity scale to

facilitate more effective communication between primary health care providers (e.g.,

optometrists and GPs), diabetes specialists and ophthalmologists.

37

Table 1.1: International Clinical Diabetic Retinopathy Severity Scales and

International Clinical Diabetic Macular Edema Disease Severity Scales8

Grades Retinal Findings

None No abnormalities

Mild non-proliferative DR

(NPDR)

Microaneurysms (MAs) only

Moderate NPDR More than just MAs, but less than severe NPDR

Severe NPDR Any of the following:

i. Extensive (>20) intra-retinal hemorrhages in each of

four quadrants

ii. Definite venous beading in 2+ quadrants

iii. Prominent IRMA in 1+ quadrant

AND no signs of PDR

PDR One or more of the following:

i. Neovascularization

ii. Vitreous/pre-retinal hemorrhage

DME apparently absent

DME apparently present

No apparent retinal thickening or hard exudates in

posterior pole

Some apparent retinal thickening or hard exudates in

posterior pole

38

Table 1.2: Classification of diabetic retinopathy into retinopathy stages

(Wisconsin level)9

DR Stage Retinal Findings

Minimal NPDR MAs and one or more of the following:

i. retinal haem

ii. Hex

iii. CWS

but not meeting the criteria for moderate NPDR

Moderate NPDR H/Ma >std photo 2A in at least one quadrant and one or

more of the following:

i. CWS

ii. VB

iii. IRMA

but not meeting severe NPDR

Severe NPDR Any of:

H/Ma>std photo 2A in all four quadrants

IRMA >std photo 8A in one or more quadrants

VB in two or more quadrants

PDR Any of:

NVE or NVD <std photo 10A,

vitreous/pre-retinal haem

NVE<1/2 DA without NVD

High-risk PDR Any of:

NVD>1/4 to 1/3 disc area

or with vitreous/ pre-retinal haem

or NVE>1/2 DA with vitreous/pre-retinal haem

Advanced PDR High-risk PDR with tractional detachment involving macula

or vitreous haem obscuring ability to grade NVD and NVE

39

1.4.2.6 Flow of the Screening Process

Patients who agreed to participate in the study had to go through the following steps

in the eye clinic.

CONSENT FOR THE STUDY (Information/Objectives of the Study)

PUPIL DILATION 0.5% Tropicamide and 1% Phenylephrine

THREE-FIELD RETINAL COLOR STILL PHOTOGRAPHY

(Optic Disc, Macula, Temporal View)

i. FF450 PLUS—Retinal Photographer ii. OIS EYESCAN—Resident Medical Officer

RETINAL DIGITAL VIDEO RECORDING

i. OIS EYESCAN—Resident Medical Officer

SLIT LAMP EXAMINATION—Senior Consultant Ophthalmologist

(GOLD STANDARD)

IMAGE DE-IDENTIFICATION AND RANDOMIZATION

40

1.4.2.7 Reference Standard

For the detection of DR, the ETDRS’s seven-field 35 mm stereoscopic color fundus

photographs using the modified Airlie House classification is currently the gold-

standard photographic technique.9 Nevertheless, we preferred using the slit lamp

biomicroscopy examination by a senior consultant ophthalmologist as the reference

standard of this study because:

a. slit lamp examination has been shown to be compared favorably with seven

field stereo-photography (sensitivity 87.4%, specificity 94.9%)10

b. slit lamp examination is easy to perform and less time-consuming

c. patients are less tolerant to seven-field stereoscopic color fundus photography

immediately after undergoing three sets of retinal examination using two

different devices

d. for patients with moderate to severe cataracts, the quality of retinal still images

is significantly compromised; thus, the patient will end up needing a slit lamp

examination by an eye specialist.

1.4.2.8 Diagnostic Modalities and Operators

All patients underwent four sets of tests for the diagnosis of DR:

i) three-field 35° non-stereo retinal color still photography by FF450 plus

ii) three-field 35° non-stereo retinal color still photography by EyeScan

iii) retinal color digital video recording by EyeScan

iv) slit lamp biomicroscopy examination with a 78-diopter lens by a senior

consultant ophthalmologist (gold standard).

41

The FF450 plus was operated by an experienced retinal photographer (with more than

10 years of experience in performing retinal still photography), while the EyeScan

was operated by a resident medical officer who had no previous ocular imaging

experience. Prior to screening the patients, the medical officer underwent a three-hour

training session on the EyeScan device for both retinal still photography and retinal

video recording.

1.4.2.9 Retinal Color Digital Video Recording

All retinal video recordings commenced at the optic disc and proceeded to the macula

and temporal regions. To obtain continuity of retinal information between the regions,

the retinal camera was tilted at a consistent pace from left to right for the right eye

(optic disc, macula and temporal retina) for at least five seconds on each view, and

vice versa for the left.

1.4.2.10 Data Processing and Interpretation

The retinal still images (FF450 plus and EyeScan) and digital videos (EyeScan) were

downloaded separately onto an external hard drive. Subsequently, they were de-

identified, randomized and interpreted by consultant ophthalmologists and a medical

officer using the International Clinical DR Severity Scales (see Table 1.1). The results

of the retinal images and videos were recorded on the data sheet.

The quality of the digital retinal videos and digital retinal photography images were

classified as either ‘unacceptable’ or ‘acceptable’. The retinal video recordings were

graded as unacceptable by the ophthalmologist if they were blurred, out of focus, dark

and/or had insufficient views (fewer than five seconds on any view). The retinal

photos were graded as ‘unacceptable’ if more than one-third of the photo was not

interpretable. All retinal still images and videos were interpreted in a dimly lit room

42

using iPhoto (Apple, CA, US) and VLC Media Player 1.1.4 (Apple, CA, US) on a

standardized monitor screen (iMac 27 inches; Apple, CA, US).

A total of 100 sets of three-field (optic disc, macula and temporal views) retinal color

still images consisting of 50 normal and 50 with DR were selected and interpreted on

two smaller reading screens—a 15-inch MacBook Pro (Apple, CA, US) and a 9.7-

inch iPad (Apple, CA, US)—apart from the 27-inch iMac (Apple, CA, US). The

selected images captured by FF450 plus (Carl Zeiss, Inc., US) were all ‘acceptable’

based on the quality rated on the 27-inch iMac. All images were graded by a retinal

specialist and a medical officer.

In addition, 36 retinal color digital videos with ‘acceptable’ quality were chosen and

compressed to four different levels/megabytes (MB) (Group 1: 100 MB; Group 2:

30 MB; Group 3: 20 MB; Group 4: 5 MB) from the uncompressed original file size

(1 GB (gigabyte)) by reducing their bit rates using Xilisoft Video Converter Ultimate

6.0, which possesses a standard video codec H.264 (see Table 1.2). The frame rate

and frame size were set at 17 frames per second and 640 x 480 pixels respectively by

the fundus camera (EyeScan). All videos were interpreted on a 27-inch iMac by a

consultant ophthalmologist and a medical officer.

1.4.2.11 Statistical Analyses

All data were entered into Microsoft Excel 2007 and analyzed using SPSS version 17

(SPSS, Chicago, IL, US). The main outcome measures of the study were:

i) retinal still photography using EyeScan and FF450 plus, graded by a consultant

ophthalmologist and a medical officer on a 27-inch iMac with reference to gold-

standard slit lamp examination by a consultant specialist (see Chapter 3)

43

ii) retinal still photography using FF450 plus on a 15-inch MacBook Pro and a 9.7-

inch iPad by a retinal specialist and a medical officer with reference to a 27-inch

iMac (see Chapter 4)

iii) retinal video recording using EyeScan on a 27-inch iMac by two consultant

ophthalmologists with reference to gold-standard slit lamp examination by a

consultant specialist (see Chapter 5)

iv) retinal digital video recording with different compression levels using EyeScan by

a consultant ophthalmologist and a medical officer on a 27-inch iMac with

reference to the raw uncompressed original file size (see Chapter 6).

44

2. Literature Review

2.1 Diabetes Mellitus

2.1.1 Definition and Diagnostic Criteria

Diabetes mellitus (DM) is a metabolic disorder that is characterized by hyperglycemia

that is secondary to either impaired insulin secretion or insulin resistance. Chronic,

uncontrolled hyperglycemia can give rise to macrovascular (cerebrovascular accident,

ischemic heart disease and peripheral vascular disease) and microvascular

(retinopathy, nephropathy and neuropathy) complications.

According to the American Diabetes Association (ADA), the current diagnostic

criteria for diabetes are as follows:11,12

1. fasting plasma glucose ≥7.0 mmol/L (fasting is defined as no caloric intake for

at least eight hours)

2. two-hour plasma glucose ≥11.1 mmol/L on oral glucose tolerance test (OGTT)

(the OGTT should be performed as per the World Health Organization’s

(WHO) guidelines using a glucose load that contains the equivalent of 75 g

anhydrous glucose dissolved in water).

3. glycosylated hemoglobin (HbA1c) ≥6.5% (the test should be performed in a

laboratory using a method that is certified by the National Glycohaemoglobin

Standardization Program and standardized to the Diabetes Control and

Complications Trial (DCCT) assay).

4. random plasma glucose ≥11.1 mmol/L in the presence of classic symptoms of

hyperglycemia or hyperglycemic crisis.

45

To confirm the diagnosis of DM, one of these tests (fasting plasma glucose, OGTT or

HbA1c) had to be repeated on a subsequent day, except if any two of the tests were

positive on the same occasion. In addition, two pre-diabetic states (impaired fasting

glucose (IFG) and impaired glucose tolerance (IGT)) that include patients with

abnormally elevated plasma glucose levels do not qualify for the diagnosis of DM

(see Table 2.1). The definition of IFG by the ADA is slightly different from that of the

WHO. For the ADA, IFG is defined as fasting plasma glucose levels from 5.6 mmol/L

to 6.9 mmol/L, while for the WHO,13 it is between 6.1 mmol/L and 6.9 mmol/L.

Conversely, IGT is defined by both the ADA and the WHO as two-hour plasma

glucose levels of 7.8 mmol/L to 11.0 mmol/L on OGTT.

Table 2.1: The current diagnostic criteria for diabetes, impaired fasting glucose

and impaired glucose tolerance published by the American Diabetes

Association11,12

Category

Fasting Plasma Glucose

(mmol/l)

Two-hour Plasma Glucose

(mmol/l)

Normal <5.6 <7.8

IFG 5.6–6.9 —

IGT — 7.8–11.0

DM ≥7.0 ≥11.1

46

2.1.2 Classification and Pathogenesis

DM exists in several forms: type 1 DM (T1DM), type 2 DM (T2DM), gestational DM

and various other secondary DM relating to genetic defects, diseases of exocrine

pancreas, endocrinopathies, chemical inducement and infections.11,14 According to the

WHO, DM is broadly categorized into two etiopathogenetic groups: T1DM and

T2DM.15 DM occurs as a result of insulin insufficiency or insulin resistance on target

tissues, which leads to the abnormal metabolism of carbohydrates, fats and proteins in

people with diabetes.11 T1DM is usually characterized by absolute insulin deficiency

that is secondary to the destruction of B-cells (responsible for insulin secretion) in the

pancreas by a cellular-mediated autoimmune pathogenic process or other unknown

factors.11 Traditionally, T1DM is also known as juvenile-onset DM or insulin-

dependent DM. At least one or more autoantibodies to B-cell of the pancreas (islet

cell autoantibodies (ICD)), autoantibodies to insulin, autoantibodies to glutamic acid

decarboxylase (GAD) 65 and autoantibodies to tyrosine phosphatases 1A-2 and 1A-

2B is present in nearly 90% of patients with T1DM.11 Often, these patients are also

susceptible to other autoimmune-mediated diseases, such as Hashimoto’s and Grave’s

thyroid disease, autoimmune hepatitis, myasthenia gravis, celiac disease and

pernicious anemia.

In contrast, T2DM is a much more common form of DM and is characterized by

varying degrees of insulin resistance and relative (rather than absolute) insulin

deficiency.11, 16-18 Traditionally, this is known as non-insulin-dependent DM (NIDDM)

or adult-onset DM. This form of DM has a strong genetic predisposition that is

polygenic, complex and not clearly defined.19 It is often associated with other

cardiovascular risk factors, such as hypertension, hyperlipidemia (high serum, low-

density-lipoprotein (LDL)) and obesity.20

47

2.1.3 Epidemiology

According to the WHO, the global prevalence of diabetes for all age groups is

estimated to increase from 6.6% (285 millions) to 7.8% (438 millions) between 2010

and 2030 as the result of the increasing rate of obesity, ageing populations,21,22 better

detection of diabetes and the survival of patients with diabetes.23 In addition,

individuals with a pre-diabetic state will also increase from 314 million to 472 million

by 2030.24 In Australia, almost 25% of Australians above the age of 25 has diabetes or

a pre-diabetes condition.25 The prevalence of diabetes among Australian males and

females is 8% and 6.8% respectively (average of 7.4%), whereas for the Australian

indigenous population, the prevalence is twice as high (15% vs. 7.4%) when adjusted

for age and gender.26 T2DM is more prevalent than T1DM (70%–90% vs. 5%–

10%).27-29 Further, T1DM and T2DM have been shown to significantly vary in

incidence according to age and gender,30-32 geographical distribution,33,34 family

history35-37 and ethnic group.29

2.1.4 Clinical Features

The clinical presentation of patients with diabetes often varies. Some patients present

with non-specific symptoms such as lethargy, weight loss, polyuria, polydipsia and

polyphagia.11 Some may experience symptoms as a result of macro- or microvascular

complications such as slurred speech, loss of power/sensation, loss of consciousness

or memory, chest pain, shortness of breath, lower-limb claudication, visual

impairment, proteinuria, polyuria, diabetic foot ulcers and cellulitis. More importantly,

a prolonged period of untreated high plasma glucose levels can have life-threatening

consequences due to diabetic ketoacidosis (DKA) and hyperosmolar hyperglycemic

syndrome (HHS), with mortality rates of 2%–5% and 15% respectively.38,39

48

DKA, which occurs in T1DM, is a serious metabolic derangement that is

characterized by a combination of hyperglycemia, hyperketonemia and metabolic

acidosis. Patients can present with abdominal pain (40%–75% of DKA),40 vomiting

and abnormal breathing patterns (also known as Kussmual’s breathing), which are

often preceded by a few days of non-specific symptoms such as polyuria and

polydipsia. Conversely, HHS is less common than DKA and can manifest as severe

dehydration and focal or global neurologic deficits.39,41,42 Patients generally have

hyperglycemia and hyperosmolality, which are secondary to severe dehydration

without significant ketoacidosis upon presentation. Table 2.2 shows the diagnostic

criteria for DKA and HHS according to the ADA.

Table 2.2: Diagnostic criteria for DKA and HHS by the ADA39

DKA HHS Mild Moderate Severe Plasma glucose (mmol/L) >13.9 >13.9 >13.9 >33.3 Arterial pH 7.25–7.30 7.00–7.24 <7.00 >7.30 Serum bicarbonate (mmol/L)

15–18 10–14 <10 >15

Urine ketone* Positive Positive Positive Positive Serum ketone* Positive Positive Positive Positive Effective serum osmolality** (mOsm/kg)

Variable Variable Variable >320

Anion gap*** >10 >12 >12 <12 Alteration in sensorium or mental obtundation

Alert Alert/ drowsy

Stuporous/ comatose

Stuporous/ comatose

DKA: diabetic ketoacidosis; HHS: hyperosmolar hyperglycemic state

*Nitroprusside reaction method

**Calculation: Effective serum osmolality: 2(measured Na (mmol/L))+glucose

(mmol/L)

***Calculation: Anion gap (Na+)–(Cl- + HCO3-)(mmol/L)

49

2.2 Diabetic Retinopathy

2.2.1 Types and Clinical Features

DR is one of the most common microvascular complications of diabetes.43 Nearly 100%

of T1DM patients and more than 60% of T2DM patients will have at least some

retinopathy after 20 years of diabetes.43,44 DR exists in several forms: NPDR, PDR and

diabetic macular edema (DME). In general, DR starts from the NPDR stage and

evolves to PDR if the plasma glucose level is left untreated or poorly controlled.

NPDR is characterized by the presence of abnormal retinal vasculature changes such

as MAs, retinal hemorrhages, cotton wool spots (CWS), venous beading (VB) and

intra-retinal microvascular abnormalities (IRMA), while PDR has features of new

vessels proliferation on the disc (NVD) or elsewhere (NVE) in response to retinal

ischemia. One of the earliest changes seen in people with DR is the formation of MAs

(10–100 microns in diameter) (see Figure 2.1). They are focal dilations of retinal

capillaries that appear as red dots. These changes are often reversible with the

optimization of the plasma glucose level. Nevertheless, persistent hyperglycemia can

subsequently lead to progressive capillary wall weakening and ruptures, forming

intra-retinal hemorrhages (Figure 2.2), which can exist in various shapes, such as dot

hemorrhages, blot hemorrhages and flame hemorrhages. Dot hemorrhages are often

indistinguishable from MAs, as they both manifest as small red dots on color fundus

photographs. They can only be differentiated on fluorescein angiograms (FFA) when

MAs appear as hyperfluorescent (bright) lesions, while dot hemorrhages appear as

hypofluorescent (black) lesions. Flame hemorrhages (flame or splinter in shape) and

CWSs (see Figure 2.3) occur superficially at the level of nerve fiber layers, and they

can also be seen in hypertensive retinopathy.

50

CWSs, also known as soft exudates, manifest as white lesions and result from the

swelling and infarction of nerve fiber layers secondary to the occlusion of the

precapillary arterioles. The tight junctions of retinal capillary cells in diabetes are

often impaired, causing the breakdown of the blood-retina barrier and the leakage of

the protein substance from the systemic circulation. The combination of the proteins

and the surrounding retinal neurosensory tissues leads to the formation of hard

exudates (see Figure 2.3), which are bright yellow. VB and IRMA (see Figure 2.4),

which are high-risk features associated with the development of PDR,9 are often

found adjacent to the non-perfused areas of the retina (see Figure 2.5). IRMA is

formed by a group of remodeled capillary beds without proliferative changes, and

they do not leak on FFA, while VB often manifests as saccular bulges on retinal veins,

and they are usually accompanied by adjacent areas of capillary dropout on FFA.

NVD (see Figure 2.2) is defined as neovascularization, which occurs within one disc

diameter of the disc margin, while NVE is for new vessels that are present elsewhere.

These new vessels are often fragile and prone to bleeding, leading to pre-retinal

hemorrhages/subhyaloid hemorrhages or vitreous hemorrhages. The ‘high-risk’ PDR

changes associated with poor visual prognosis are: 1) NVD >¼ disc area; 2) any NVD

with vitreous or pre-retinal hemorrhage; 3) NVE ≥½ disc area with vitreous or pre-

retinal hemorrhage; or 4) vitreous or pre-retinal hemorrhage obscuring ≥1 disc area.45

In addition, neovascularization can take place on the iris and over the iris angle,

leading to the formation of fibrovascular membranes (often invisible on gonioscopic

examination). The fibrovascular membranes and neovascularization over the iris angle

will progressively obstruct the trabecular meshwork, causing raised intraocular

pressure and secondary open-angle glaucoma. These fibrovascular membranes can

51

also contract, drawing the iris to the trabecular meshwork and forming peripheral

anterior synechiae. This disease process can lead to secondary angle-closure glaucoma.

DME, which is another form of retinopathy, is defined as the presence of any retinal

thickening within two disc diameters of the fovea. If the retinal thickening is within

500 microns of the fovea, it is labeled as clinically significant macular edema

(CSME).46 Sight-threatening DR (STDR) usually refers to people with severe NPDR,

PDR and DME who require prompt medical investigation and intervention to prevent

severe visual impairment secondary to DR-related complications.

Figure 2.1: Microaneurysms and retinal hemorrhages

Microaneurysms

Retinal hemorrhages

52

Figure 2.2: New vessels formation at the disc (NVD) and a cotton wool spot

(CWS)

Figure 2.3: Hard exudates, microaneurysms and retinal hemorrhages

NVD CWS

Microaneurysms

Retinal hemorrhages

Hard Exudates

53

Figure 2.4: Intraretinal microvascular abnormalities (IRMA) and venous

beading (VB)

Figure 2.5: Area of non-perfusion accompanying venous beading (white arrow)

IRMA

VB

Non-perfused area

Non-perfused area

Venous beading

54

2.2.2 Pathogenesis

The pathogenesis of DR is multifactorial. Chronic hyperglycemia is thought to be the

primary cause for its development.47,48 It compromises a retinal autoregulatory

mechanism, vascular changes and retinal ischemia. Various other biochemical

pathways have also been suggested to play a role in the development and progression

of DR in people with diabetes—for example, the accumulation of sorbitol and

advanced glycation end (AGE) products,49,50 impaired autoregulation of retinal blood

flow,51 increased level of protein kinase C,52 intraocular and serum angiotensin-

converting enzyme (ACE),53-55 plasma prekallikrein,56 erythropoietin57 and various

growth factors (e.g., vascular endothelial growth factor (VEGF), insulin-like growth

factor (IGF), platelet-derived growth factors and pigment-derived factor (PEDF)).58-60

The accumulation of sorbitol damages vascular cells and pericytes, leading to the

thickening of retinal vascular endothelial cell basement membranes and the closure of

retinal capillaries.

55

Development of diabetic retinopathy changes61

Classification

Microvascular dysfunction à narrowing of blood vessels and increasing vessel permeability

Chronic hyperglycemia

Formation of microaneurysms and retinal hemorrhages

Persistent capillary non-perfusion à development of cotton wool spots, venous beading and intra-retinal microvascular abnormalities

Prolonged period of hypoxia à formation of new vessels on retina, posterior vitreous and iris (rubeosis iridis)

If left untreated à increased risk of developing blinding complications such as vitreous hemorrhage, pre-retinal/subhyaloid hemorrhage and rubeiotic glaucoma

56

2.2.3 Epidemiology

According to the WHO, it is estimated that DR accounts for 4.8% of the number of

cases of blindness (37 millions) worldwide.62 A pooled analysis of data from eight

population-based eye surveys conducted by the US National Eye Institute (NEI)

showed that the overall crude prevalence of DR and STDR in people aged 40 years or

over who were known to have DM were 40.3% and 8.2% respectively.63 The

prevalence of DR and STDR in T1DM was found to be much higher than T2DM.

According to the Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR)

and the New Jersey population study, 82% of T1DM had DR changes, including 32%

with STDR changes.64 In contrast, the Liverpool Diabetic Eye Study65 and the United

Kingdom Prospective Diabetes Study (UKPDS)66 reported that the prevalence of DR

in T2DM was 25% and 27% respectively, with 1% of STDR. In Australia, the

prevalence rate of DR in patients with known DM ranges from 22% to 35%, while for

STDR, it ranges from 1.2% to 7.1% (AusDiab study,2 Blue Mountains Eye Study

(BMES),67 Melbourne Visual Impairment Project (MVIP)68 and Newcastle Diabetic

Retinopathy Study69,70). Further, the BMES reported a five-year incidence of DR and

PDR of 22.2% and 1.5% respectively for people with diabetes.

2.2.4 Diabetic Retinopathy Screening

2.2.4.1 Screening Practices among Primary Eye Care Providers in Australia

In Australia, DR screening is mainly performed by primary care physicians, including

GPs and optometrists. There were approximately 941,000 Australians with diabetes in

2000,1 and it is projected that this figure will increase to 1.6 million by 2030.71 In

1997, the NHMRC published clinical practice guidelines for DR management,

including different screening and treatment modalities. The guidelines recommended

57

that all examiners should assess patients’ best-corrected visual acuity and perform a

dilated fundus examination at the time of the diagnosis of diabetes. Alternatively, a

dilated fundus examination could be replaced by retinal photography. In addition, it is

crucial to obtain a thorough diabetes history, such as HbA1c (glycosylated

hemoglobin), blood pressure profile, lipid profile, smoking status and other diabetes-

related complications, as these risk factors may affect the urgency for referral by

primary eye care providers to ophthalmologists.

To evaluate DR screening and referral patterns in the community, McCarty et al.

conducted a self-administered questionnaire survey among primary eye care providers

and ophthalmologists in 1997 and 1999, in conjunction with the release.3-5 A self-

administered two-page questionnaire was sent to ophthalmologists and optometrists. It

consisted of the primary eye care providers’ demographics, practice location, barriers

to performing dilated fundoscopy, practice patterns and 10 hypothetical case scenarios

on different levels of DR. The researchers found that optometrists were generally

more conservative and less compliant with the guideline recommendation in terms of

screening intervals. The distribution of NHMRC guidelines to optometrists and

ophthalmologists had a minimal effect on management practices. To increase the

awareness of DR among primary eye care providers, the Australian Diabetes Society

disseminated a color information sheet on the management of DR.72 A revised version

of these guidelines was released 10 years later, in 2007. To date, no studies have

evaluated the practice patterns in DR management among primary eye care providers

(optometrists and GPs).

58

2.2.4.2 Diabetic Retinopathy Screening Classification System

Early detection and prompt treatment have been reported to be able to prevent up to

98% of diabetes-related visual impairment.73 At present, various DR screening

programs around the world use different DR classification systems. 9,74,75 For research

studies, the most commonly used retinal photography screening method is the Airlie

House seven standard 30° stereoscopic fields (see Figure 2.6)74, graded using the

ETDRS grading system (see Table 1.2), which consists of six levels of retinopathy for

one eye or 11 levels for both eyes.75 To simplify the DR classification system, the

WHO divided the DR severity into three levels: i) lesions that could be reviewed in

the clinic in a few months; ii) lesions that need a referral as soon as possible; and iii)

sight-threatening retinopathy, which requires immediate referral.76 Due to the

complexity of ETDRS, Wilkinson et al. released the International Clinical Diabetic

and Diabetic Macular Edema Disease Severity Scales in 2003 (see Table 1.1).8 This

classification is much simpler and more user-friendly among primary eye care

physicians, allowing better communication between ophthalmologists and other health

care professionals.

59

Figure 2.6: Airlie House seven standard 30° stereoscopic fields on right eye

Field 1: optic disc; field 2: macula; field 3: lateral macula; field 4: superior temporal;

field 5: inferior temporal; field 6: superior nasal; field 7: inferior nasal

2.2.4.3 Diabetic Retinopathy Screening Modalities

DR screening is performed in various ways by different health care professionals such

as optometrists77,78 and GPs,78,79 including direct ophthalmoscopy,6 dilated slit lamp

biomicroscopy with a hand-held lens (90 D or 78 D),10 mydriatic or non-mydriatic

retinal photography6 and tele-retinal screening.80 In addition, the sensitivity and

specificity of the number of retinal fields performed during DR screening were

evaluated, as well as the type of devices and screeners with variable ophthalmic

experience.

Of the screening methods, 30° stereoscopic seven-field fundal photography by a

trained practitioner remains the gold standard for DR screening.9 According to the

National Institute for Clinical Excellence (NICE) guidelines, DR screening tests

should have sensitivity and specificity of at least 80% and 95% respectively, with a

technical failure rate of no more than 5%.81 The technical failure rate has been

60

reported to be higher in non-mydriatic and wide-angle field (e.g., 50°) retinal

photography.82,83 At present, few studies have reported on the technical failure rate of

screening exercises, which may affect the effectiveness and cost-effectiveness of a

particular screening device/method. For unreadable or ungradeable retinal images,

patients will ideally warrant referrals to see ophthalmologists due to the undetermined

DR state and other co-existing visually significant pathologies such as dense cataracts,

rubeotic glaucoma with hazy cornea or vitreous hemorrhages.

Non-mydriatic retinal photography does not require any dilating drops (which can

potentially blur vision for up to six hours and precipitate acute angle closure in

patients with narrow iridocorneal angles). Hence, it is a popular screening technique

in the primary eye care setting. Nevertheless, non-mydriatic retinal photography has

drawbacks, including a higher technical failure rate resulting from media opacity or

small pupils, and difficulty in obtaining stereoscopic views. Various studies have

reported on the sensitivity and specificity for the detection of any DR, PDR or STR by

practitioners with varied ophthalmic experience, including research associates,

physicians, optometrists using different reference standards such as seven-field

stereoscopic view, dilated indirect and biomicroscopy (78D) by ophthalmologists and

clinical assistants.82,83 One study showed sensitivity and specificity up to 98% and 90%

respectively for non-mydriatic retinal photography in detecting threshold

retinopathy.84 However, very few studies could reproduce similar detection rates.

Further, Lin et al. reported sensitivity of 78% and specificity of 86% in DR needing

referral for non-mydriatic monochromatic digital photographs, with reference to

stereoscopic seven-field retinal photography,83 demonstrating another potential

modality for DR screening.

61

Conversely, mydriatic retinal photography is more reliable and allows not only better-

quality retinal images, but also a lower technical failure rate. It has been shown to

have a minimum sensitivity of at least 80% in the detection of any grade of DR.81 For

STR, sensitivity and specificity increased to 97% and 92% respectively. Despite the

significant increase in the detection rate of DR for mydriatic compared to non-

mydriatic retinal photography, the safety of pupil dilation remains one of the fearful

complications among primary eye care physicians. The incidence of acute angle-

closure glaucoma (AACG) caused by pupil dilation is up to six per 20,000 people, and

narrow angles have also been shown to be a poor predictor of the likelihood of

mydriatic-induced AACG.85 Compared to tropicamide alone, a combination of 2.5%

phenylephrine and 1% tropicamide has been shown to better dilate pupils.86

2.2.4.3 Novel Imaging Modalities for Diabetic Retinopathy Screening

Optical coherence tomography (OCT) is a non-invasive, noncontact imaging modality

that uses low-coherence interferometry to measure optical wave reflectivity and

capture cross section image of retina. Spectral domain OCT (SD-OCT, Spectralis,

Heidelberg Engineering, Germany) and swept source optical coherence tomography

(SS-OCT, Topcon Medical System, Japan) are the two main OCT devices in the

market and they are useful in detecting diabetic macular edema (DME) and clinically

significant macular edema (CSME). The use of OCT may reduce the false positive

referrals of patients suspected with DME/CSME, based on the presence of a few

microaneurysms or hard exudates around the macula area. Furthermore, optical

coherence angiography (OCA) is the latest OCT device that has the ability to analyze

blood flow and the retinal vascular bed without the need of intravenous contrast. This

device is potentially useful to prognosticate the visual potential of people with

62

diabetic retinopathy, based on the amount of capillary dropout around the macula

area.87

Ultra wide-field (UWF) fundus imaging technology has been implemented for DR

screening. It is able to capture a 200° wide-field image in a single photograph by

combining an ellipsoid mirror with a scanning laser ophthalmoscope. This technology

has been incorporated into various devices, including Optos (Marlborough, MA),

Optomap 200Tx, Daytona imaging systems, and Heidleberg Engineering (Carlsbad,

CA). As compared to the standard imaging, the ultrawidefield imaging was reported

to improve DR diagnostic accuracy by 15% to 17% by detecting more peripheral

lesions, with a lower technical failure rate of 3% and quicker image evaluation

time.88,89 This screening modality may improve the physicians’ ability to diagnose and

manage diabetic eye disease, given that 10 to 15% of standard fundus images captured

in multiple retinal locations is incorrect.89

Volk Pictor (Volk Optical, Inc., USA) is a light and portable digital imaging device

that provides a variety of imaging capabilities with interchangeable modules. It allows

digital still and video images to screen for DR. In addition, a few other imaging

modalities such as the adaptive optics,90-92 retinal function imager93,94 and metabolic

imaging95 of the retina had been described to help with DR diagnosis and

prognostication. Nevertheless, more research is required to further evaluate the

clinical and cost-effectiveness of these devices in the setting of DR screening.

63

SECTION 2: EVALUATION OF DIABETIC RETINOPATHY SCREENING

PRACTICES IN THE PRIMARY EYE CARE SETTING IN AUSTRALIA

3. CHAPTER 1: DIABETIC RETINOPATHY MANAGEMENT BY

AUSTRALIAN OPTOMETRISTS

3.1 Summary

This is the first tenth-year survey on the current DR screening and management

practices of Australian optometrists following the release of the 1997 NHMRC

Diabetic Retinopathy Management Guidelines. It is a cross-sectional national survey

in which a self-administered questionnaire was sent to 1,000 optometrists throughout

all states during 2007/2008, inquiring on the use of retinal cameras, screening

practices/attitudes and behavior in DR management. A total of 568 optometrists (57%)

responded to the survey. Patients’ unpreparedness to drive post-dilation (51%) and the

fear of angle-closure glaucoma (13%) were the two main barriers to optometrists

performing dilated ophthalmoscopy. Those who had a strong desire to screen for DR

were more likely to use a retinal camera (p<0.005). The use of a retinal camera was

significantly associated with increased confidence in detecting clinical signs of DR,

including macular edema (p<0.001). Optometrists who read the guidelines at least

once were 2.5 times (p<0.001) more likely to have confidence in detecting macular

edema than those who had never read the guidelines. Although they may be confident

in diagnosis and may use retinal cameras for screening, nearly 60% of optometrists

would not refer patients with macular edema to an ophthalmologist. In conclusion, the

management of macular edema by Australian optometrists needs improvement,

64

despite their strong self-reported desire for involvement in DR. The use of retinal

cameras and the promotion of the 2008 NHMRC guidelines should be encouraged to

improve overall optometric DR management, particularly with macular edema.

3.2 Introduction

The prevalence of diabetes mellitus is growing rapidly worldwide.1 In 2000, there

were approximately 941,000 Australians living with diabetes and it is estimated that

by 2030, this will rise to 1.6 million.1 Diabetic retinopathy (DR) occurs in 25% of

patients with diabetes in Australia.2 As 98% of visual impairment secondary to DR

can be prevented by timely treatment, early detection is crucial.73

As the primary eye care providers, the optometrists and general practitioners play an

enormous role in diabetic retinopathy screening in the community. As part of routine

DR screening, the National Health and Medical Research Council (NHMRC)

Guidelines96 recommended that all examiners should assess patients’ best corrected

visual acuity and perform dilated fundus examination at time of diagnosis of diabetes.

Alternately, the dilated fundus examination may be replaced by retinal photography.

Additional information such as HbA1c (glycosylated hemoglobin), blood pressure

profile, lipid profile, smoking status and other diabetes-related complications may

also help in determining the urgency for referrals.

Australian optometrists have previously been surveyed in 1999 and 2001 3,4 following

the release of the original 1997 NHMRC guidelines on DR management.97 A revised

version of these guidelines was released in late 2008.96 However, to date no published

studies have examined the long-term impact of these guidelines on DR screening and

management practices among Australian optometrists.

65

Our aim was to identify any changes in DR screening and management practices that

have occurred over the last decade following the release of these national guidelines.

This will provide information that will guide the implementation of the revised

guidelines, as well as establishing updated data for future evaluation.

3.3 Methods

We conducted a cross-sectional survey of currently practicing Australian optometrists.

A random sample of 1,000 optometrists was selected from the Optometrists

Association of Australia membership database (4,414 members). A self-administered

three-page questionnaire, an information pamphlet about the objectives of this study

and a postage-paid return envelope were mailed to each selected optometrist in

November 2007. A repeat mail-out of surveys to non-respondents was conducted after

three months to maximize responses. The University of Western Australia Human

Research Ethics Committee approved this study.

The questionnaire used for this study was adapted from two previous surveys

conducted by McCarty et al 3,4 to allow temporal comparison regarding DR

management practices by Australian optometrists. The survey instrument comprised

questions relating to general professional and practice details; and DR screening

attitudes and practices (e.g. perceived barriers and estimated frequency of performing

dilated fundoscopy on diabetic patients, confidence in detecting sight-threatening DR,

desire to participate in community screening and perceived need for further education

on DR).

Optometrists were surveyed about their management practices using 12 hypothetical

clinical scenarios. The first seven scenarios involved patients of different ages (7, 18

and 60 years of age), varying diabetic treatment (diet alone, oral hypoglycemic agent

66

or insulin) and who had no DR detected at their first visit. The last five scenarios

focused upon DR management following the detection of microaneurysms, retinal

hemorrhages, new vessels formation and macular edema. Optometrists were given

five choices of performing dilated fundoscopy in less than 6 months, 1 year, 2 years, 5

years or prompt referral to an ophthalmologist.

All participants remained anonymous throughout data collection and analyses.

Responses were analyzed using Stata 10.0 (StataCorp, College Station, TX) with

significance set at p<0.05 for all analyses. Descriptive statistics were calculated for all

continuous variables. Relationships between categorical variables were explored using

Pearson Chi-square tests. Multivariate logistic regression models were used where

appropriate to explore outcomes of interest (such as the use of a retinal camera and

confidence of detecting DR signs and diabetic macular edema) while controlling for

possible confounding factors of practice location, place of training and years of

practice.

3.4 Results

A total of 568 (57%) optometrists currently practicing in Australia responded to the

survey. Demographic characteristics of the respondent optometrists are shown in

Table 3.1. Our sample size was 13% of the total optometrists practicing in Australia

(total optometrists = 4,414) and the percentage of state and urban/rural distribution of

the respondent optometrists was reasonably comparable to the data published by the

Australian Institute of Health and Welfare (AIHW) in 2006.98 More than 80%

reported receiving a copy of the 1997 NHMRC guideline while only 65% reported

having read them at least once.

67

Of the ophthalmic equipment used by optometrists to examine patients with diabetes,

the direct ophthalmoscope was most frequently used (72%), followed by slit lamp

biomicroscopy (65%), binocular indirect ophthalmoscope (56%) and retinal camera

(51%). Almost 15% of optometrists never performed direct ophthalmoscopy whereas

55% of optometrists used a retinal camera in their practices on more than half of their

diabetic patients.

Table 3.2 shows the perceived barriers to optometrists not performing dilated

ophthalmoscopy. Patients’ unpreparedness to drive (51%) and a fear of precipitating

angle closure glaucoma (13%) post dilation were the two leading reported ‘moderate’

to ‘major’ barriers. Only a small number of optometrists reported a lack of confidence

in detecting changes (2%) and uncertainty about DR management (1%) as moderate

to major barriers to performing dilated fundoscopy.

Table 3.3 summarizes reported current practices, examinations procedures and routine

enquiry of risk factors by optometrists. About 90% of optometrists reported either

‘often’ or ‘almost always’ performing dilated fundoscopy on patients with known

diabetes while only 23% would routinely perform dilated fundoscopy on patients

without any history of diabetes or glaucoma. Two thirds always questioned about a

positive diagnosis of diabetes in patients older than 40 years. As part of routine

follow-up for patients with diabetes, 95% of optometrists ‘often’ or ‘almost always’

enquired about blood glucose level, as well as factors such as assessment of blood

pressure, cholesterol level, smoking status and importance of risk factors control to

prevent diabetes complications (Table 3.3).

Almost all optometrists reported being ‘often’ or ‘always’ confident in detecting

microaneurysms (93%) and retinal hemorrhages (97%) but fewer were confident in

68

detecting new vessel formation (85%). More than 50% of the optometrists were ‘often’

or ‘always’ unsure in detecting macular edema.

Most optometrists (95%) used a recall system to follow up patients with diabetes for

examination. However, only half reported that more than 70% of their patients would

return to see them within the suggested time frame. When patients were referred to

ophthalmologists, the majority (83%) of referring optometrists reported that more

than 70% of their patients would see their ophthalmologist within the suggested time

frame. The percentage of optometrists expressing ‘moderate’ to ‘strong’ desire to

screen for, and receive further education regarding diabetic retinopathy was 78% and

72% respectively.

Changes in reported management since 19993

The percentage of optometrists performing dilated fundoscopy on diabetic patients

has increased significantly from 75% in 1999 to 89% in 2009 (p<0.001). In addition,

the confidence in detecting sight threatening DR changes (new vessels elsewhere and

macular edema) improved significantly from 1999 and 2009 (new vessels elsewhere:

75% to 85%, p< 0.01; macular edema: 33% to 47%, p<0.001). Significantly, more

optometrists reported using a recall system in 2009 (95%) compared with 1999 (83%).

From 1999 to 2009, there were no significant changes in the potential perceived

barriers such as fear of inducing angle closure glaucoma post dilation (1999: 17%;

2009: 13%), lack of confidence in detecting DR changes (1999: 4%; 2009: 2%),

uncertainty about DR management (1999: 2%; 2009: 1%) and the desire to screen for

DR (1999: 84%; 2009: 78%). In contrast, there was significantly less desire for

further education about DR diagnosis and management from 1999 to 2009 (84% to

72%, p<0.0001).

69

Responses to the hypothetical clinical scenarios (Table 3.4)

Only 11% of optometrists reported that they would perform dilated fundoscopy on a

seven year-old child newly diagnosed with diabetes in five years while 17% would

refer this child to an ophthalmologist. Optometrists would still recommend a 12-

month examination for diabetic patients with good glycemic control with no signs of

DR compared with the two-year follow-up recommended by the NHMRC Guidelines

(Table 3.3). Nonetheless, the overall percentage of optometrists following the

NHMRC recommended management practices (dilated fundoscopy within one or two

years of diagnosis, or referral to an ophthalmologist) was greater than observed in

1999. In 2009, the majority of optometrists reported that they would refer patients

with severe non-proliferative DR (90%) and proliferative DR (98%) to an

ophthalmologist. However, only 42% would refer patients with macular edema to an

ophthalmologist for prompt investigation and treatment.

From multivariate logistic regression analyses, optometrists who had a strong desire

to screen for DR were almost twice as likely to ‘often’ or ‘almost always’ use a retinal

camera to examine patients with diabetes after controlling for previous training

location, duration and location of current practice (OR=1.98, 95%CI=1.27-3.10,

p<0.005). When controlled for reading the guidelines, previous training location,

duration and location of practice, optometrists who ‘often’ and ‘always’ used a retinal

camera were more confident in detecting retinal diabetic changes such as

microaneurysms (OR=5.29, 95%CI=2.22-12.40, P<0.001), new vessels formation

elsewhere (OR=4.62, 95%CI=2.56-8.34, P<0.001) and macular edema (OR=2.49,

95%CI=1.51-4.12, P<0.001).

70

Reading the guidelines at least once was also associated with increased confidence in

detecting macular edema (OR=2.49, 95%CI=1.51-4.12, P<0.001). However reading

the guidelines did not improve referrals for patients with macular edema (OR=1.04,

95%CI= 0.65-1.65, p=0.88). Likewise confidence in detecting macular edema after

controlling for other factors such as previous training location, duration and location

of practice, and use of a retinal camera was not associated with optometric referrals of

patients with macular edema to an ophthalmologist (OR=0.99, 95%CI=0.71-1.39,

p=0.97).

71

Table 3.1: Demographics of the optometrists responding to a survey on diabetic

retinopathy screening

Total number

(%)

State or Territory of practice - 566

New South Wales 195 (34%)

Victoria 160 (28%)

Queensland 97 (17%)

Western Australia 44 (7%)

South Australia 42 (7%)

Tasmania 20 (4%)

ACT 7 (1%)

Northern Territory 1 (<1%)

Number of years of practice

0 - 10

11 - 20

21 - 30

>30

Location of practices

42 (7%)

169 (30%)

260 (46%)

97 (17%)

Metropolitan 359 (64.1%)

Rural 170 (30.4%)

Metropolitan and rural 31 (5.5%)

Location of training

Australia 520 (92%)

United Kingdom 25 (4%)

South Africa 9 (2%)

New Zealand 8 (1%)

United States 2 (<1%)

Canada 1 (<1%)

Table 3.2: Barriers to optometrists performing dilated fundoscopy

Barriers Not a barrier Minor barrier Moderate barrier Strong Barrier

Patients' unpreparedness to drive 68 (12%) 205 (37%) 179 (32%) 107 (19%)

Worry of angle closure glaucoma 305 (55%) 185 (33%) 48 (9%) 22 (4%)

Time consuming 297 (53%) 195 (35%) 61 (11%) 9 (2%)

Lack of dilating drops 539 (97%) 9 (2%) 2 (<1%) 8 (1%)

Lack of ophthalmoscopes 545 (98%) 3 (1%) 3 (1%) 6 (1%)

Lack of confidence in detecting changes 468 (84%) 76 (14%) 11 (2%) 1 (<1%)

Unsure of diabetic retinopathy management 514 (92%) 37 (7%) 6 (1%) 1 (<1%)

Table 3.3: Current optometrists’ management and attitudes to diabetes and diabetic retinopathy

Screening questions and examinations Almost never Sometimes Half the time Often Almost always

On patients >40 years,

1. Routine questioning about diagnosis of diabetes 20 (4%) 56 (10%) 21 (4%) 80 (14%) 389 (69%)

2. Routine dilated fundoscopy without history of diabetes

or glaucoma 102 (18%) 258 (46%) 73 (13%) 56 (10%) 74 (13%)

3. Routine of dilated fundoscopy with history of diabetes 13 (2%) 27 (5%) 22 (4%) 58 (10%) 448 (79%)

Frequency of risk factors enquiries

Blood sugar control 2 (<1%) 3 (<1%) 21 (4%) 95 (17%) 446 (79%)

Blood pressure control 10 (2%) 15 (3%) 77 (14%) 185 (33%) 279 (49%)

Blood cholesterol control 14 (3%) 53 (9%) 121 (21%) 167 (30%) 211 (37%)

Smoking status 48 (9%) 116 (21%) 161 (28%) 140 (25%) 102 (18%)

Advice regarding complications 11 (2%) 33 (6%) 98 (17%) 221 (39%) 202 (36%)

Table 3.4: Optometrists’ management of hypothetical clinical scenarios and specific signs of diabetic retinopathy

Case scenarios

Referral to

Ophthalmologists

Review in

5 years

Review in

2 years

Review in

1 year

Review in

<6 months

If no signs of DR at baseline examination

7 yo - newly diagnosed diabetic† 95 (17%) 63 (11%)* 125 (22%) 183 (33%) 96 (17%)

18 yo - newly diagnosed with DM† 40 (7%) 16 (3%) 141 (25%)* 255 (45%) 113 (20%)

60 yo with good HbA1c control – diet alone† 5 (1%) 1 (<1%) 211 (37%)* 290 (51%) 58 (10%)

60 yo, 10 years diabetes, commenced on OHA† 9 (2%) 0 (0%) 86 (15%)* 374 (66%) 98 (17%)

60 yo, 10 years diabetes with good HbA1c on OHA† 10 (2%) 1 (<1%) 72 (13%)* 417 (74%) 65 (12%)

60 yo, 10 years diabetes with good HbA1c on insulin† 21 (4%) 0 (0%) 31 (6%)* 409 (72%) 105 (19%)

60 yo, poorly controlled HbA1c despite insulin† 79 (14%) 0 (0%) 2 (<1%) 127 (22%)* 360 (63%)

Diabetic Retinopathy Signs

Occasional MAs on peripheries 43 (8%) 14 (3%) 297 (53%) 167 (30%)* 44 (8%)

Macular edema (not clinically significant) 234 (41%)* 6 (1%) 43 (8%) 138 (25%) 139 (25%)

Peripheral MAs and retinal hemorrhages 224 (40%) 0 (0%) 81 (14%) 169 (30%) 93 (16%)*

Extensive MAs, retinal hemorrhages and CWSs 509 (90%)* 0 (0%) 4 (1%) 18 (3%) 36 (6%)

New vessel formation 557 (98%)* 0 (0%) 0 (0%) 1 (<1%) 10 (2%)

† yo: year-old; OHA: oral hypoglycemic agents; HbA1c – glycosylated hemoglobin; Mas – microaneurysms; CWS – cotton wool spots * : Recommended time frame suggested by NHMRC guidelines in 1997

75

3.5 Discussions

Our results indicate that DR management practices of Australian optometrists have improved

since the release of NHMRC guidelines in 1997.97 Compared with the two national surveys

conducted in 1999 and 2001,3,4 more optometrists now perform dilated fundoscopy on

diabetic patients, use recall notices and had greater confidence in detecting and managing DR

changes in their patients. Additionally, nearly 80% of optometrists had ‘moderate’ to ‘strong’

desire to screen for DR. This could significantly reduce the waiting period of diabetic patients

to see an ophthalmologist in the public setting, and may triage the urgency of ophthalmic

review by severity, especially for patients with sight-threatening DR. We also found an

approximate 10% rise in the use of recall notices over the last decade. This helps to ensure

regular eye screening of patients, as a patient reminder system was reported to be an effective

way of enforcing and increasing the patients’ adherence to clinical guidelines.99

Both the 1997 and 2008 NHMRC guidelines6-7 recommend that screening of diabetic children

should start at the time of puberty, with the screening interval determined by the clinical

findings. Those with moderate to severe non-proliferative DR, proliferative DR, or macular

edema warrant prompt referrals to an ophthalmologist. We found that optometrists generally

reviewed diabetic patients with no signs of DR more frequently than recommended.97 The

optometrists should be encouraged to read the guidelines more frequently in order to review

the patients with diabetes in an appropriate time frame to reduce their unnecessary financial

burden.

The responses we obtained regarding the management of diabetic macular edema by

Australian optometrists were of concern. More than 50% of optometrists reported that they

lacked confidence in detecting macular edema and only 40% would refer patients with

macular edema to an ophthalmologist.

76

Although both the use of a retinal camera and reading the guidelines were significantly

associated with increased confidence in detecting macular edema, they were not associated

with appropriate referral of patients with macular edema to an ophthalmologist. Confidence

in detecting macular edema was also not associated with referrals to an ophthalmologist. In

other words, the optometrists were not likely to refer patients with macular edema to an

ophthalmologist despite having read the guidelines, being confident in detecting macular

edema and using a retinal camera. Given macular edema is a major cause of significant visual

impairment, optometrists need to improve their management (confidence to detect and

referrals) of this condition to ensure prompt laser treatment for patients with diabetic

maculopathy. Although early stages of macular edema may be difficult to detect without

indirect ophthalmoscopy, any reduction in visual acuity should raise suspicion and prompt a

referral.

Concern about use of dilating drops inducing angle closure glaucoma seems unwarranted as it

is a rare event (1 in 20,000).100 Promoting the use of non-mydriatic fundus cameras may help.

We found there was a strong association between the frequency of retinal camera use and the

desire to screen, as well as confidence in detecting DR changes. Others have found non-

mydriatic retinal camera fundus photography yielded a reasonable sensitivity (95%) and

specificity (99%).101

The present study provides an overview of DR management by Australian optometrists which

has improved over the last decade following the release of 1997 NHMRC guideline.97 Given

that macular edema causes significant visual impairment in patients with diabetes, further

education about the detection and referral of subjects with macular edema is important. The

use of retinal cameras and promotion of the new 2008 NHMRC guidelines7 should be

encouraged to improve the overall optometric DR management and reduce the incidence of

77

this preventable blinding disease in the future. A further survey may help assess the impact of

the new 2008 NHMRC guidelines, especially the management of diabetic macular edema by

Australian optometrists.

78

4. CHAPTER 2: DIABETIC RETINOPATHY MANAGEMENT BY

AUSTRALIAN GENERAL PRACTITIONERS

4.1 Summary

This is the first nationwide survey on DR screening and management practices among

Australian GPs. A self-administered questionnaire on DR management was mailed to 2,000

rural and urban GPs across Australia in 2007–2008. Only 29% of GPs had read the NHMRC

guidelines at least once, and 41% had a ‘moderate’ to ‘strong’ desire to screen for DR. The

majority of GPs (74%) reported not routinely examining their diabetic patients for DR. Lack

of confidence in detecting DR changes (86.4%) and time constraints (73.4%) were the two

major barriers to GPs performing dilated fundoscopy on diabetic patients. Given that access

to optometry is not evenly distributed across the country and that ophthalmology is under-

resourced to review all but those with DR and the potential for complications, GPs are still

the most proficient health care providers to manage and screen for DR in the community.

4.2 Introduction

Diabetes mellitus is rising in prevalence within Australia and internationally with estimates

indicating that the global prevalence of diabetes will double by 2030.1 In Australia, the

prevalence of diabetes is 8% in adult men and 6.5% in adult women,25 of which 1 in 4 people

with diabetes will be diagnosed with diabetic retinopathy (DR).2 Early detection and prompt

treatment can prevent 98% of visual impairment.73

Primary health care providers such as general practitioners (GPs) and optometrists comprise

the ‘front line’ in service provision and have a crucial role in screening for DR in the

community. However, a 1994 survey of Victorian GPs found over half had little interest in

79

DR screening and that most routinely examined less than half of their patients with diabetes

for DR.102 The National Health and Medical Research Council (NHMRC) released clinical

practice guidelines for DR in 1997,97 outlining evidence based DR management practices and

encouraging physicians to increase the DR screening rate in order to reduce DR-related visual

impairment. Nonetheless, a subsequent Victorian survey reported that despite the NHMRC

guidelines, 48% of GPs still had little or no desire to screen for DR.5

Following the local Victorian study,7 our study is the first national survey of Australian GPs

on DR management. The purpose of our study is to investigate the current Australian GPs

management practices and attitudes towards DR and its management since the 1997 NHMRC

guidelines.97 It also serves as the baseline survey findings prior to the release of the new 2008

NHMRC guidelines.96

4.3 Methods

We surveyed a random sample of 2,000 currently practicing Australian GPs selected from the

Royal Australian College of General Practitioners (RACGP) membership database. A

package consisting of a self-administered three-page questionnaire, an information leaflet

detailing the aim of this study and a postage-paid return envelope was mailed to selected GPs

in December 2007. A repeat mail-out of surveys to non-respondents was conducted after

three months to maximize responses. This study was approved by the University of Western

Australia’s Human Research Ethics Committee.

The questions related to general professional and practice details (location of previous

training, duration and location of practices - rural or metropolitan); frequency of

measurement of HbA1c, blood pressure, cholesterol, smoking status; perceived barriers to

DR screening (time factors, patient refusal, fear of angle closure glaucoma, lack of

confidence in detecting and managing DR and lack of dilating drops and ophthalmoscopes in

80

practices); frequency of referral of diabetic patients to optometrists and ophthalmologists for

DR screening; eye examination routine (measurement of visual acuity, dilated or non-dilated

fundoscopy); and their desire to undertake DR screening.

GPs were also required to respond to five case scenarios regarding the management of DR

clinical signs (microaneurysms, retinal hemorrhages, cotton wool spots, new vessels

formation and presence of hard exudates near the macula). A further seven hypothetical case

scenarios involving patients of varying ages (7, 18 and 60 years old), diabetic management

(diet control, oral hypoglycemic agents and insulin) and glycemic control (poorly and well

controlled) who had no signs of DR detected at baseline examination were also included.

Analyses were performed using SPSS version 17 (SPSS, Chicago, IL, USA). Descriptive

statistics (mean, standard deviation) were calculated for continuous variables. Relationships

between categorical variables were explored using Chi-square tests. Additionally, we used a

multivariate logistic regression model to study the possible factors relating to GPs confidence

to detect DR clinical signs such as their years of practice, previous training location and

reading the NHMRC guidelines.

4.4 Results

There were 429 (21%) respondents to the survey and their demographics are displayed in

Table 4.1. Almost half reported having received the 1997 NHMRC guidelines for DR

management, however of these, only 29% had read the guidelines at least once. Apart from

the NHMRC guidelines, GPs also reported using other resources such as the RACGP

guidelines,103 Endocrine Therapeutic Guidelines,104 National Prescribing Service

Guidelines,105 American Diabetic Association Guidelines,106 various online websites and

diabetes-focused peer-reviewed journals.

81

Almost all GPs reviewed their diabetic patients’ blood pressure (92%) and HbA1c (99%) at

least every six months (Table 4.2). Assessment of lipid profile, smoking status and advice on

diabetes complications were conducted less frequently by the respondent GPs (Table 4.2).

Nearly 75% of GPs did not routinely examined their diabetic patients for DR and of those, 89%

of them would refer their diabetic patients to see an ophthalmologist within two years. More

GPs ‘usually’ and ‘always’ performed non-dilated (61%) than dilated fundoscopy (13%) to

detect DR signs and only 65% of GPs ‘usually’ or ‘always’ checked visual acuity.

Only 21% of GPs responded that they were ‘often’ or ‘almost always’ confident in detecting

DR changes. Lack of confidence in detecting DR changes (86%) and time constraints (73%)

were the primary barriers to performing dilated fundoscopy on diabetic patients for GPs.

Additional reported barriers included patient refusal, concern of angle closure glaucoma, lack

of dilating drops and uncertainty surrounding DR management (Table 4.3).

Less than half of GPs (41%) expressed ‘moderate’ to ‘strong’ desire to screen for DR in

community setting. Nearly all GPs (91%) referred diabetic patients to ophthalmologists one

to two yearly whilst 68% referred their diabetic patients to optometrists in the first instance.

A fifth of GPs never referred any diabetic patients to see an optometrist. Rather, they all

preferred to refer their patients to see an ophthalmologist every one to two yearly. A small

number of GPs (8%) would only refer their patients to an ophthalmologist or optometrist if

visual symptoms were present. Nearly 80% of GPs felt that their patients would be compliant

to see an ophthalmologist should it be necessary.

Table 4.4 shows responses to the management of specific DR signs and hypothetical case

scenarios. Most GPs (63%) would refer their patients with occasional microaneurysms with

normal vision within a month to see an ophthalmologist while only 3% were confident to not

refer these patients to the ophthalmologist. For patients with hard exudates near the macula

82

and normal vision, 87% GPs indicated that they would refer them to an ophthalmologist

within one month. Following the detection of peripheral microaneurysms and retinal

hemorrhages, 95% GPs would refer their patients to see an ophthalmologist within the

recommended time frame. For new vessel formation, nearly all GPs would refer within a

month to an ophthalmologist.

A majority of GPs (82%) would inappropriately refer a 7 year-old child with diabetes and no

signs of DR for regular eye screening while 18% would refer such patients in five years. As

shown in Table 4.3, the majority of the GPs would refer patients of varying ages (7, 18 and

60 year-old), diabetic management (diet control, oral hypoglycemic and insulin) and

glycemic control (well and poorly controlled) elsewhere for regular eye screening even

without any signs of DR rather than perform the review themselves.

GPs desire to screen for DR in the community were strongly associated with reading the 1997

NHMRC guidelines at least once (χ2=17.64, p<0.001) and confidence in detecting DR

clinical changes (χ2=28.5, p<0.001). In multivariate logistic regression analyses, GPs who

were confident in detecting DR clinical signs were 3.31 times more likely to have desire to

screen for DR after controlling for reading the guidelines at least once, their years of practice

and previous training location (OR=3.31, SE=0.85, 95%CI=2.00-5.47, p<0.001). However,

the frequency of GPs performing visual acuity measurement and dilated fundoscopy as part

of the routine eye examination for patients with diabetes was not associated with reading the

NHMRC guidelines.

Additionally, GPs confidence in detecting DR changes was strongly associated with having

read the guidelines at least once (χ2=7.48, p<0.01); GPs in practice more than 15 years (χ

2=7.71, p<0.01); and Australian trained GPs (χ2=3.88, p<0.05). When controlled for years

of practice and previous training location, GPs who had read the guidelines at least once were

83

2.11 times more likely to have confidence in screening for DR (OR=2.11, SE=0.54,

95%CI=1.27-3.50, p<0.005). The location of the GP practice (rural or metropolitan) was not

associated with confidence to detect DR changes, desire of DR screening and the frequency

of referral to ophthalmologists or optometrists.

Table 4.1: Demographics of general practitioners responding to a survey on diabetic retinopathy screening

Total

number (%)

State or Territory of practice

New South Wales 101 (24%)

Victoria 122 (29%)

Queensland 74 (18%)

Western Australia 68 (16%)

South Australia 40 (9%)

Tasmania 11 (3%)

Australian Capital Territory 5 (1%)

Northern Territory 1 (<1%)

Number of years of practice

0 – 10

11 - 20

21 - 30

>30

Locality of practices

47 (11%)

61 (14%)

136 (32%)

185 (43%)

Metropolitan 279 (66%)

Rural 143 (34%)


Australia 343 (81%)

United Kingdom 39 (9%)

India 12 (3%)

New Zealand 6 (1%)

Others 26 (6%)

84

Table 4.2: Current general practitioners management and attitudes to diabetes and

diabetic retinopathy

At diabetes follow-up Never and rarely Sometimes

Usually

and Always

1. Visual acuity measurement 30 (13.5%) 49 (22%) 144 (64.5%)

2. Fundsocopy (undilated) 54 (25.1%) 29 (13.5%) 132 (61.4%)

3. Dilated fundoscopy 155 (79.0%) 16 (8.2%) 25 (12.8%)

Percentage of diabetic patients

examined for diabetic retinopathy

None

197 (46.4%)

Some

119 (28%)

100%

109 (25.6%)

Frequency of patients with diabetes Yearly or less More than yearly

were reviewed 137 (57.3%) 102 (42.7%)

Frequency of risk factor

management

Six monthly or

less

More than six

monthly

HbA1c 387 (92.2%) 33 (7.8%)

Blood pressure 416 (98.6%) 6 (1.4%)

Cholesterol 245 (58.2%) 176 (41.8%)

Smoking 278 (65.9%) 144 (34.1%)

Advice regarding complications 334 (79.5%) 85 (20.5%)

Table 4.3: Barriers to general practitioners performing dilated fundoscopy

Barriers

Not/Minor barrier

Moderate/Major barrier

No confidence in detecting diabetic retinopathy signs 55 (13.6%) 349 (86.4%) Time consuming 109 (26.7%) 300 (73.3%) Patients' refusal to dilation 145 (36.5%) 252 (63.5%) Worry of angle closure glaucoma 223 (55.8%) 177 (44.2%) Lack of dilating drops 251 (62.4%) 151 (37.5%) Unsure of diabetic retinopathy management 328 (82.2%) 71 (17.8%) Lack of ophthalmoscopes 396 (97.8%) 9 (2.2%)

85

Table 4.4: General practitioners management of hypothetical clinical scenarios and specific signs of diabetic retinopathy

Clinical scenarios

Appropriate referral

Inappropriate referral

Recommended referral time frame (2008)*

Recommended referral time frame (1997)**

Occasional microaneurysms with normal vision 101 (24.7%) 308 (75.3%) 1 year Non-urgent, routine referral Hard exudates near macula with normal vision 348 (86.6%) 54 (13.4%) 1 month or less Refer urgently Peripheral microaneurysms and retinal hemorrhages 384 (94.8%) 21 (5.2%) 6 months or less Refer urgently Extensive microaneurysms, retinal hemorrhages and cotton wool spots (all peripherally) 407 (99.5%) 2 (0.5%) 3 months or less

Refer urgently

New vessel formation 407 (99.7%) 1 (0.3%) 1 month or less Refer urgently If no signs of DR at baseline examination, 7 yo - newly diagnosed diabetic 338 (80.3%) 83 (19.7%) At puberty Refer in 5 years 18 yo - newly diagnosed diabetic*** 351 (84.0%) 67 (16.0%) 1 to 2 years Yearly, no later than 2 yearly 60 yo with good HbA1c control – diet**** 284 (67.2%) 139 (32.8%) 2 years Yearly, no later than 2 yearly 60 yo, 10 years diabetes, commenced on OHA*** 361 (84.9%) 64 (15.1%) 1 to 2 years Yearly, no later than 2 yearly 60 yo, 10 years diabetes with good HbA1c on OHA**** 289 (68.5%) 133 (31.5%) 2 years Yearly, no later than 2 yearly 60 yo, 10 years diabetes with good HbA1c on insulin**** 269 (63.4%) 155 (36.6%) 2 years Yearly, no later than 2 yearly 60 yo, poorly controlled HbA1c despite insulin 306 (72.0%) 119 (28.0%) 1 year Yearly, no later than 2 yearly

yo – year-old, OHA- oral hypoglycemic agents, HbA1c – glycosylated hemoglobin

* Referral time frame recommended by 2008 NHMRC Clinical Practice Guideline: Diabetic Retinopathy Management8

** Referral time frame recommended by 1997 NHMRC Clinical Practice Guideline: Diabetic Retinopathy Management6

*** Should HbA1c is unavailable, all diabetic patients should be referred within 1 to 2 years

**** Patients with good HbA1c control and no signs of DR are recommended to undergo 2 yearly fundus examinations for diabetic retinopathy

86

4.5 Discussion

In response to World Health Assembly resolution on the elimination of avoidable

blindness,115 the National Eye Health Framework was developed following the Australian

Health Ministers’ Conference in 2005.116 It identified five potential key areas which may help

prevent avoidable blindness and low vision. Two of these related to increasing early detection

and improving access to eye health care services.116 At the time of the release of NHMRC

guidelines on DR management in 1997,6 the Victorian GP DR survey showed that half the

GPs expressed a desire to regularly screen for DR in patients attending their practice.5

Unfortunately, we have found that since the last survey, even fewer GPs (41%) expressed a

desire to screen for DR. This is of concern given that primary health care screening provides

an excellent opportunity to differentiate those patients who require specialist ophthalmic care

with others who can continue to be managed by their GPs.

We also found that similar to the prior Victorian study,5 GPs’ perceived the lack of

confidence in detecting clinical DR signs was the leading barrier to performing dilated

fundoscopy. A fear of inducing angle closure glaucoma (a rare 1:20,000 event post dilation100)

was perceived as another major barrier to performing dilated fundoscopy. Despite the low

numbers of confident GPs in detecting DR clinical signs (21%), most GPs were generally

confident and proficient to manage DR once DR changes were detected based on the

hypothetical clinical scenarios. Given that the GPs who had confidence to detect DR clinical

signs were three times more likely to have strong desire to screen for DR, more education

will need to be directed towards DR clinical signs detection to increase the DR screening rate

in Australian community.

In a similar survey of optometrists, we found that the optometric DR management

was generally more inferior to that of GPs.15 Nearly 60% of optometrists would not refer

87

patients with diabetic maculopathy to see an ophthalmologist and 10% of them would not

refer patients with severe non-proliferative DR. The comparison of these two surveys

highlighted that GPs are still the most proficient health care providers to manage/screen for

diabetic retinopathy in the community.

From the same optometrists study,117 the retinal cameras were shown to have increased

optometrists’ confidence to detect DR changes such as microaneurysms, retinal hemorrhages,

new vessels formation and macular edema. It is unknown if GPs would feel more confident

with retinal photographic screening but a pilot study of photographic screening by GPs for

DR found that they were willing to expand their roles into DR screening if such infrastructure

is readily accessible.118 That study also showed that GPs had good diagnostic accuracy

(sensitivity 87%; specificity 95%) for DR. Given that it was a pilot study with a relatively

small sample size, a larger study using cheaper, portable retinal cameras would provide more

insight into the enthusiasm and effectiveness of retinal photographic screening in primary

care.

The updated 2008 NHMRC Guidelines suggests that mydriatic retinal photography is the

most effective DR screening tool with a sensitivity of at least 80%.8 Cheaper retinal cameras

are available in the market now and should be encouraged in primary health care, especially

for large practices. Retinal photographs taken by staff could be instantly interpreted by the

GP to determine the need for referrals to ophthalmologists during the general diabetes

consultations.

The strength of our study was being the first nationwide GP DR survey. Our study findings

not only revealed the current management practices of Australian GPs on DR management

but also could be utilized as baseline findings to assess the impact of the new 2008 NHMRC

guideline87 in the future. On the other hand, our study findings were limited by several factors.

88

Firstly, low response from the GPs (21.3%) despite sending repeat mail-out of surveys may

affect generalizability of our results on Australian GPs DR management. In addition, we have

no data regarding the current use of retinal cameras among GPs, their previous education on

eye training and the size of their practice which may possibly affect the GPs confidence in

screening for DR.

IMPLICATIONS FOR GENERAL PRACTICE

In conclusion, Australian GPs in general reported sound management on sight-threatening

DR. Given that the access to optometry is not evenly distributed across the country, and

ophthalmology is under resourced to review all but those with DR and the potential for

complications, GPs are still the most proficient health care providers to manage and screen

for DR in the community. In the absence of retinal photography, dilated ophthalmoscopy is

still the most convenient and effective method to examine for DR and thus, it is a basic

clinical examination skill that should be encouraged for all GPs.

89

SECTION 3: RETINAL STILL PHOTOGRAPHY: NOVEL, EASY-TO-OPERATE

AND EFFECTIVE DIAGNOSTIC DEVICES FOR DIABETIC RETINOPATHY

SCREENING IN THE PRIMARY EYE CARE SETTING IN AUSTRALIA

5. CHAPTER 3: LIGHT AND PORTABLE DEVICE FOR DIABETIC

RETINOPATHY SCREENING

5.1 Summary

The previous surveys showed that GPs lacked the confidence and desire to screen for DR in

the primary eye care setting, despite having sound knowledge of DR management. Given that

the use of retinal cameras may be associated with increased desire/confidence to screen for

DR in the community, we performed a study to validate the use of a novel, light and portable

multipurpose ophthalmic imaging device called EyeScan (OIS, CA, US) for DR screening.

This single-center study evaluated the diagnostic device in 136 patients (272 eyes) recruited

from the DR screening clinic at Royal Perth Hospital, WA. All patients underwent three-field

(optic disc, macular and temporal view) mydriatic retinal digital still photography captured

by EyeScan and FF450 plus (Carl Zeiss, Meditec, US), and they were subsequently examined

by a senior consultant ophthalmologist using the slit lamp biomicroscopy (reference standard).

All retinal images were interpreted by a consultant ophthalmologist and a medical officer.

The main outcome measures were the sensitivity, specificity and kappa statistics of EyeScan

and FF450 plus, with reference to the slit lamp examination findings by a senior consultant

ophthalmologist. For the detection of any grade of DR, EyeScan had sensitivity and

specificity of 93% and 98% (ophthalmologist) and 92% and 95% (medical officer)

respectively. In contrast, FF450 plus images had sensitivity and specificity of 95% and 99%

90

(ophthalmologist) and 92% and 96% (medical officer) respectively. The overall kappa

statistics for DR grading for EyeScan and FF450 plus were 0.93 and 0.95 (ophthalmologist)

and 0.88 and 0.90 (medical officer) respectively. Given that EyeScan requires minimal

training to use and has excellent diagnostic accuracy in screening for DR, it could be utilized

by primary eye care providers to widely screen for DR in the community.

5.2 Introduction

The worldwide prevalence of diabetes is estimated to double to 366 million (4.8%) by 2030.1

People with diabetes should receive regular eye screening, as early detection could prevent

diabetic retinopathy-related visual impairment.73,119,120 Retinal still photography remains the

mainstay screening tool in various diabetic retinopathy screening programs worldwide, and it

has been shown to increase the primary eye care providers’ confidence and desire to detect

diabetic retinopathy in the community.121 To date, mydriatic retinal photography has been

shown to be the most effective means to detect diabetic retinopathy in a screening setting.6,122

The rising prevalence of diabetes will require implementation of more community screening

programs. This will lead to a rise in health care costs123 and maintenance of an effective

quality assurance system.124 The cost of a retinal camera is and will continue to be an

important factor in the health care cost equation. Numerous retinal cameras (mydriatic and

non-mydriatic) with various functions, including anterior and posterior segment imaging, and

fluorescein angiography, are currently available on the market. However, many are not

inexpensive.

The purpose of our study was to validate the efficacy of the EyeScan (Ophthalmic Imaging

System, CA) machine to screen for diabetic retinopathy in the community. The EyeScan is a

flash camera which weighs around 0.9 kilogram (kg). Apart from retinal imaging, this device

also captures the anterior segments of the eye such as the cornea and the lens and this is

91

especially important in determining the causes of some ‘ungradeable’ retinal still images

secondary to cataracts or pterygium. It has a maximum image capture rate of 3 images per

second and 5.3 Mega Pixel Sensor. It is able to capture the photos with near infrared or

visible light and has a field of view of 35 degrees. This machine was approved by United

States Food and Drug Administration (US FDA) in 2010 and currently available in United

States, Australia and other European countries. Given that this is an extremely portable

device which can be carried around in a suitcase, it will be a suitable imaging device in

routine, mobile and tele-retinal diabetic retinopathy screening clinics in both metropolitan

and rural areas.

5.3 Methods

Design

This single center study aimed to validate and compare a new diagnostic device, EyeScan,

with the currently accepted diagnostic device FF450 plus (Carl Zeiss Meditec, Inc., North

America), with reference to slit-lamp examination by a consultant ophthalmologist.

Study sample

We enrolled 136 consecutive patients (272 eyes) from the diabetic retinopathy screening

clinic of Royal Perth Hospital, Western Australia, into our study. All patients signed an

informed consent for participation. This study was approved by the Royal Perth Hospital

Ethics Committee.

Patients’ characteristics and diabetes history

We collected information on patients’ demographics (e.g., age and ethnicity), current and

past ocular history, diabetes history [e.g., type, duration, glycated hemoglobin (HbA1c) level,

macro- and microvascular complications (cerebrovascular accidents, ischemic heart disease,

92

peripheral vascular disease, nephropathy, and neuropathy)] and other associated

cardiovascular risk factors (e.g., smoking status, blood pressure and lipid profile).

Screening process

Upon arrival at the screening clinic, all patients received pupil-dilating drops (2.5%

phenylephrine and 0.5% tropicamide) in both eyes. They underwent 3 sets of retinal

examinations in the following orders: (i) non-stereo color retinal still photography (FF450

plus); (ii) non-stereo color retinal still photography (EyeScan); and (iii) slit-lamp

biomicroscopy examination with a 78 diopter lens by a senior consultant ophthalmologist.

Retinal still photography using the EyeScan and FF450 plus was performed by a medical

officer and a retinal photographer, respectively. In order to assess the usability of the

EyeScan device, a medical officer with no previous experience in ocular imaging was

recruited and compared to the retinal photographer who has had 10 years experience in

performing retinal still photography for diabetic retinopathy screening. Three retinal fields

(optic disc, macula, and temporal views) were captured using both devices, and the images

were subsequently de-identified, randomized, and interpreted by a consultant ophthalmologist

and a medical officer (who has graded more than 1000 color fundus photos of patients with

diabetes) on a 27-inch iMac (Apple, CA, USA) with a display resolution of 2560 x 1440

pixels in a dimly lit room.

The retinal digital still images of the EyeScan and FF450 plus were all downloaded in Joint

Photographic Experts Group (JPEG) format. The color resolution of the still images of

EyeScan and FF450 plus were 640 x 480 pixels and 2392 x 1944 pixels, respectively. The

field angle of EyeScan and FF450 plus was 35 degrees and 30 degrees respectively, centering

on optic disc, macular and temporal views.

93

The retinal images were graded on the basis of the presence or absence of diabetic

retinopathy signs (microaneurysms, retinal hemorrhages, hard exudates, cotton wool spots,

venous beading, intraretinal microvascular abnormalities, new vessel formation, and

preretinal/vitreous hemorrhage) using the International Clinical Diabetic Retinopathy

Severity Scale (Table 3.1). The retinal photographs were classified as “unacceptable,”

“average,” or “excellent” depending on their quality; the retinal photograph was graded as

“unacceptable” if more than one-third of it was “blurred” or “uninterpretable.”

Statistical analyses

We calculated the sensitivity and specificity of the 2 imaging devices (EyeScan and FF450

plus) in detecting and grading diabetic retinopathy, with reference to slit-lamp examination.

In addition, Cohen’s kappa coefficient was utilized as a measure of agreement for diabetic

retinopathy signs and grading using the 2 types of imaging devices. The technical failure rate

was defined as the fraction of the “unacceptable” retinal images captured by both devices;

such images were excluded from the calculation for sensitivity, specificity, and kappa

coefficient. All data were analyzed using SPSS version 17 (SPSS, Chicago, IL, USA).

94

5.4 Results

Patients’ demographics and clinical characteristics

A total of 136 patients (272 eyes) participated in our study. The mean (± standard deviation)

age of participants was 53.9 ± 15.3 years, duration of diabetes was 13.9 ± 9.9 years, and

HbA1c was 8.0% ± 1.7%. Among the recruited patients, 74% (n = 101) were Caucasians, 17%

(n = 23) were Asians, and 9% (n = 12) were from other ethnic groups. Of these, ninety-six

patients (71%) had type 2 diabetes. Figure 5.1 shows the color fundus images captured by

EyeScan and FF450 plus.

The best corrected visual acuity of 240 eyes (88%) was 6/6 or 6/9, while that of 23 eyes (9%)

was between 6/12 and 6/36, and that of 9 eyes (3%) was 6/60 or less. Of the consecutively

recruited eyes, nearly 35% had diabetic retinopathy ranging from mild non-proliferative

diabetic retinopathy to proliferative diabetic retinopathy (Table 5.2). Nearly 15% (n = 37) of

eyes had previously received panretinal photocoagulation, and cataracts were diagnosed in 28

eyes (10.3%) on the basis of slit-lamp biomicroscopy examination using Lens Opacities

Classification III.107 Almost 45% (n = 118) of the patients had never undergone any diabetic

retinopathy screening. Of the self reported diabetes-related complications, diabetic

neuropathy (23%, n = 62) and nephropathy (22%, n = 60) were the leading complications

(Table 5.3).

Main outcome measure

Compared with the slit-lamp biomicroscopy examination, EyeScan in detecting any grade of

diabetic retinopathy had a sensitivity and specificity of 93% (95% confidence interval [CI]:

84.9–97.1) and 98.2% (95% CI: 94.3–99.5), respectively, when graded by the

ophthalmologist; however, they were 91.7% (95% CI: 83.2–96.3) and 94.7% (95% CI: 89.9–

95

97.4), respectively for the medical officer (Table 5.4). In contrast, FF450 plus images graded

by the ophthalmologist had a sensitivity and specificity of 95.1% (95% CI: 87–98.4) and 98.8%

(95% CI: 95.4–99.8), respectively whereas for the medical officer, he had a sensitivity and

specificity of 91.9% (95% CI: 83.4–96.4) and 95.9% (95% CI: 91.5–98.2), respectively. For

the detection of sight-threatening diabetic retinopathy (severe non-proliferative diabetic

retinopathy and proliferative diabetic retinopathy), the sensitivity and specificity of images

from both devices (EyeScan and FF450 plus) graded by both readers increased to 100%.

The technical failure rate for EyeScan and FF450 plus were 8.5% and 7%, respectively, and

they were not statistically significant (χ² = 0.23, d.f. = 1, p = 0.63). Of the failed retinal

photographs captured by the EyeScan, 39% (n = 9) were photographs of eyes with cataracts

and 9% (n = 2) with dark fundi; in 52% (n = 12) of the photographs, failure was due to

intolerance to bright flash. On the other hand, the ‘uninterpretable’ Zeiss retinal images were

secondary to cataracts (42.1%, n = 8), dark fundi (10.5%, n = 2) and intolerance to bright

flash (47.4%, n = 9).

The overall kappa statistics for diabetic retinopathy grading for EyeScan and FF450 plus

were 0.93 and 0.95 for ophthalmologist and 0.88 and 0.90 for medical officer, respectively

(Table 5.4). The kappa coefficients for all diabetic retinopathy signs except macular edema

based on the analysis of EyeScan and FF450 plus images by both readers, with reference to

the slit-lamp biomicroscopy examination, were more than 0.8 (Table 5.5). The kappa

coefficients for the ophthalmologist in detecting diabetic maculopathy using EyeScan and

FF450 plus were 0.70 and 0.74, respectively, whereas for the medical officer, they were 0.71

and 0.76, respectively.

96

Table 5.1: International Clinical Diabetic Retinopathy Severity Scales8

Grades Retinal findings


Mild NPDR Microaneurysms only

Moderate NPDR More than just microaneurysms

but less than severe NPDR


i. Extensive (>20) intraretinal hemorrhages in each of 4

quadrants



AND no signs of PDR



ii. Vitreous/Preretinal hemorrhage

NPDR: Non-proliferative diabetic retinopathy

PDR: Proliferative diabetic retinopathy

IRMA: Intraretinal microvascular abnormalities

Table 5.2: Diabetic retinopathy grading of the study patients based on slit lamp biomicroscopy examination

Diabetic retinopathy severity Eyes (n) % Normal 181 66.5 Mild non-proliferative diabetic retinopathy 51 18.8 Moderate non-proliferative diabetic retinopathy 26 9.6 Severe non-proliferative diabetic retinopathy 8 2.9 Proliferative diabetic retinopathy 6 2.2 Maculopathy 6 2.2

97

Table 5.3: The self reported diabetes micro- and macrovascular complications of the

enrolled study population

Diabetes complications Patients (%) Microvascular Neuropathy 31 (23%) Nephropathy 30 (22%) Macrovascular Ischemic heart disease 23 (17%) Peripheral vascular disease 13 (10%) Cerebrovascular disease 9 (7%)

Table 5.4: Sensitivity, specificity and Kappa correlations of overall diabetic retinopathy

grading by a consultant ophthalmologist and a medical officer from color fundus photographs of EyeScan (Ophthalmic Imaging System, CA) and FF450 (Carl Zeiss,

North America), with reference to slit lamp biomicroscopy examination by a consultant

ophthalmologist

Sensitivity Specificity Kappa correlation Ophthalmologist (95%CI) (95%CI) EyeScan

93% 98.20% 0.93 (84.9 - 97.1) (94.3-99.5)

Zeiss

95.10% 98.80% 0.95 (87-98.4) (95.4-99.8)

Medical Officer

EyeScan

91.70% 94.70% 0.88 (83.2-96.3) (89.9-97.4)

Zeiss

91.90% 95.90% 0.9 (83.4-96.4) (91.5-98.2)

98

Table 5.5: Kappa statistics for retinal photography using EyeScan and FF450 plus in

comparison with the gold standard slit-lamp biomicroscopy examination by an ophthalmologist and a medical officer

Retinal Findings Kappa statistics

Retinal Photography

Retinal Photography

EyeScan FF450 plus Ophthalmologist Microaneurysms 0.94 0.97 Retinal hemorrhages 0.87 0.95 Cotton wool spots 1 1 Venous beading 1 0.9 Intraretinal microvascular abnormalities 0.93 0.93 New vessels formation 1 1 Vitreous hemorrhage 1 1 Hard exudates 0.97 1 Macular edema 0.7 0.74 Cupped optic disc 1 1 Professional Grader Microaneurysms 0.88 0.88 Retinal hemorrhages 0.88 0.91 Cotton wool spots 1 1 Venous beading 0.89 0.89 Intraretinal microvascular abnormalities 0.76 0.76 New vessels formation 1 1 Vitreous hemorrhage 1 1 Hard exudates 0.94 0.94 Macular edema 0.71 0.76 Cupped optic disc 0.87 0.89

99

Table 5.6: Classification of diabetic retinopathy into retinopathy stages (Wisconsin

level)9

Diabetic Retinopathy Stage

Retinal Findings

Minimal NPDR

MA and one or more of the following: retinal haem, Hex, CWS, but not meeting the criteria for moderate NPDR

Moderate NPDR

H/Ma >std photo 2A in at least one quadrant and one of more of: CWS, VB, IRMA, but not meeting severe NPDR

Severe NPDR

Any of : H/Ma>std photo 2A in all four quadrants, IRMA >std photo 8A in one or more quadrants, VB in two or more quadrants

PDR

Any of: NVE or NVD <std photo 10 A, vitreous/preretinal haem NVE<1/2 DA without NVD

High-risk PDR

Any of : NVD>1/4 to 1/3 disc area, or with vitreous/ preretinal haem, or NVE>1/2 DA with vitreous/preretinal haem

Advanced PDR

High-risk PDR with tractional detachment involving macula or vitreous haem obscuring ability to grade NVD and NVE

CWS: Cotton Wool Spots DA: Disc Area Haem: hemorrhages H/Ma: Hemorrhages and microaneurysms Hex: Hard exudates IRMA: Intraretinal microvascular abnormalities MA: Microaneurysms NPDR: Non-proliferative diabetic retinopathy NVD: New vessels disc NVE: New vessels elsewhere PDR: Proliferative diabetic retinopathy Std: Standard VB: Venous beading

100

Figure 5.1: Images captured by EyeScan and FF450 plus

EyeScan: Normal Fundus FF450 plus: Normal Fundus

EyeScan: Subhyaloid hemorrhage FF450plus: Subhyaloid hemorrhage

5.5 Discussion

In our study, we utilized the International Clinical Diabetic Retinopathy Severity Scale

(Table 5.1) 8 as the grading system as it is much more simplified with less severity levels and

diagnostic criteria, as compared with the Early Treatment Diabetic Retinopathy Study

(ETDRS) classification system (Table 5.6).9 In addition, the slit lamp examination was

chosen as the reference standard given that it has been shown to be compared favorably with

the 7-field stereoscopic 30 degrees ETDRS.10 In addition, it is easy to perform, less time

consuming and more tolerable especially when all patients had to undergo two sets of retinal

still imaging on EyeScan and FF450 plus. Should the quality of the retinal still images were

compromised due to the presence of cataracts or pterygiums (encroaching the visual axis), the

101

slit lamp examination by an ophthalmologist would be more accurate than the retinal still

images captured using 7-field ETDRS.

In this study, we have shown that with reference to the slit lamp examination, both readers

had comparable sensitivity and specificity in grading diabetic retinopathy from retinal

photographs captured using the EyeScan machine or FF450 plus (Table 5.4). Both devices

had a sensitivity and specificity of 100% in detecting sight-threatening diabetic retinopathy

changes. Additionally, the kappa statistics of EyeScan and FF450 plus for the detection of all

diabetic retinopathy signs, except for macular edema, were more than 0.8 (Table 5.5). Since

the images obtained from EyeScan showed excellent sensitivity, specificity, and kappa

coefficients in diagnosing diabetic retinopathy, it could be used as reliably as currently used

cameras for the screening of diabetic retinopathy.

The color fundus images from both devices graded by the ophthalmologist and medical

officer had kappa coefficients of less than 0.8 for detecting diabetic maculopathy (Table 5.5).

Since retinal photographs provide only two-dimensional views of the retina, color fundus

images do not afford easy identification of any retinal thickening or macular edema. In this

study, we diagnosed diabetic maculopathy on the basis of the presence of hard exudates,

microaneurysms, and retinal hemorrhages close to the macular area. Presence of these lesions

in the macula region should always prompt an urgent referral.

In order to evaluate the usability of EyeScan, a medical officer with no previous experience

in ocular imaging was recruited as the operator of the device. He received a single day’s

training on the device prior to the screening. The technical failure rate of the FF450 plus

operated by an experienced retinal photographer (10 years of experience on retinal still

photography for diabetic retinopathy screening) and the EyeScan were similar (8.5% versus

7%, p > 0.05). These figures show that EyeScan may be used by non-experienced personnel

102

with minimal training. Further studies will be of great value in evaluating the user-

friendliness of the EyeScan for both medical and non-medical personnel in the primary health

care and tele-ophthalmology setting.

In our study, assessments by both readers had excellent sensitivity, specificity, and kappa

coefficients in detecting and grading diabetic retinopathy using the color fundus images

captured by both devices (Tables 5.4 and 5.5). These results indicate that non-specialist

personnel such as primary care physicians and allied health personnel can be trained to screen

for diabetic retinopathy in the primary health care setting. The EyeScan will be a suitable

device to be utilized in a tele-ophthalmology setting as it requires small storage capacity for

its lower resolution files and thus, the available bandwidth would better handle the transfer

and archives of the images form community to centralized screening center. Further research

is required to evaluate the cost effectiveness of using EyeScan in a community and tele-

ophthalmology setting, using primary eye care providers, such as optometrists, general

practitioners, orthoptists, and diabetes nurses, to screen for diabetic retinopathy. The

ophthalmologist will not be able to service the projected increase in numbers of patients with

diabetes. Optimizing use of people resources, embracing new and affordable technology will

be necessary to effectively screen these patients and therefore reduce the impact of diabetic

retinopathy related visual impairment.

The strength of our study was that all recruited patients underwent color fundus photo

imaging by using both devices (EyeScan and FF450 plus), and this allowed a head-to-head

comparison between the 2 devices. In addition, we have utilized 2 different statistical

methods (sensitivity/specificity and Cohen’s kappa statistics) and 2 readers (an

ophthalmologist and a medical officer) to increase the reliability and validity of our study

findings. In contrast, our study carries a few limitations, one of which was that the slit-lamp

103

examination (reference standard) was performed by a single senior consultant

ophthalmologist. In addition, all patients successively underwent 2 sets of retinal still

photography within a short period of time, and this could have contributed to the technical

failure that had occurred as a consequence of patients’ intolerance to bright light from both

the devices.

104

6. CHAPTER 4: DIABETIC RETINOPATHY SCREENING: CAN THE

VIEWING MONITOR INFLUENCE THE READING AND

GRADING OUTCOMES

6.1 Summary

To increase primary eye care providers’ interest and desire to screen for DR, we conducted a

study to explore the possibility of using smaller, portable reading devices for the image

reading and grading of DR. This single-center experimental case series evaluated reading

devices for DR screening. A total of 100 sets of three-field (optic disc, macula and temporal

views) color retinal still images (50 normal and 50 with DR) captured by FF450 plus (Carl

Zeiss, Meditec, US) were interpreted on a 27-inch iMac, a 15-inch MacBook Pro and a 9.7-

inch iPad. All images were interpreted by a retinal specialist and a medical officer. We

calculated the sensitivity and specificity of a 15-inch MacBook Pro and a 9.7-inch iPad in the

detection of DR signs and grades, with reference to the reading outcomes obtained using a

27-inch iMac reading monitor. In the detection of any grade of DR, the 15-inch MacBook Pro

had sensitivity and specificity of 96% (95% confidence interval (CI): 85.1–99.3) and 96%

(95% CI: 85.1–99.3) respectively for the retinal specialist and 91.5% (95% CI: 78.7–97.2)

and 94.3% (95% CI: 83.3–98.5) respectively for the medical officer, while for the 9.7-inch

iPad, they were 91.8% (95% CI: 79.5–97.4) and 94.1% (82.8–98.5) respectively for the

retinal specialist and 91.3% (95% CI: 78.3–97.1) and 92.6% (95% CI: 81.3–97.6)

respectively for the medical officer. In conclusion, the 15-inch MacBook Pro and the 9.7-inch

iPad had excellent sensitivity and specificity in detecting DR; hence, both screen sizes can be

utilized to effectively interpret color retinal still images for DR remotely in a routine, mobile

or tele-ophthalmology setting. Future studies could explore the use of more economical

105

devices with smaller viewing resolutions to reduce the cost implementation of DR screening

services.

6.2 Introduction

Diabetic retinopathy screening programs have been implemented worldwide to enable early

detection of diabetic retinopathy, which if treated appropriately, will minimize severe visual

impairment.73 By 2030, it is estimated that the total number of people with diabetes will rise

to 366 million.1 Due to the rising prevalence of diabetes, the current screening services in

developing and developed countries will be faced with increasing costs of implementation

and maintenance of a screening program for the people with diabetes.114 It is therefore

prudent that stakeholders continue to look for different ways of servicing the increasing

diabetic population, and at the same time minimizing the economical impact of screening

programs within the community.

To date, various studies have evaluated various parameters which may affect the sensitivity,

specificity and cost effectiveness in screening diabetic retinopathy, including numbers of

retinal fields,10 color or monochromatic fundus photographs,83 mydriatic status,6,117

photographers and readers with different medical qualifications,117 automated grading

system,109 use of an economical retinal camera110 and retinal video recording technique.120

However, none of the studies have compared the use of different viewing monitors for the

reading and grading of diabetic retinopathy from the digital color fundus photographs.

To date, the use of small viewing monitors in screening for diabetic retinopathy has become

an emerging trend among the ophthalmologists or professional graders as they can utilize

them remotely without being physically present at the reading center. Due to the growing

popularity of these mobile and portable technologies, the purpose of our study is to evaluate

the efficacy of using different portable and mobile devices with varying viewing monitors

106

(15-inch MacBook Pro and 9.7-inch iPad) to detect subtle diabetic retinopathy changes

(microaneurysms and dots hemorrhages) and diagnose the severity level. This also helps to

determine the suitability of using smaller and more affordable portable devices to interpret

the color retinal images for grading of diabetic retinopathy.

6.3 Methods

Design and Data Acquisition

This is a single center case series to evaluate different screening resolutions to interpret

retinal color still images for diabetic retinopathy screening. A total of 100 sets of non-stereo

mydriatic 3-field (optic disc, macula and temporal views) 35 degrees color retinal still images

consisting of 50 normal and 50 with diabetic retinopathy were selected into our study. The

quality of all recruited retinal images were at least ‘acceptable’ (more than two-third of

retinal images were ‘interpretable’) based on the reading outcomes using a 27-inch iMac (the

standard viewing screen in our reading center). All images were captured using FF 450 plus

(Carl Zeiss, Inc., North America) by an experienced retinal photographer at the Diabetic

Retinopathy Screening Clinic of Royal Perth Hospital and downloaded in Joint Photographic

Experts Group (JPEG) format. The color resolution of all images was fixed at 2588 x 1958

pixels. This study has been approved by the Royal Perth Hospital Human Research Ethics

committee.

Data Interpretation

All images were de-identified, randomized and interpreted by two readers (a retinal specialist

and a medical officer) in a dark room using the standardized Apple software – iPhoto (Apple,

CA, USA) on three monitors with different sizes – 27-inch iMac (Apple, CA, USA), 15-inch

MacBook Pro (Apple, CA, USA) and 9.7-inch iPad (Apple, CA, USA). All images were

107

interpreted on the specific three monitors with calibrated brightness at 100%; target white

point at D65; and target gamma at 2.2 using software named Display Calibrator Assistant.

The specification of the reading devices were listed in Table 6.1. The quality of retinal

images was rated as ‘acceptable’ or ‘uninterpretable’ by the readers. The diabetic retinopathy

severity level was graded based on presence/absence of microaneurysms, retinal hemorrhages,

cotton wool spots, venous beading, intraretinal microvascular abnormalities, new vessels

formation and hard exudates using the International Clinical Diabetic Retinopathy Severity

Scales8 (Table 6.2).

Sample size estimation

To allow for a power of 95%, desired precision of 0.10, expected sensitivity and specificity

of 90%, the total number of eyes required for each device was 71 (prevalence was set at 0.50

as the selected samples consisted of 50% normal and 50% abnormal retinal color still images).

Statistical Analyses

All data were analyzed using SPSS version 17 (SPSS, Chicago, IL, USA). The sensitivity,

specificity and Kappa correlation coefficient of 15-inch MacBook Pro and 9.7-inch iPad in

detecting diabetic retinopathy lesions and grading were calculated with reference to the

findings on the 27-inch iMac (as the reference standard). Moreover, the Kappa coefficient

was performed on the diabetic retinopathy grading detected on 27-inch iMac for both readers

and diabetic retinopathy lesions detected by 15-inch MacBook Pro and 9.7-inch iPad with

reference to 27-inch iMac.

108

6.4 Results

The mean age [± standard deviation (SD)] of the recruited participants was 51.3 ± 13.8 with

HbA1c of 8.4 ± 1.6 and duration of diabetes of 12.1 ± 8.7. Of the retinal images, 50 (50%)

had no diabetic retinopathy, 16 (16%) had mild non-proliferative diabetic retinopathy

(NPDR), 25 (25%) had moderate NPDR, 7 (7%) had severe NPDR and 2 (2%) had

proliferative diabetic retinopathy (PDR). All retinal images were rated as ‘acceptable’ by the

retinal specialist and medical officer on 15-inch MacBook Pro and 9.7-inch iPad.

For 27-inch iMac (the ‘reference standard’ of our study), both retinal specialist and medical

officer had a Kappa correlation of 0.88 in detecting the overall diabetic retinopathy grading.

In detection of any grade of diabetic retinopathy on 15-inch MacBook Pro, the retinal

specialist had sensitivity and specificity of 96% and 96% respectively while the medical

officer had 91.5% and 94.3% respectively with reference to the 27-inch iMac (Table 6.3). On

the other hand, the sensitivity and specificity in detecting any grade of diabetic retinopathy on

9.7-inch iPad for retinal specialist were 91.8% and 94.1% respectively whereas for medical

officer, they were 91.3% and 92.6% respectively. For sight threatening diabetic retinopathy,

the retinal specialist had 100% sensitivity and specificity on both reading devices whereas for

medical officer, the sensitivity and specificity were 100% and 97.7% respectively on both

devices.

The iPad had lower sensitivity and specificity (retinal specialist: 89.1% and 96.3%

respectively; medical officer: 87.5% and 98.1% respectively) in detecting microaneurysms by

both readers compared to MacBook Pro (sensitivity- retinal specialist: 100% and 96.3%

respectively; medical officer: 95.8% and 100% respectively) (Table 4.4). Both devices had

comparable sensitivity and specificity in detecting retinal hemorrhages by both readers (Table

6.4).

109

For retinal specialist, the Kappa coefficient for 15-inch MacBook Pro and 9.7-inch iPad in

detection of any grade of diabetic retinopathy were 0.94 and 0.89 respectively whereas for

medical officer, they were 0.89 and 0.88 respectively with reference to 27-inch iMac. The

Kappa coefficient in detecting all diabetic retinopathy signs (microaneurysms, retinal

hemorrhages, cotton wool spots, new vessels formation and hard exudates) by both readers

were more than 0.80 (Table 6.5).

Table 6.1: Specifications and prices of 27-inch iMac, 15-inch MacBook Pro and 9.7-inch iPad (The indicated prices are obtained in United States Dollars)

27-inch iMac

15-inch MacBook Pro

9.7-inch iPad

(Wi-Fi + 3G)

Monitor Screen (inch) 27 15 9.7

Screen resolution (pixels) 1920 x 1080 1440x900 1024x768

Graphics Processor

ATI Radeon HD

4670

NVIDIA GeForce

320M A4 1Ghz

Weight (kg) 13.8 2.54 0.73

Price (USD) $1,699 $1,799 $629

110

Table 6.2: International Clinical Diabetic Retinopathy Severity Scales9








quadrants



AND no signs of proliferative DR

Proliferative DR One or more of the following:




DR: Diabetic retinopathy

111

Table 6.3: The sensitivity, specificity and Kappa coefficient of 15-inch MacBook Pro

and 9.7-inch iPad in detecting diabetic retinopathy grading by a retinal specialist and a medical officer with reference to 27-inch iMac

Any Grade of Diabetic Retinopathy Sensitivity Specificity Kappa

Retinal Specialist (95% CI) (95% CI)

15-inch MacBook Pro 96% (85.1-99.3) 96% (85.1-99.3) 0.94

9.7-inch iPad 91.8% (79.5-97.4) 94.1% (82.8-98.5) 0.89

Medical Officer

15-inch MacBook Pro 91.5% (78.7-97.2) 94.3% (83.3-98.5) 0.89

9.7-inch iPad 91.3% (78.3-97.1) 92.6% (81.3-97.6) 0.88

Sight Threatening Diabetic Retinopathy

Retinal specialist

15-inch MacBook Pro 100% 100% 1.00

9.7-inch iPad 100% 100% 1.00

Medical Officer

15-inch MacBook Pro 100% 97.7% 0.92

9.7-inch iPad 100% 97.7% 0.92

NB: Sight-threatening diabetic retinopathy comprises of severe non-proliferative diabetic

retinopathy and proliferative diabetic retinopathy.

112

Table 6.4: The sensitivity, specificity and Kappa coefficient of 15-inch MacBook Pro

and 9.7-inch iPad in detecting microaneurysms and retinal hemorrhages by a retinal specialist and a medical officer with reference to 27-inch iMac

Microaneurysms Sensitivity Specificity

Retinal Specialist (95% CI) (95% CI)

15" Macbook Pro 100% (90.4-100) 96.3% (86.1-99.4)

9.7" iPad 89.1% (75.6-95.9) 96.3% (86.2-99.4)

Medical Officer

15" Macbook Pro 95.8% (84.6-99.3) 100% (91.4-100)

9.7" iPad 87.5% (74.1-94.8) 98.1% (88.4-99.9)

Retinal Hemorrhages

Retinal specialist

15" Macbook Pro 93.8% (77.8-98.9) 98.5% (91-99.9)

9.7" iPad 93.8% (77.8-98.9) 95.6% (86.8-98.9)

Medical Officer

15" Macbook Pro 93.9% (78.4-98.9) 98.5% (90.9-99.9)

9.7" iPad 93.9% (78.4-98.9) 95.5% (86.6-98.8)

113

Table 6.5: The Kappa correlation of diabetic retinopathy changes interpreted by a

retinal specialist and a medical officer on 15-inch MacBook Pro and 9.7-inch iPad with reference to the retinal findings detected on 27-inch iMac

Retinal Findings Kappa statistics

15-inch MacBook Pro 9.7-inch iPad

Retinal specialist

Microaneurysms 0.96 0.86

Retinal hemorrhages 0.93 0.89

Cotton wool spots 1.00 1.00

New vessels formation 1.00 1.00

Hard exudates 0.96 0.96

Cupped optic disc 1.00 1.00

Medical officer

Microaneurysms 0.92 0.86

Retinal hemorrhages 0.93 0.91

Cotton wool spots 1.00 1.00

New vessels formation 1.00 1.00

Hard exudates 0.96 0.96

Cupped optic disc 1.00 1.00

6.5 Discussion

The success of a screening process relies on multiple factors including the photographers’

factor, patients’ factor and readers’ factor. In the presence of an experienced photographer,

patients with good ocular media and experienced readers, the influence of viewing monitors

114

also plays a role in determining the diagnostic accuracy of retinal images grading. In order to

evaluate the effectiveness of 15-inch Macbook Pro and 9.7-inch iPad in detecting diabetic

retinopathy lesions and grading, we compared the retinal findings of each of the devices with

the respective findings on 27-inch iMac. In our study, the retinal specialist and the medical

officer as the trained reader had extremely strong inter-observer agreement (Kappa = 0.88) in

grading diabetic retinopathy on the 27-inch iMac. Both readers had excellent sensitivity and

specificity in diagnosing diabetic retinopathy using 15-inch MacBook Pro (retinal specialist-

96%, 96%; medical officer: 91.5%, 94.3%) and 9.7-inch iPad (retinal specialist- 91.8%,

94.1%; medical officer: 91.3%, 92.6%) (Table 6.3). In addition, the Kappa correlation

between 15-inch MacBook Pro and 9.7-inch iPad versus 27-inch iMac in detection of

diabetic retinopathy changes (microaneurysms, retinal hemorrhages, cotton wool spots,

neovascularization and hard exudates) and diabetic retinopathy grading were excellent

(greater than 0.8). These results indicated that the specialist (retinal specialist) and non-

specialist (medical officer) screeners could effectively interpret and diagnose diabetic

retinopathy from the color retinal still images using a 15-inch or a 9.7-inch reading screen.

In detection of sight-threatening diabetic retinopathy (severe non-proliferative diabetic

retinopathy), the medical officer had 100% sensitivity and 97.7% specificity on both devices

(MacBook Pro and iPad). The discrepancy of the specificity between the retinal specialist and

the medical officer were due to two false positives which had been graded by the medical

officer as severe NPDR instead of moderate NPDR. Nevertheless, a screener especially the

non-ophthalmologist personnel such as the optometrists and general practitioners should

always be suspicious and have lower threshold in referring patients with uncertain diabetic

retinopathy lesions detected on the retinal still images, even if this will result in some

‘unnecessary’ referrals to the specialists.

115

The native image resolution of 2588 x 1958 pixels exceeded any of the compared displays

spatial capabilities. This image resolution of 2588 x 1958 pixels was set by the fundus

camera (Zeiss FF 450 plus) and all images were interpreted using a common software –

iPhoto (Apple, CA, USA). The image size exceeded the compared displays spatial resolution

and therefore, all the images were set to 100% to fit the full screen during the viewing and

interpretation process. However, the full image may still be navigated on different screens

using the iPhoto display program with ease.

In our study, we utilized the 27-inch iMac as the reference standard of our study to avoid any

diagnostic error secondary to small screen size and low screen resolution. We adopted the

mydriatic 50 degrees 3-field retinal still photography in our screening clinic given that its

sensitivity in detecting any grade of diabetic retinopathy has been shown to be more than

90%.82 Compared with the gold standard 7-field 30 degrees stereoscopic views, it is more

time saving and practical to be implemented in the routine screening setting. For the

displaying program, the ‘iPhoto’ was utilized instead of the more specialized programs such

as ‘Visupac’ or ‘IMAGEnet system’ as the latter programs often will need to be purchased.

On the other hand, it is more economical to use ‘iPhoto’ program as it is a free software

which is included in the Mac computers.

In our study, the iPad had slightly lower sensitivity, specificity and Kappa correlation with

the 27-inch iMac compared to the 15-inch MacBook Pro (Table 6.4). Both devices, however,

had excellent diagnostic accuracy in detecting sight-threatening diabetic retinopathy lesions

(retinal hemorrhages, cotton wool spots and new vessels) and diabetic retinopathy grading by

both readers (Table 6.3). In a screening setting, it is rather critical to detect and refer patients

with sight threatening diabetic retinopathy changes such as multiple retinal hemorrhages,

cotton wool spots and neovascularization such that pan-retinal photocoagulation could be

116

applied without delay to prevent severe visual impairment. Therefore, an occasional missed

microaneurysm often will not result in severe visual impairment and this is consistent with

the findings of our study given that both readers had 100% sensitivity in diagnosing sight-

threatening diabetic retinopathy on both devices. Given the Kappa correlation between iPad

and 27-inch iMac, graded by retinal specialist and medical officer, in detection of

microaneurysms was within the excellent range (k=0.86), it will be feasible for the specialist

and non-specialist readers to utilize a small reading screen (e.g. 9.7-inch iPad) with a spatial

resolution of 1024 x 768 pixels to effectively screen for diabetic retinopathy in the

community. A further study will be of great value to explore the efficacy of other cheaper PC,

tablet computers [e.g., Galaxy Tab (Samsung, Korea)] and cell/smart phones [e.g., 3.5-inch

iPhone (Apple, CA, USA) and 4-inch Galaxy S (Samsung, Korea)] with smaller reading

screen sizes to screen for diabetic retinopathy in a routine, mobile or tele-ophthalmology

setting.

The strength of our study was being one of the recent studies which evaluated the effect of

using devices with varying monitor resolution to screen for diabetic retinopathy. Moreover,

we utilized two statistical methods (sensitivity/specificity and Kappa coefficient) and two

readers (retinal specialist and medical officer) in order to justify the diagnostic accuracy of

each monitor size by the specialist and non-specialist personnel. On the other hand, one of the

weaknesses of our study was that the color resolution of the 3 display monitors was different.

Despite having a similar color depth for all screens, the color gamut, the range and set of

colors that they can produce, were not the same for all 3 monitors. The iMac can display

much wider color gamut than other two displays used in this study. The Macbook Pro and

iPad displays have much less display color gamut than iMac. The color gamut can influence

the accuracy of the colors and may show under saturated colors and hence, potentially

affecting the interpretation of fundus images.

117

In addition, our results may be potentially subject to a selection bias as we only selected the

good quality color retinal images into our study. It is unknown if the color retinal images with

suboptimal quality due to media opacity or dark fundi will affect the sensitivity and

specificity in detecting diabetic retinopathy changes using the 15-inch or 9.7-inch reading

screens. Thus, it will be of great significance to recruit all patients with diabetes

consecutively from the screening clinic in the future study to evaluate the overall efficacy of

different monitor resolutions in detecting diabetic retinopathy lesions from the color retinal

images with good and suboptimal quality.

118

SECTION 4: NOVEL VIDEO-BASED IMAGING TECHNOLOGY FOR DIABETIC

RETINOPATHY SCREENING

7. CHAPTER 5: RETINAL VIDEO RECORDING: A NEW WAY TO

IMAGE AND DIAGNOSE DIABETIC RETINOPATHY

7.1 Summary

In addition to the portable retinal camera and reading devices, we also explored a novel

technique using retinal video recording to screen for DR. We conducted a study to compare

this technique with standard traditional retinal photography, with reference to slit lamp

examination performed by two senior ophthalmologists. This single-center study evaluated a

new diagnostic technique: retinal video recording for 100 patients with DR recruited from

Royal Perth Hospital, WA. All fundus images were captured using standard retinal still

photography (FF450 plus; Carl Zeiss, Meditec, US) and retinal video (EyeScan; OIS, CA,

US), followed by a gold-standard slit-lamp biomicroscopy examination. All videos and still

images were de-identified, randomized and interpreted by two senior consultant

ophthalmologists. Kappa statistics, sensitivity and specificity for all DR signs and grades

were calculated with reference to slit lamp examination results as the gold standard. The main

outcome measures were the sensitivity and specificity of video recordings for detecting DR

signs and grades. The mean age (± standard deviation (SD)) of participants was 52.8±15.1

years, the mean duration of diabetes (±SD) was 13.7±9.7 years and the mean glycosylated

hemoglobin level was 8.0±1.7%. Compared with the gold-standard slit lamp examination

results, the sensitivity and specificity of video recording for detecting the presence of any DR

was 93.8% and 99.2% (ophthalmologist 1) and 93.3% and 95.2% (ophthalmologist 2)

119

respectively. In contrast, the sensitivity and specificity of retinal photography was 91.8% and

98.4% (ophthalmologist 1) and 92.1% and 96.8% (ophthalmologist 2) respectively. Both

imaging methods had 100% sensitivity and specificity in detecting STDR. For overall DR

grading by both ophthalmologists, the measurements of agreement (Cohen’s coefficient)

between the overall grading obtained from the retinal video versus slit lamp examination, and

retinal photography versus slit lamp examination, were more than 0.90. The technical failure

rate for retinal video recording and retinal photography were 7.0% and 5.5% respectively. In

conclusion, this study demonstrated that retinal video recording was equally as effective as

retinal photography in the subjects who were evaluated. It is a novel, alternative DR

screening technique that not only gives primary eye care providers the opportunity to view

numerous retinal fields within a short period, but it is also easy to learn by inexperienced

personnel with minimal training.

7.2 Introduction

Diabetes is one of the world’s fastest growing chronic diseases and a leading cause of

acquired vision loss.131 According to the World Health Organization (WHO), it is estimated

that the total number of people with diabetes for all age-groups will more than double from

171 million (2.8%) in 2000 to 366 million (4.8%) by 2030.1 On average, 25% of diabetic

patients have identifiable signs of diabetic retinopathy2 and the prevalence of diabetic

retinopathy can increase in up to 75% in patients who have had diabetes for 20 years or

more.123,124 Nevertheless, early detection and prompt treatment has been reported to be able to

prevent up to 98% of diabetes-related visual impairment.73

In the United States, federal savings of $624 million and 400,000 person-years of sight could

be achieved annually if everyone with diabetes underwent regular diabetic retinopathy

screening and received treatment according to the severity of their condition.125-127 A national

120

tele-retinal imaging diabetic retinopathy screening program was set up between the Veterans

Health Administration (VHA), Joslin Vision Network and the Department of Defense and the

Veterans Integrated Service Network in order to improve the access to diabetic retinopathy

screening for the United States population.128-130 Similarly, the National Health Service has

also set up a National Diabetic Retinopathy Screening Program in United Kingdom with the

aim to achieve a 100% screening rate for the patients with diabetes.131 In Australia, the use of

tele-ophthalmology in diabetic retinopathy screening has also been shown to be cost effective

and cheaper than any of the alternative options with a minimum number of patients per year

to be 128.132,133

To date, all tele-retinal screening services worldwide utilize retinal photography (mydriatic or

non-mydriatic) as the diabetic retinopathy screening tool. For detection of diabetic

retinopathy, the Early Treatment Diabetic Retinopathy Study (ETDRS) seven standard field

35-mm stereoscopic color fundus photographs using the modified Airlie House classification

is currently the gold standard classification.9 However, the use of a single field mydriatic 45º

retinal photography with sensitivity of more than 80% has been shown to be adequate and

effective (level 1 evidence)6,134 in a screening setting and a mydriatic 3-fields retinal

photography could further increase the sensitivity to 97%.135

A good quality retinal image is highly dependent on the operators’ skills to perform retinal

photography, patient compliance with instructions and their tolerance to the bright light. As a

result, retinal photography is often performed by highly trained professionals, such as

optometrists and orthoptists, which makes it less accessible in a remote community.

Moreover, even when performed by experienced retinal photographers, the reported technical

failure rate of mydriatic retinal photography has been shown to be as high as 12%.82,136

121

Our study describes a new technique using retinal video recording to screen for diabetic

retinopathy. Unlike traditional retinal photography, retinal video recording is easier and

quicker to learn. This technique also provides a view of the retina which simulates what is

seen with a slit lamp examination. As it also provides a greater field of view compared to the

3-field retinal photography, this technique of image acquisition offers an easier alternative

and may even supplant what is currently in use.

7.3 Methods

Design

This is a single center, blinded study to determine the validity of a new diagnostic technique -

retinal video recording, for assessing presence and stage of diabetic retinopathy, by

comparison with currently accepted diagnostic practice using retinal still photography.

Study sample

A total of 100 patients (200 eyes) were recruited for this study from the Ophthalmology

Clinic of Royal Perth Hospital, a public teaching hospital in Western Australia into this study.

No patients were excluded based on age, gender, socioeconomic status or clinical

characteristics. All participants provided a signed informed consent for participation and the

study was approved by the Royal Perth Hospital Ethics Committee.

122

Patients’ demographics and diabetes history

Data collected included patient demographics (age and ethnicity); best corrected visual acuity;

ocular history (e.g., cataracts, glaucoma, laser therapy); self reported diabetic and

cardiovascular history (type and duration of diabetes, glycosylated hemoglobin (HbA1c)

levels, smoking status, blood pressure, lipid profile) as well as diabetic complications

(cerebrovascular accidents, ischemic heart disease, peripheral vascular disease, retinopathy,

nephropathy and neuropathy).

Retinal video and retinal photography protocol

All patients underwent three separate tests for diagnosis of diabetic retinopathy: i) retinal

video recording, ii) still retinal photography and iii) slit lamp biomicroscopy examination.

Upon presentation at the clinic, they received 2.5% phenylephrine and 0.5% tropicamide in

both eyes to maximize pupil dilation. Three fields (optic disc, macula and temporal views)

retinal video recording and still retinal photography were captured using EyeScan

(Ophthalmic Imaging Systems, Sacramento, California) and FF450 plus (Carl Zeiss, Inc.

North America) respectively. The retinal video recording was performed by a senior

Ophthalmology resident medical officer who received a full day of training for using the

device. Retinal photography was performed by a trained photographer using FF450 plus. All

retinal video recordings commenced at the optic disc and proceeded to the macula and

temporal regions. In order to obtain continuity of retinal information between the regions, the

retinal camera was tilted at a consistent pace from left to right for the right eye (optic disc,

macula and temporal retina) for at least five seconds and vice versa for the left.

123

Interpretation of retinal photography and retinal videos

All retinal still images and videos were de-identified, randomized and then interpreted by a

senior consultant ophthalmologist. The quality of the videos and retinal photography images

were classified as ‘unacceptable’, ‘average’ or ‘excellent’. The retinal video recordings were

graded as unacceptable by the ophthalmologist if they were blurred, out of focus, dark and/or

had insufficient views (less than five seconds on any view). The retinal photos were graded

as ‘unacceptable’ if more than one-third of the photo was blurred.

Diabetic retinopathy was diagnosed based on the presence of the following: microaneurysms,

dot/blot hemorrhages, cotton wool spots, intraretinal microvascular abnormalities (IRMAs),

venous beading, neovascularization, preretinal/vitreous hemorrhage, hard exudates, macular

edema.

Diabetic retinopathy severity was graded using the International Clinical Diabetic

Retinopathy Severity Scale (Table 7.1).8 This scale was established by the American

Academy of Ophthalmology (AAO) in the Global Diabetic Retinopathy Project with the aim

to improve the communication between the medical and non-medical personnel such as

ophthalmologists, diabetic specialists, primary care physicians and retinal photographers by

promoting the development of a common clinical severity scale for diabetic retinopathy.8 All

the retinal photographs and retinal videos were viewed using the same monitor (iMac 27

inches; Apple, California, USA) and VLC media player 1.1.4 (Mac, Apple, California, USA).


The main outcome measures investigated were the sensitivity and specificity of the two

modes of imaging (retinal video recording and retinal photography) in detecting diabetic

retinopathy grading with reference to the currently accepted ‘gold-standard’ slit lamp

124

biomicroscopy examination. Technical failure rate, defined as the fraction of the

unacceptable images or videos, was determined for both photographic methods.

As a measurement of agreement, Cohen’s Kappa coefficient137 was obtained for diabetic

retinopathy signs and grading between retinal video recording and retinal photography versus

the gold standard slit lamp biomicroscopy examination were calculated. All data were

analyzed using SPSS version 17 (SPSS, Chicago, IL, USA).

5.4 Results

A total of 200 eyes (100 patients) were enrolled in our study. A summary of the patients’

demographic characteristics and diabetes history is shown in Table 7.2. The mean (±SD) age

of participants was 52.9±15.1 years, duration of diabetes was 13.7 ± 9.7 years and HbA1c

was 8.0 ± 1.7%. Most of the patients were Caucasian (75.0%, n=75) and had type II diabetes

(70.0%, n=70).

Patients’ clinical characteristics

Most eyes (88%, n=176) had best corrected visual acuity of 6/6 or 6/9, 22 eyes (11%) had

6/12 to 6/36 and 2 eyes (1%) had 6/60. Cataracts were diagnosed in eighteen eyes (9%) based

on slit lamp biomicroscopy examination. Of the 200 eyes, 70 eyes (35%) were diagnosed

with diabetic retinopathy. Of those diagnosed, 36 (18%) had mild non-proliferative diabetic

retinopathy, 23 (11.5%) had moderate non-proliferative diabetic retinopathy, 2 (1%) had

severe non-proliferative diabetic retinopathy, 5 (2.5%) had proliferative diabetic retinopathy

and 4 (2%) had diabetic maculopathy.

Of the diabetes-related complications, more patients suffered from microvascular

complications than macrovascular complications. Diabetic neuropathy (23%, n=23) was the

commonest complication, followed by diabetic nephropathy (22%, n=22); ischemic heart

125

disease (16%, n=16); peripheral vascular disease (10%, n=10) and cerebrovascular disease

(6%, n=6). Of the 100 recruited patients, 40% had never undergone any previous diabetic

retinopathy screening while 10% reported having previously received retinal laser therapy.

Cross-source agreement of diagnostic methods

For detection of any grade of diabetic retinopathy, the sensitivity and specificity of the retinal

video imaging technique, when compared with gold-standard slit lamp biomicroscopy

examination, were 93.9% (95%CI: 84.4-98.0) and 98.5% (95%CI: 94.1-99.7) respectively.

Retinal still photography had a sensitivity, compared with the gold-standard, of 92.4%

(95%CI: 82.5-97.2) and specificity of 98.5% (95%CI: 94.2-99.7). For detection of sight-

threatening diabetic retinopathy (severe non-proliferative diabetic retinopathy, proliferative

diabetic retinopathy), the sensitivity and specificity of retinal video recordings and retinal

photography compared with the gold-standard measurement were both 100%. Moreover,

both imaging devices had 100% sensitivity in detecting diabetic maculopathy but the

specificity was 98.9% and 97.9% for retinal videos and retinal photography respectively.

The Kappa coefficient for retinal video recordings and retinal photography for all diabetic

retinopathy grades were 0.98 and 0.92 respectively (Table 3). In detection of diabetic

maculopathy, the Kappa coefficient was 0.79 and 0.66 for grading based on retinal video

recordings and retinal photography respectively. Additionally, Cohen’s Kappa coefficient for

retinal video recordings and retinal photography for all diabetic retinopathy signs

(microaneurysms, retinal hemorrhages, venous beading, IRMAs, new vessels formation,

preretinal hemorrhage, hard exudates) and optic disc abnormalities were all greater than 0.80

(Table 7.3). Of the captured 200 eyes, 19 (9.5%) of the retinal videos and 14 (7%) of the

retinal photographic images were graded as unacceptable (technical failure).

126

Of the 19 failed retinal videos, 7 eyes had grade III nuclear sclerotic cataract based on the

Lens Opacities Classification III107, 2 eyes were from a darkly pigmented patient and 10 eyes

were due to intolerance to bright light. In contrast, the failed retinal photographs were due to

cataracts (5 eyes) and blurriness of the images secondary to eye movement (4 eyes) and

intolerance to bright flash (5 eyes).

Table 7.1: International Clinical Diabetic Retinopathy Severity Scales and International

Clinical Diabetic Macular Edema Disease Severity Scales8








quadrants



AND no signs of PDR



DME apparently absent

DME apparently present


No apparent retinal thickening or hard exudates in posterior

pole

Some apparent retinal thickening or hard exudates in posterior

pole


PDR: Proliferative diabetic retinopathy


DME: Diabetic macular edema

127

Table 7.2: Patient characteristics and their diabetes history

Patient demographics

Mean (± SD) age (years) 52.8 ± 15.1

Ethnicity (n,%)

i. Caucasian

ii. Asian

75 (75%)

15 (15%)

iii. Indigenous 6 (6%)

iv. Others 4 (4%)

Diabetes History

Mean (±SD) duration of diabetes (years) 13.7 ± 9.7

Mean (±SD) HbA1c (%) 8.0 ± 1.7

Type 1 (Insulin Dependent) 60 (30%)

Type 2

i. Non Insulin Dependent 94 (47%)

ii. Insulin Dependent 46 (23%)

SD: Standard deviation, HbA1c: Glycosylated Hemoglobin

Table 7.3: The sensitivity, specificity and Kappa correlation coefficient for retinal

photography (FF450 plus, Carl Zeiss Inc., North America) and retinal video recording (EyeScan, Ophthalmic Imaging System, CA, US) with reference to slit lamp

biomicroscopy examination

Sensitivity Specificity Kappa

(95% CI) (95% CI)

Ophthalmologist 1 Retinal still photography

91.8%

(81.1 - 96.9) 98.4%

(93.8-99.7) 0.93

Retinal video recording

93.8% (84.2 - 98.0)

99.2% (94.7-99.9)

0.95

Ophthalmologist 2 Retinal still photography

92.1%

(81.7-97.0) 96.8%

(91.6-99.0) 0.91

Retinal video recording

93.3% (83.0-97.8)

95.2% (89.4-98.0)

0.90

128

Table 7.4: The Kappa statistics for retinal videos (EyeScan) and retinal photography

(FF450 plus) by both consultant ophthalmologists compared to the gold standard slit lamp biomicroscopy examination

Retinal findings Kappa statistics Retinal Video Recording Retinal still photography Ophthalmologist 1

Microaneurysms 0.93 0.95 Retinal hemorrhages 0.93 0.94 Cotton wool spots 1.00 1.00 Venous beading 0.91 0.89 IRMA 1.00 0.85 Preretinal hemorrhages 1.00 1.00 New vessels formation 1.00 1.00 Hard exudates 1.00 1.00 Cupped optic disc 1.00 1.00 Macular edema 0.80 0.66

Ophthalmologist 2

Microaneurysms 0.9 0.93 Retinal hemorrhages 0.9 0.92 Cotton wool spots 1.00 1.00 Venous beading 0.86 0.86 IRMA 0.8 0.83 Preretinal hemorrhages 1.00 1.00 New vessels formation 1.00 1.00 Hard exudates 1.00 1.00 Cupped optic disc 1.00 1.00 Macular edema 0.66 0.66


5.5 Discussion

Our study has demonstrated the possibility of using retinal video recording as an alternative

diabetic retinopathy screening technique given that the sensitivity and specificity of retinal

video recording were comparable to color retinal photography with reference to slit lamp

biomicroscopy examination in determining the diabetic retinopathy grading (retinal video

recordings- sensitivity: 93.9%, specificity: 98.5%; retinal photography- sensitivity: 92.4%,

129

specificity: 98.5%). Moreover, retinal video also possessed a comparable Kappa coefficient

to retinal photography in detecting diabetic retinopathy signs such as microaneurysms (0.94

vs. 0.94), retinal hemorrhages (0.97 vs. 0.94), cotton wool spots (1.0 vs. 1.0), venous beading

(0.90 vs. 0.89) and intraretinal microvascular abnormalities (1.0 vs. 0.91) when compared

with the slit lamp examination.

Both imaging devices demonstrated excellent sensitivity and specificity in detecting diabetic

maculopathy. Nevertheless, this may be due to the limited sample size of patients with

macular edema in our study (n=4). Given that the retinal videos and retinal photography only

offer two-dimensional retinal views, it will be impossible to detect any signs of retinal

thickening from a retinal photo or a retinal video. Nevertheless, presence of hard exudates or

microaneurysms close to the macular area and an unexplained drop in a patient’s visual

acuity should prompt a referral to see an ophthalmologist urgently given that diabetic

maculopathy could cause significant visual impairment in patients with diabetes.

In our study, the retinal video recording technique was performed by a medical officer (no

previous experience in retinal photography) while the retinal photography was performed by

an experienced retinal photographer (had more than 10 years of retinal still photography

experience in screening for diabetic retinopathy). The difference of technical failure rate

between the video recording and the retinal photography was not statistically significant (9.5%

versus 7%, χ²=0.83, d.f.=1, p=0.36). This finding demonstrates the ‘user friendliness’ of

the device as the medical officer only received single day training on the video recording

technique. As a result, any medical or non-medical personnel such as nurses, allied health

workers or any volunteer could be trained to screen for diabetic retinopathy within a short

period of time (one to two days training sessions) and this could largely increase the diabetic

retinopathy screening rates in the community, particularly in the developing countries.

130

Of the 19 failed retinal videos, 10 eyes were due to intolerance to bright light. In our study,

the retinal video recordings were performed after retinal photography which could potentially

result in the irritation or exhaustion of patients’ eyes during the retinal video recording

process. Given that the retinal video recording technique has only been utilized for a short

period of time, the failed videos may be also partially due to the operator’s technique. Hence,

the technical failure rate could be potentially further diminished with improvement of

operator’s technique over time. Of the failed retinal photographs, they were secondary to eye

movement (4 eyes) and intolerance to bright flash (5 eyes). Retinal photography always

requires patients’ compliance to fixate their eye during the procedure to avoid image

blurriness whereas retinal video recordings could tolerate slightly more eye movement and

this is particularly useful for patients who cannot fixate their eyes to the external/internal

fixator of the retinal camera.

Compared with retinal photography, retinal video recording carries some disadvantages. It

does not possess a flash function to briefly increase the light intensity. As a result, the light

intensity required throughout the entire retinal video recordings will be slightly higher, but

less than the intensity of the flash, for the period of recording, especially in patients with dark

fundi. Additionally, the videos saved as AVI format need relatively high storage capacity as

one second in the video takes up approximately 20 megabytes. Depending on patients’

compliance and tolerance, an average retinal video recording will at least take 15 seconds and

thus, a large storage capacity will be required to accommodate the size of the videos. In order

to incorporate retinal video recordings as a future diabetic retinopathy screening tool via a

web-based telemedicine program, it will need to be readily compressible to a smaller file size

to increase the data transfer speed.

131

In our study, we have discovered that the difference in the size of a reading screen is a

potential factor which affects the detection of subtle diabetic retinopathy changes such as

microaneurysms. We have utilized the iMac 27 inch screen (Apple, California, US) in our

study as a reading monitor to avoid any bias secondary to reading screen resolution. As a

result, a further study will be of valuable to determine the appropriate minimum screen

resolution without compromising the detection of diabetic retinopathy.

Given that this is a relatively new technique, the cost effectiveness of using retinal video

screening still needs to be evaluated. The potential cost will involve purchasing a retinal

camera (with a storage system) which possesses retinal video recording function and hiring a

photographer who does not necessarily need to be a professional such as an optometrist or an

orthoptist. In order to evaluate the user friendliness of retinal video recording, a further study

could be carried out by recruiting both medical and non-medical personnel to perform and

grade the difficulty level of using this technique.

Interestingly, spontaneous pulsation of the central retinal vein was discovered incidentally in

the retinal videos. Loss of spontaneous venous pulsation is known to be associated with open

angle glaucoma.138 To date, spontaneous venous pulsation has not been utilized as a screening

criterion for glaucoma as it involves posterior segment examination and it is often hard to be

visualized using direct ophthalmoscopy by any primary health care physicians. Thus,

incorporating the loss of spontaneous of venous pulsation as one of the glaucoma screening

criteria using retinal video recording could be further explored in future studies.

In conclusion, we have demonstrated that the retinal video recording is equally as effective as

retinal photography in our study. Hence, it is a new and alternative technique to screen for

diabetic retinopathy. Although it will not be a substitute to the comprehensive ophthalmic

examination for diabetic patients, it offers the primary eye care providers the opportunity to

132

view a greater field of retinal view within a short period of time in the setting of diabetic

retinopathy screening. Given that it is quick and easy to perform, both medical and non-

medical personnel would be able to utilize it with minimal training. By making it easier to

screen and monitor diabetic retinopathy in the community, particularly in remote areas and

developing countries, this potentially sight-threatening condition may be diagnosed earlier

and treated appropriately.

133

8. CHAPTER 6: RETINAL VIDEO RECORDINGS AT DIFFERENT

COMPRESSION LEVELS: A NOVEL, VIDEO-BASED IMAGING

TECHNOLOGY FOR DIABETIC RETINOPATHY SCREENING

8.1 Summary

In the previous study (see Chapter 5), retinal video recordings saved in AVI format required

large storage capacity because of the large file sizes. To improve the user-friendliness of this

technique, we conducted a subsequent study to determine the maximum compression levels

for retinal video recordings while still preserving the sensitivity and specificity of DR

detection. This single-center study evaluated 36 retinal videos that were captured using

EyeScan (OIS, CA, USA) and compressed from the original uncompressed file size of 1 GB

to four different compression levels: 100 MB (Group 1), 30 MB (Group 2), 20 MB (Group 3)

and 5 MB (Group 4). The videos were subsequently interpreted by an ophthalmologist and a

resident using the International Clinical Diabetic Retinopathy Severity Scales. The main

outcome measures were the sensitivity, specificity and kappa coefficient for DR grading

detected by each compression level (groups 1–4), with reference to the original

uncompressed retinal videos. Groups 1–3, which were graded by both readers, had sensitivity

and specificity greater than 90% in detecting DR, whereas for group 4, the sensitivity and

specificity were 70.6% and 94.7% (ophthalmologist) and 80.0% and 72.2% (medical officer)

respectively. The kappa correlation in detecting DR for groups 1–3 was greater than 0.95,

whereas for group 4, the kappa was 0.76 and 0.66 for the ophthalmologist and medical officer

respectively. In conclusion, retinal video recording is a novel and effective DR screening

technique with high sensitivity, specificity and kappa correlation. Its compressibility makes it

134

a potentially effective technique that can be widely implemented in a routine, mobile and

tele-ophthalmology setting for DR screening services.

8.2 Introduction

Diabetes mellitus is a metabolic disorder characterized by hyperglycemia secondary to either

impaired insulin secretion or insulin resistance. The chronic uncontrolled hyperglycemia can

give rise to macrovascular (cerebrovascular accident, ischemic heart disease and peripheral

vascular disease) and microvascular (retinopathy, nephropathy and neuropathy)

complications. Diabetic retinopathy is one of the commonest microvascular complications of

diabetes. Nearly 100% type I diabetic patients and more than 60% type 2 diabetic patients

will have at least some retinopathy after 20 years of diabetes.1,2 Therefore, it is crucial for

primary eye care providers to regularly screen people with diabetes for diabetic retinopathy

as early detection can prevent severe visual impairment.3

Traditionally, retinal still photography has been the gold standard diabetic retinopathy

screening tool in the primary health care setting. However, the video-based imaging

technology using retinal video recording has been recently proposed to be a novel diabetic

retinopathy screening method.4 This technique is quick to perform, easy to learn (single day

training) and does not require any previous ophthalmic imaging experience. Compared with

the retinal still photography, the retinal video not only provides a greater field within shorter

period of time but also mimics what is seen with a slit lamp examination. Nevertheless, this

technique was limited by its large video file size which requires high storage capacity.4 On

average, a retinal video of 30 seconds takes up approximately 500 Megabytes (MB) and thus,

will be impractical to archive and transmit such large video files during routine diabetic

retinopathy screenings.

135

The objective of our study is to investigate and determine the optimal compression level for

retinal videos to screen for diabetic retinopathy. The effect of retinal still image compression

for diabetic retinopathy screening has been studied previously5-7 but none of the which has

investigated the use of retinal videos compression technique. Compression of retinal videos

will help to reduce the need for a large storage capacity and to increase the data transmission

speed for diabetic retinopathy screening in a routine, mobile and tele-ophthalmology setting.

8.3 Methods

Sample population

A total of 36 retinal videos (18 normal and 18 with diabetic retinopathy changes) at 4

different compression levels - 100 Megabyte (MB) (Group 1), 30MB (Group 2), 20MB

(Group 3) and 5MB (Group 4), captured by retinal video recording using OIS EyeScan

(Ophthalmic Imaging System, CA, US), were selected for our study. Given that three-field

(optic disc, macula and temporal views) retinal still photography has been the standard of

care for patients with diabetes in our center, the retinal video recordings were also captured in

a similar manner for the recruited patients in our study. All study subjects were enrolled from

the Diabetic Retinopathy Screening Clinic of Royal Perth Hospital and this study has been

approved by the Royal Perth Hospital Human Research Ethics Committee.


To allow for a power of 95%, desired precision of 0.10, expected sensitivity and specificity

of 96%, the total number of eyes required for each compression level was 31 (prevalence was

set at 0.50 as selected samples consisted of 50% normal and 50% abnormal retinal digital

videos).

136

Conversion process

A digital video can exist in various different file formats such as Moving Picture Experts

Group (MPEG), Audio Video Interleave (AVI), Windows Media Video (WMV), QuickTime

Movie File (MOVIE) and etc. It consists of a series of bitmap digital images displayed in a

rapid succession at a constant rate. The size of a video is determined by bit rates (BR) and

time (T). Bit rate, defined as the number of bits processed or conveyed per unit time,

represents the amount of information stored per unit of time of a recording. It is determined

by the frame rate (FR), frame size (FS) and color depth (CD) of a video. The frame rate is

defined as the number of displayed digital images per unit time and it is often measured in a

second (frames per second – FPS). The frame size is the total number of pixels in terms of

width (W) and height (H) of an image. The color depth (CD) represents the amount of bits

that form a single pixel. By altering one of these properties, one can modify the size of a

digital video using various video compression software packages and codecs which are

readily available on the Internet.

For our study, the uncompressed raw retinal videos were compressed to 4 different levels

(Group 1, 2, 3 and 4) from the original file size using a Video Converter Software, Xilisoft

Video Converter Ultimate 6.0 (Xilisoft Corporation, British Virgin Island), which utilizes a

standard video codec H.264 (Table 8.1). The average size of an original uncompressed 30-

second video was 500 MB. In our study, we reduced the 500 MB video file to 4 different

levels (Group 1: 100MB, Group 2: 30MB, Group 3: 20MB and Group 4: 5MB) by changing

the bit rates. The frame rate and frame size were set at 17 frames per second and 640 x 480

pixels respectively as these settings have been preset by the fundus camera (OIS EyeScan).

137

Reference Standard

We utilized the raw uncompressed raw retinal videos as the reference of our study as from

our previous study,4 the sensitivity and specificity of an uncompressed raw retinal videos in

detecting any grade of diabetic retinopathy by a retinal specialist and a consultant

ophthalmologist with special interest in diabetes, compared to the slit lamp examination by a

senior consultant ophthalmologist (with more than 10 years experience in screening diabetic

retinopathy), were more than 90%.

Data Interpretation

The converted videos were randomized and sent to two different readers (a consultant

ophthalmologist and a medical officer) who were blinded to the compression level of the

videos. Each reader interpreted a total of 144 retinal videos (36 retinal videos at 4 different

compression levels) using a standard monitor screen (iMac 27”, Apple, Cupertino, California,

USA) with a VLC media player 1.1.4 (Apple, California, USA) in a dimly lit room. The

International Clinical Diabetic Retinopathy Severity Scales8 was utilized to interpret and

grade the diabetic retinopathy severity by determining the presence/absence of lesions

including microaneuryms, retinal hemorrhages, cotton wool spots, venous beading,

intraretinal microvascular abnormalities, new vessels formation, vitreous hemorrhage,

preretinal hemorrhage and hard exudates. In addition, the retinal videos were rated as

‘acceptable’, ‘pixelated but interpretable’ and ‘unacceptable’ by the readers.

Data Analysis

The data analysis was performed using SPSS version 17 (SPSS, Chicago, IL, USA). The

main outcome measures were the sensitivity, specificity and Kappa coefficient between the

uncompressed raw retinal videos and compressed retinal videos in diagnosing any level of

138

diabetic retinopathy. In addition, the Kappa correlation was performed for other variables

such as 1) the microaneurysms and retinal hemorrhages between the uncompressed and

compressed retinal videos and; 2) the diabetic retinopathy grading between the consultant

ophthalmologist and medical officer using the uncompressed retinal videos. Values of kappa

of 0.8 and above were considered as excellent agreement between two groups for our study.9

6.4 Results

All retinal videos in group 1 (100MB), 2 (30MB) and 3 (20MB) were rated as ‘acceptable’ by

the ophthalmologist and medical officer (Table 8.2). For group 4 (10MB), only 11% and 3%

of the retinal videos were rated as ‘acceptable’ by the ophthalmologist and medical officer

respectively. Of the 18 retinal videos with diabetic retinopathy, 22% (n=4) had mild non-

proliferative diabetic retinopathy (NPDR), 61% (n=11) had moderate NPDR, 11% (n=2) had

severe NPDR and 6% (n=1) had proliferative diabetic retinopathy.

The conversion time for the retinal videos to different compression levels are shown in Table

8.3. Using the uncompressed videos as gold standard, group 1, 2 and 3 graded by both

readers had excellent sensitivity and specificity of more than 90% in detecting diabetic

retinopathy (Table 8.4). On the other hand, group 4 had the lowest sensitivity and specificity

in detecting diabetic retinopathy changes (ophthalmologist 1– sensitivity: 70.6%, specificity:

94.7%; medical officer: sensitivity: 80.0%, specificity: 72.2%).

Also, the Kappa between the gold standard uncompressed videos and group 1, 2 and 3 to

detect diabetic retinopathy-related changes for both readers were more than 0.96 whereas for

group 4, the Kappa for the ophthalmologist and medical officer were 0.76 and 0.66,

respectively. On the other hand, the Kappa between the ophthalmologist and medical officer

for raw uncompressed videos, group 1, group 2, group 3 and group 4 were 0.84, 0.84, 0.80,

0.76 and 0.43, respectively. Similarly, the Kappa correlation of the microaneurysms, retinal

139

hemorrhages, detected by both readers for group 1, 2 and 3 with reference to the gold

standard video files (1GB) were more than 0.9 (Table 8.5). For group 4, the Kappa for the

detection of microaneurysms by the ophthalmologist and medical officer were both 0.87

whereas for retinal hemorrhages, they were 0.88 and 0.87 respectively. The Kappa

correlation of new vessels formation and subhyaloid hemorrhage were 1.0 in group 1 to 4

graded by both ophthalmologist and medical officer, with reference to the uncompressed

videos.

Table 8.1: The file size of a retinal video with different compression levels by reducing

its bit rate while keeping other parameters constant (frame rate, frame size, zoom)

Groups Bit rate (kbps)

Approx. File Size for 60 Seconds

Compression Level Percentage of file size from its original size

Original Uncompressed Raw Video)

165000 1GB - 100%

1 15000 100MB 90%

10%

2 5000 30MB 97%

3%

3 3000 20MB 98%

2%

4 512 5MB 99%

1%

GB: Gigabytes

MB: Megabytes

The setting of other parameters:

1) Frame rate – 17 frames per second; 2) frame size – 640 x 480 pixels; 3) Zoom - Full

140

Table 8.2: The quality of retinal videos (with/without diabetic retinopathy changes) of

different compression levels rated by an ophthalmologist and a medical officer

Ophthalmologist Uninterpretable Pixelated but interpretable Acceptable

Group 1 (100MB) 0 (0%) 0 (0%) 36 (100%)

Group 2 (30MB) 0 (0%) 0 (0%) 36 (100%)

Group 3 (20 MB) 0 (0%) 0 (0%) 36 (100%)

Group 4 (5 MB) 3 (8.3%) 29 (80.6%) 4 (11.1%)

Medical officer

Group 1 (100MB) 0 (0%) 0 (0%) 36 (100%)

Group 2 (30MB) 0 (0%) 0 (0%) 36 (100%)

Group 3 (20 MB) 0% 0% 36 (100%)

Group 4 (5 MB) 3 (8.3%) 32 (88.9%) 1 (2.8%)

MB: Megabytes

Table 8.3: The average conversion timing of an uncompressed raw retinal video (1

Gigabytes) to different compression levels

Compression level Timings (seconds)

Group 1 (100MB) 25

Group 2 (30MB) 18

Group 3 (20 MB) 17

Group 4 (5 MB) 16

MB: Megabytes

The compression was performed using a 32-bit Windows XP, 2.5 RAM, Intel® Xeon®

X5550 Processor 2.67GHz with NVIDIA Quadro FX 580 graphics card (Timings may vary

on different computers with different performance).

141

Table 8.4: The sensitivity, specificity and Kappa correlation of different compression

levels for retinal videos in detecting diabetic retinopathy grading by an ophthalmologist and a medical officer with reference to uncompressed raw retinal videos (1GB)

Ophthalmologist Sensitivity (%) Specificity (%) Kappa

correlation (95% CI) (95% CI) Group 1 (100MB) 100 (78.1-100) 100 (78.1-100) 1.00 Group 2 (30MB) 100 (77.1-100) 94.7 (71.9-99.7) 0.96 Group 3 (20 MB) 94.4 (70.6-99.7) 100 (78.1-100) 0.96 Group 4 (5 MB) 70.6 (44.0-88.6) 94.7 (71.9-99.7) 0.76 Medical officer Group 1 (100MB) 100 (80.0-100) 100 (75.9-100) 1.00 Group 2 (30MB) 100 (80.0-100) 100 (75.90100) 1.00 Group 3 (20 MB) 100 (80.0-100) 93.8 (67.7-99.7) 0.96 Group 4 (5 MB) 80.0 (51.4-94.7) 72.2 (46.4-89.3) 0.66

GB: Gigabytes

MB: Megabytes

CI: Confidence Interval

Table 8.5: The Kappa correlation of the microaneurysms and retinal hemorrhages

detected by both ophthalmologist and medical officer at different compression levels with reference to the uncompressed raw video files (1GB)

Ophthalmologist Microaneurysms Retinal hemorrhages Group 1 (100MB) 1.00 1.00 Group 2 (30MB) 1.00 1.00 Group 3 (20 MB) 1.00 1.00 Group 4 (5 MB) 0.87 0.87 Medical officer Group 1 (100MB) 1.00 1.00 Group 2 (30MB) 1.00 1.00 Group 3 (20 MB) 0.94 1.00 Group 4 (5 MB) 0.88 0.87

GB: Gigabytes

MB: Megabytes

142

6.5 Discussion

Our study showed that the retinal videos could be significantly compressed down to 20MB of

its original file size with excellent sensitivity (ophthalmologist: 94.4%; medical officer:

100%) and specificity (ophthalmologist: 100%; medical officer: 93.8%) in detecting diabetic

retinopathy changes. The file size of a compressed 30-second retinal video consisting of optic

disc, macula and temporal views is approximately 20 MB. For an uncompressed color fundus

photo captured by FF450 plus (Carl Zeiss Meditec, Inc., CA, USA) in the Tagged Image File

Format (TIFF) or BIT MAP format, each photo takes up approximately 13 MB. In other

words, a total of 39 MB is required for a three-field color retinal still photography. In

comparison, a compressed video will have a lower storage capacity compared to the three

uncompressed raw color fundus photos.

For group 4 (5MB), most of the retinal videos were rated as ‘pixelated but interpretable’ by

both readers (Table 8.2). Nonetheless, they did not possess comparable sensitivity and

specificity with other three groups (Table 8.4). In addition, the feedback from the

ophthalmologist and medical officer were similar. Whilst reading the video recordings from

groups 1 to 3 was comfortable, interpreting the ‘pixelated’ videos was ‘extremely tiring’ and

‘time consuming’ as more time was required to differentiate a true lesion from the normal

retina and most of them were very ‘disjointed’ and ‘blurred’. Hence, this compression level

will not be ideal in the setting of diabetic retinopathy screening.

To the authors’ knowledge, no data was published on retinal digital video recording for

diabetic retinopathy screening. Given that this is one of the first studies which evaluate the

effectiveness of retinal video recordings at different compression levels, we selected the

uncompressed raw retinal videos which are considered to be of ‘good’ quality. Our results

indicated that a retinal video can be compressed down to 20MB without compromising its

143

quality and diagnostic accuracy. This study will need to be further expanded on retinal videos

with varying quality, especially ones with compromised quality due to media opacities and

dark fundi, to assess the effects of video compression.

The medical officer in this study had interpreted more than 1000 color fundus photos of

diabetic patients prior to this study. The difference in the detection of diabetic retinopathy

between the ophthalmologist and medical officer was minimal (Table 8.4). These results

indicated that the compressed retinal videos could be potentially interpreted by trained non-

ophthalmologist personnel and thus, the consultant specialist input can be redirected to more

useful areas such as provision of consultations and surgical intervention for patients with

sight-threatening diabetic retinopathy.

Only one retinal video was graded differently to the uncompressed recording by the

consultant ophthalmologist in group 2 and 3 and medical officer in group 3. Due to the

relatively small sample size, this has significantly reduced the sensitivity and specificity of

detecting diabetic retinopathy grading by the consultant ophthalmologist (specificity of 94.6%

in group 2 and sensitivity of 94.4% in group 3) and medical officer (specificity of 93.8% in

group 3) (Table 8.4). Despite fulfilling the sample size estimation, we felt that this was still

one of the weaknesses of our study and thus, further research with a larger sample size will

be of great value to explore the diagnostic accuracy of video compression at 30MB, 20 MB

and other compression levels such as 15 MB and 10 MB.

In detecting diabetic retinopathy changes for group 3 (20 MB) and 4 (5MB), the sensitivity of

the medical officer in detecting diabetic retinopathy changes was higher than the

ophthalmologist but in contrast, the ophthalmologist had a higher specificity than the medical

officer (Table 8.4). The difference in diagnostic sensitivity and specificity indicated that the

medical officer had a lower threshold in diagnosing a patient with diabetic retinopathy

144

compared with the ophthalmologist, especially in cases where he was uncertain about a

suspicious lesion. Although false positive diagnoses may result in more ‘unnecessary’

referrals to an ophthalmologist, it is critical to note that a screener should always refer the

patients with a suspicious lesion to an ophthalmologist to avoid any misdiagnosis of a sight-

threatening condition.

The Xilisoft Video Converter Ultimate 6.0 utilizes a standard video codec H.264 which is

compatible for both Windows and Macintosh. The retinal video recording device (OIS

EyeScan) does not possess a built-in retinal video compression program which performs

different compression levels for the retinal videos. Given that the OIS EyeScan is one of the

first fundus cameras to perform color retinal video recording, future research could be

conducted for OIS EyeScan or other devices to incorporate a built-in retinal video

compression program to further shorten the process of converting a retinal video.

Depending on each compression level, the conversion time for a retinal video from an

uncompressed format (500 MB) requires 16 to 25 seconds using a 32-bit Windows XP ( 2.5

RAM, Intel® Xeon® X5550 Processor 2.67GHz with NVIDIA Quadro FX 580 graphics card)

(Table 8.3). This timing may vary slightly on different computers with different performance.

This is a rapid conversion process as nearly 250 retinal videos can be converted within an

hour. Since the retinal videos can be readily compressible to be interpreted and stored, the

retinal video recording technique can be potentially utilized as a tele-ophthalmology

screening tool. It will be of great value to conduct further research to evaluate the cost and

clinical effectiveness of performing tele-retinal video recording for diabetic retinopathy

screening in the remote underserved areas.

In conclusion, a retinal video recording can be compressed from 500 MB to 20 MB whilst

maintaining its quality, sensitivity and specificity of diabetic retinopathy grading. As network

145

bandwidth is an issue in most of the rural and remote locations, the retinal video transmission

speed could be increased by reducing the file size without significant loss of diagnostic

information using efficient compression techniques. Given that the retinal videos are easily

compressible whilst retaining excellent diagnostic accuracy, retinal video recording may be

used as an alternative technique in diabetic retinopathy screening centers. As this technique is

still in its infancy, more research is required to improve the usability of this technique.

146

SECTION 5: DISCUSSION, CONCLUSIONS AND FUTURE DIRECTIONS

9. DISCUSSION

Diabetic Retinopathy Screening Practices

Early detection can prevent 98% of DR-related visual impairment.73 In Australia, patients

with diabetes often present to primary eye care physicians such as optometrists and GPs

before ophthalmologists, as most of them are often asymptomatic, especially during the

initial phase. Due to a rise in health consciousness, the majority of them are diagnosed with

diabetes during a routine health check. A small proportion of patients present with diabetes-

related systemic complications such as a cerebrovascular accident or ischemic heart disease.

Rarely, patients initially present with diabetic ocular complications such as diabetic macular

edema, vitreous hemorrhage, non-arteritic ischemic optic neuropathy, retinal vascular

occlusions or ischemic cranial neuropathy resulting in oculomotility disorders.

Given the importance of DR screening, it is important for primary eye care physicians to

understand and keep abreast of updated evidence-based DR guidelines. In 1997, the NHMRC

released the first national guidelines on DR management. Subsequently, McCarty et al.

conducted national surveys in 1999 and 2001 on DR management for optometrists and

ophthalmologists, but not GPs. In 2008, the revised NHMRC DR management guidelines

were released. It is important to study the current trend of DR management in the 10 years

since the first guidelines were released.

Compared to the screening practices reported in 1999 and 2001, we found that there is an

increasing trend of optometrists performing dilated fundoscopy on diabetic patients and using

recall notices, and they had greater confidence in detecting and managing DR changes (see

147

Chapter 1). Nearly 80% of optometrists had a ‘moderate’ to ‘strong’ desire to screen for DR.

This could reduce the waiting time for diabetes patients to see an ophthalmologist in the

public setting, hence allowing a better training system for patients with STDR.

From this study, we found that optometrists are more inclined to screen patients earlier than

the recommended timeframe. Despite the NHMRC only recommending first eye-screening at

puberty, almost 80% of optometrists would review a seven-year-old diabetic child in less

than five years, while 10% would refer the patient to the ophthalmologist. This management

pattern will not only place an unnecessary financial burden on the diabetic child’s family, but

it will also increase the waiting time to see an ophthalmologist. To improve DR screening

practices and reduce unnecessary reviews and referrals, optometrists should be encouraged to

read the guidelines at least once every few years or refer to the guidelines whenever they are

in doubt, as this study showed that even though 80% of optometrists reported receiving a

copy of the NHMRC guidelines, only two-thirds reported having read them at least once.

More than 50% of optometrists reported that they lacked confidence in detecting macular

edema, and only 40% would refer patients with macular edema to an ophthalmologist.

Interestingly, reading the guidelines did not seem to increase optometrists’ referrals of

patients with macular edema to an ophthalmologist. Given that macular edema is a major

cause of significant visual impairment, optometrists should improve their management

(confidence to detect and referrals) of this condition to ensure prompt laser treatment or

intravitreal anti-VEGF injection for patients with clinically significant macular edema. Any

reduction in visual acuity should raise suspicion and prompt a referral to an ophthalmologist,

as early stages of macular edema may be difficult to detect without indirect ophthalmoscopy

examination.

148

In the Caucasian population, concern about the use of dilating drops should not be a limiting

factor for dilated fundus examination, as the incidence of AACG is only one in 20,000.100 The

use of a retinal camera is associated with a stronger desire to screen and higher confidence in

detecting DR changes. Nevertheless, the retinal camera was the least commonly used

ophthalmic equipment among optometrists, with only 50% reporting using a retinal camera in

their practice on more than half of their diabetic patients. Of the various ophthalmic

equipment used, the direct ophthalmoscope was the most frequently used (72%), followed by

the slit lamp (65%) and the binocular indirect ophthalmoscope (56%). Hence, an economical

but effective retinal camera would be useful for optometrists to screen for DR.

Compared to optometrists, GPs had much less desire (40%) and confidence to screen for DR

in the community (see Chapter 2). Despite having sufficient knowledge in managing DR with

different severity levels (based on the hypothetical questions), the self-reported major

screening barriers for GPs were lack of confidence in detecting changes using dilated

fundoscopy (86%), followed by time limitations in consultations (86%), patients

unpreparedness to drive (64%) and fear of angle-closure glaucoma (44%). This is of major

concern, as this is a basic sight-saving skill that is taught and assessed at the medical school

level in Australia. In response to the two key areas (increasing early detection of DR and

improving access to eye health services) identified in the National Eye Health Framework,98

GPs should make it a habit to practice the dilated fundoscopy examination on a regular basis

in order to increase their confidence in detecting DR changes and to reduce unwarranted

referrals to ophthalmologists whenever possible.

In our survey, the majority of GPs (92%) were compliant with the recommended guidelines,

which state that HbA1c status must be assessed every six months or less for patients with

diabetes. However, based on Medicare’s data analysis, this figure is significantly lower in

149

reality (25%–80%), depicting a gap between what GPs think they should do and what is done

in reality on a routine basis. The other possibility is that the GPs who completed the survey

may have been more motivated to measure HbA1c than others. It is important for GPs to

strictly monitor and control the HbA1c of diabetic patients below 7%, as this has been shown

to significantly reduce diabetes-related complications.66

Comparing the 10 hypothetical case scenarios between the optometrists and GPs, the survey

found that GPs are generally more proficient in managing and screening DR in the

community, despite their low desire and confidence to detect DR changes in fundoscopy. For

diabetic maculopathy and severe NPDR, nearly 60% and 10% of optometrists respectively

would choose not to refer, whereas it was only 0.5% and 13% respectively for GPs.

In Chapter 1, we found that the retinal camera is associated with increased confidence in

detecting DR changes, including maculopathy, among optometrists. A pilot study of

photographic screenings by GPs for DR also found that GPs had good diagnostic accuracy

(sensitivity: 87%, specificity: 95%) in detecting any DR grade and, more importantly, they

were more willing to expand their role into DR screening if such infrastructure were readily

accessible. With the increasing burden on the Australian health care system, the use of

economical retinal cameras in GP clinics could be considered for DR screening in primary

health care settings in order to significantly reduce waiting times and costs to see an

ophthalmologist.

From the previous surveys, we found that the desire and confidence to screen for DR among

primary eye care providers were suboptimal. To improve DR screening interest in the

primary health care setting, we have explored the use of various novel screening modalities

(e.g., portable retinal camera, portable reading devices, video-based imaging technology) that

have not been reported in the past. Due to the need for an economical retinal camera for DR

150

screening, we developed and validated an economical and portable retinal camera called

EyeScan, which was later manufactured and marketed by the OIS (CA, USA) (see Chapter 3).

It weighs around 0.9 kg, has a maximum image capture rate of three images per second and a

5.3 MP sensor. It can capture anterior (cornea and lens) and posterior segment photos with

visible light or near infrared light at an affordable cost.

With reference to the slit lamp examination, our results showed comparable sensitivity,

specificity and kappa correlation in the overall DR grading between the medical officer and

retinal specialists (see Chapter 3). These results demonstrated that EyeScan is a portable,

light, economical and effective screening device for DR that can be utilized by an

inexperienced medical officer. Given that it requires small storage capacity for its lower-

resolution images (640 x 480 pixels), the available bandwidth would better handle the

transfer and archive of images, thereby allowing its use in a tele-retinal screening in rural

areas.

To evaluate the user-friendliness of the EyeScan machine, it was used by a medical officer

with no prior ophthalmic imaging experience, and this was compared to the FF450 images

captured by an experienced retinal photographer with 10 years’ experience. In the study, the

technical failure rate, defined as the fraction of ‘unacceptable’ retinal images, was not

statistically significant for both machines (EyeScan: 8.5% vs. FF450: 7%, χ²=0.23, p=0.63).

This showed that EyeScan could be utilized by non-experienced personnel with minimal

training. Further research will be of great value in evaluating the user-friendliness of EyeScan

for both medical and non-medical personnel in the primary health care setting.

Due to the growing popularity of mobile and tablet technologies worldwide, we also explored

the use of small viewing monitors (including the 15-inch MacBook Pro and the 9.7-inch iPad)

in DR screening to create a more convenient way for eye care providers to read retinal

151

images (see Chapter 4). In this study, we showed that both the retinal specialist and the

medical officer had excellent inter-observer agreement (kappa=0.88), sensitivity and

specificity (>90%) in the overall grading of DR on the 27-inch iMac, 15-inch MacBook Pro

and 9.7-inch iPad. In addition, the kappa correlation between the 27-inch iMac versus the 15-

inch MacBook Pro and the 9.7-inch iPad in detecting DR changes, including MAs, retinal

hemorrhages, CWSs, neovascularization and hard exudates, was greater than 0.8. These

results indicated that the retinal specialist and the non-specialist screener could effectively

interpret and diagnose DR from the color retinal still images using a reading screen as small

as 9.7 inches.

For sight-threatening retinopathy (severe NPDR and PDR), the sensitivity and specificity for

the retinal specialist were 100%, whereas for the medical officer, the sensitivity was 100%

but specificity was 97.7% on both the 15-inch MacBook Pro and the 9.7-inch iPad. The mild

discrepancy in the specificity between the retinal specialist and the medical officer resulted

from the over-estimation of two patients with severe NPDR instead of moderate NPDR. For

primary eye care providers, it is advisable to always be suspicious and have a lower threshold

to refer patients with uncertain DR lesions detected on retinal still images to see an

ophthalmologist, even if this results in some ‘unnecessary’ referrals to ophthalmologists.

In this study (see Chapter 4), we utilized a 27-inch iMac as the reference standard to avoid

any diagnostic errors caused by a small screen size and low screen resolution. Instead of the

gold-standard seven-field 30° stereoscopic view, we performed three-field retinal still

photography consisting of optic disc, macula and temporal views, as this is more time-saving

and practical for implementation in busy DR screening clinics. For the viewing software in

all of the laptops, we chose iPhoto as opposed to more specialized programs such as Visupac

or IMAGEnet system, as the former is free software that is included on all Apple computers.

152

Despite having a similar color depth for all screens, the 27-inch iMac, 15-inch MacBook Pro

and 9.7-inch iPad possess different color resolutions, color ranges and sets of colors.

Although this could affect the interpretation of retinal images, the study showed little to no

effect on the diagnosis of DR—especially sight-threatening DR. In view of the feasibility of

the 9.7-inch iPad in interpreting retinal images for DR screening, future studies should

evaluate the efficacy of other cheaper and smaller tablet computers or mobile devices (e.g.,

Samsung tablets, iPhone, mini iPad, LG or Sony). The study excluded suboptimal retinal

images resulting from media opacity or dark fundi. In the routine retinal still photographic

screening setting, patients with unreadable retinal images should be referred to an

ophthalmologist to determine the cause. Therefore, it will not be clinically significant to

compare these images on monitors of different sizes.

The EyeScan also possesses a retinal video recording function; to date, no studies have

reported on the use of retinal video recording in diagnosing DR (see Chapter 5). Retinal still

photography often relies on patients’ compliance to instructions and their tolerance to bright

light for good-quality retinal images. The technical failure rate for retinal still photography

could be as high as 12%—even when captured by an experienced photographer.82,127 Retinal

video recording is easier and faster to learn, and it provides a panoramic view of the retina

that simulates what is seen with a slit lamp examination. Each video consists of the optic disc,

macula and temporal view, lasting for 30–45 seconds. During the video recording, the patient

only needs to look straight ahead, and the process is slightly faster than the usual three-field

retinal still photography, which usually takes up to 90–120 seconds).

We compared the retinal video recordings and the retinal still photography using FF450, with

reference to slit lamp examination by two senior consultant ophthalmologists. All retinal

videos and images were de-identified, randomized and interpreted by two consultant

153

specialists. The results showed excellent sensitivity and specificity (greater than 90%) in

detecting any grade of DR for both retinal video recordings and retinal still photography, with

reference to slit lamp biomicroscopy. In addition, retinal video possessed a comparable kappa

coefficient (greater than 0.8) to retinal photography in detecting DR changes, including MAs,

retinal hemorrhages, CWSs, VB and IRMAs when compared with the slit lamp examination.

For macular edema, both imaging devices had a moderate kappa correlation with the

reference standard (slit lamp examination). As the retinal videos and still images only offer

two-dimensional retinal views, it is impossible to detect any signs of retinal thickening from

retinal photography or video alone. For retinal images or videos with hard exudates, MAs and

retinal hemorrhages within 500 um of the fovea, patients should be referred to an

ophthalmologist.

The retinal video recording was performed by an inexperienced medical officer with no prior

ocular imaging experience, whereas the retinal photography was performed by an

experienced retinal photographer. Nevertheless, the technical failure rate was similar for both

imaging modalities (retinal videos: 7.5%, retinal photography: 7%, X2=0.04, p=0.85),

suggesting that the retinal video recording is potentially an easy-to-operate technique. The

failed videos may have resulted from the irritation or exhaustion of patients’ eyes, as the

retinal videos were performed after the retinal still photography. As this is a relatively new

technique that is still in its infancy, the failed videos may in part be caused by the operator’s

technique. With improvement of the operator’s technique over time, the technical failure rate

may further diminish. The comparable technical failure rate showed that inexperienced

personnel, including GPs, nurses, allied health workers or volunteers, could be trained to

screen for DR within a short time, and this may increase DR screening rates in the

community—particularly in developing countries. Nevertheless, the use of this technique was

based on the technical failure rate and subjective experience of a sole user in this study.

154

Hence, further studies should recruit both experienced and inexperienced personnel in order

to widely assess and compare the usability of the technique of image acquisition with

standard retinal still photography in the DR screening setting.

One disadvantage of retinal video recordings is the constant light intensity throughout the

entire screening process, which may make it less tolerable. The videos were all saved in the

AVI format, which requires relatively higher storage capacity, as one second of video is

around 20 MB in size. Therefore, a 30-second retinal video could require up to 600 MB;

therefore, a large amount of storage is required. To incorporate retinal video recording as a

potential tele-retinal screening method, it would be useful to evaluate the optimal or

maximum video compression level for DR screening.

In Chapter 6, we evaluated 36 retinal video recordings in AVI format at four different

compression levels: 100 MB, 30 MB, 20 MB and 5 MB. All retinal videos were interpreted

by an ophthalmologist and a medical officer. The study showed that the retinal videos could

be compressed to 20 MB while retaining sufficient clarity/resolution to detect DR changes.

The sensitivity of the 20 MB retinal videos was interpreted as 94.4% (ophthalmologist) and

100% (medical officer) respectively, while the specificity was 100% and 93.8% respectively.

At 5 MB, almost all of the retinal videos were rated as ‘pixelated but interpretable’ by both

readers, but they did not possess comparable sensitivity and specificity with the other three

groups. In addition, the readers found it ‘extremely tiring’ and ‘time-consuming’, as more

time was required to differentiate a true lesion from the normal retina, and most of them were

very ‘disjointed’ and ‘blurred’.

The EyeScan did not have an automated video compression function; therefore, we utilized

software called Xilisoft Video Converter Ultimate 6.0 with video codec H.264, which is

compatible with both Windows and Macintosh. The conversion time for a retinal video from

155

an uncompressed format (500 MB) required 16–25 seconds using a 32-bit Windows XP. This

timing may vary slightly on different computers with different specifications. This is a rapid

conversion process, as nearly 250 retinal videos can be converted in one hour. As the retinal

videos can be readily compressible to be interpreted and stored, the retinal video recording

technique can potentially be utilized as a tele-ophthalmology screening tool. It would be

useful to conduct further research to evaluate the cost and clinical effectiveness of performing

tele-retinal video recording for DR screening in remote, underserved areas. Further, it would

be useful if the manufacturer incorporated a built-in retinal video compression program to

save time during the DR screening process.

156

10. CONCLUSIONS

Early detection can prevent most permanent DR-related visual impairment; thus, primary eye

care providers such as optometrists and GPs play a significant role in performing DR

screening in the community. Given that the risk of acute-angle closure is low, secondary to

pupillary dilation in the Caucasian population, dilated fundoscopy is a basic examination

technique that should be used on all patients with diabetes. Alternatively, retinal still

photography or retinal video recording can be utilized to screen for DR. The implementation

of large-scale, cost-effective community DR screening requires an economical and effective

retinal camera (e.g., EyeScan) that is used by trained medical or non-medical personnel, with

primary eye care providers such as optometrists and GPs grading the retinal images. In

addition to the available DR screening tools, retinal video recording is a novel DR screening

method that can potentially be utilized by primary eye care providers. More research studies

need to be conducted to evaluate the cost and clinical effectiveness of large-scale retinal

video DR screening in the community and in a tele-ophthalmology setting. Optimizing the

use of human resources and embracing new and affordable technology will improve access to

ophthalmologists and allow earlier detection among patients who require treatment (e.g.,

lasers or anti-VEGF), thereby reducing the effect of DM-related visual impairment.

157

11. FUTURE DIRECTIONS

The following areas should be considered in order to increase the desire and interest to screen

for DR in the community.

1. The use of a multifunctional, portable retinal camera with in-built, automated image

analysis:

To improve DR screening in the community, both medical and non-medical personnel

should be trained to operate portable, low-cost, non-mydriatic retinal cameras with

automated focusing and image capturing. This will significantly reduce the learning curve

and waiting time for DR screening.

2. Large-scale video screening for DR in the routine, mobile and tele-ophthalmology setting:

Retinal video recording has been shown to be an easy-to-use and effective novel DR

screening modality. It would be useful to further develop the retinal video camera with

in-built retinal video compression software and to subsequently evaluate the user-

friendliness and cost-effectiveness of this DR screening modality in various urban and

rural practices. More research should be conducted in rural practices to evaluate the

minimum bandwidth required for the network in order to transmit and store retinal videos

effectively in a tele-retinal video screening setting.

3. The use of a smart phone to image and diagnose DR screening:

Smart phones and tablets have been shown to be effective tools for reading retinal images.

Recently, it has been shown that smart phones (Portable Eye Examination Kit (PEEK))

can be used to help prevent blindness in low-income countries. More research should be

conducted to explore the usage of the camera and software in other tablets and mobile

devices in order to make DR screening more accessible.

158

4. Follow-up national survey on DR management by primary eye care providers:

It is important to understand and survey DR screening in the primary eye care setting to

prevent DR-related blindness in Australia. It would be useful to conduct a follow-up

national survey on DR management by optometrists, GPs and other allied health

professionals following the release of the 2008 NHMRC guidelines. It would also be

useful to further explore the mode of DR screening by primary eye care providers in

Australia to include the brand or cost of retinal cameras, as well as screening modalities

such as two- or three-field screening, red free, video screening or tele-retinal screening.

159

12. REFERENCES

1. Wild S, Roglic G, Green A, Sicree R, King H. Global prevalence of diabetes: estimates

for the year 2000 and projections for 2030. Diabetes Care 2004;27:1047–53.

2. Tapp RJ, Shaw JE, Harper CA, de Courten MP, Balkou B, McCarty DJ, et al. The

prevalence of and factors associated with diabetic retinopathy in the Australian

population. Diabetes Care 2003;26:1731–7.

3. McCarty CA, McKay R, Keeffe JE. Management of diabetic retinopathy by Australian

optometrists. Working Group on Evaluation of NHMRC Retinopathy Guideline

Distribution. National Health and Medical Research Council. Aust N Z J Ophthalmol

1999;27:404–9.

4. McCarty CA, Taylor KI, McKay R, Keeffe JE. Diabetic retinopathy: effects of national

guidelines on the referral, examination and treatment practices of ophthalmologists and

optometrists. Clin Experiment Ophthalmol 2001;29:52–8.

5. McCarty CA, Taylor KI, Keeffe JE. Management of diabetic retinopathy by general

practitioners in Victoria. Clin Experiment Ophthalmol 2001;29:12–6.

6. Hutchinson A, McIntosh A, Peters J, O'Keeffe C, Khunti K, Baker R, et al. Effectiveness

of screening and monitoring tests for diabetic retinopathy—A systematic review. Diabet

Med 2000;17:495–506.

7. Lwanga SK, Lemeshow, S. Sample size determination in health studies: a practical

manual. Geneva: World Health Organization; 1991.

8. Wilkinson CP, Ferris FL 3rd, Klein RE, Lee PP, Agardh CD, Davis M, et al. Proposed

international clinical diabetic retinopathy and diabetic macular edema disease severity

scales. Ophthalmology 2003;110:1677–82.

9. Early Treatment Diabetic Retinopathy Study Research Group. Grading diabetic

retinopathy from stereoscopic color fundus photographs—An extension of the modified

Airlie House classification. ETDRS report number 10. Early Treatment Diabetic

Retinopathy Study Research Group. Ophthalmology 1991;98:786–806.

10. Scanlon PH, Malhotra R, Greenwood RH, Aldington SJ, Foy C, Flatman M, et al.

Comparison of two reference standards in validating two field mydriatic digital

photography as a method of screening for diabetic retinopathy. Br J Ophthalmol

2003;87:1258–63.

11. American Diabetes Association. Diagnosis and classification of diabetes mellitus.

Diabetes Care 2010;33:S62–9.

160

12. Genuth S, Alberti KG, Bennett P, Buse J, Defronzo R, Kahn R, et al. Follow-up report on

the diagnosis of diabetes mellitus. Diabetes Care 2003;26:3160–7.

13. World Health Organization. Definition, diagnosis and classification of diabetes mellitus

and its complications: report of a WHO Consultation—Part 1: diagnosis and classification

of diabetes mellitus; 2006.

14. McCullough D. Classification of diabetes mellitus and genetic diabetic syndromes.

UpToDate; 2010.

15. World Health Organization. Definition, diagnosis and classification of diabetes mellitus

and its complications—Part 1: diagnosis and classification of diabetes mellitus; 1999.

16. Beck-Nielsen H, Groop LC. Metabolic and genetic characterization of prediabetic states:

sequence of events leading to non-insulin-dependent diabetes mellitus. J Clin Invest

1994;94:1714–21.

17. Kahn CR. Banting lecture: insulin action, diabetogenes, and the cause of type II diabetes.

Diabetes 1994;43:1066–84.

18. Robertson RP. Antagonist: diabetes and insulin resistance—Philosophy, science, and the

multiplier hypothesis. J Lab Clin Med 1995;125:560–4.

19. McCullough D, Robertson RP. Pathogenesis of type 2 diabetes mellitus. UpToDate; 2010.

20. DeFronzo RA, Ferrannini E. Insulin resistance: a multifaceted syndrome responsible for

NIDDM, obesity, hypertension, dyslipidemia, and atherosclerotic cardiovascular disease.

Diabetes Care 1991;14:173–94.

21. Shaw JE, Sicree RA, Zimmet PZ. Global estimates of the prevalence of diabetes for 2010

and 2030. Diabetes Res Clin Pr 2010;87:4–14.

22. World Health Organization. Diabetes facts. 2010.

http://www.worlddiabetesfoundation.org/composite-35.htm (accessed February 13, 2011).

23. Colagiuri S, Borch-Johnsen K, Glumer C, Vistisen D. There really is an epidemic of type

2 diabetes. Diabetologia 2005;48:1459–63.

24. International Diabetes Federation. Diabetes atlas, 2nd ed. Belgium; 2003.

25. Dunstan DW, Zimmet PZ, Welborn TA, de Courten MP, Cameron AJ, Sicree RA, et al.

The rising prevalence of diabetes and impaired glucose tolerance: the Australian Diabetes,

Obesity and Lifestyle Study. Diabetes Care 2002;25:829–34.

26. Daniel M, Rowley KG, McDermott R, O'Dea K. Diabetes and impaired glucose tolerance

in Aboriginal Australians: prevalence and risk. Diabetes Res Clin Pr 2002;57:23–33.

161

27. McCulloch DK. Classification of diabetes mellitus and genetic diabetic syndromes.

UpToDate; 2010.

28. Zimmet P, Alberti KG, Shaw J. Global and societal implications of the diabetes epidemic.

Nature 2001;414:782–7.

29. Ismail AA, Gill GV. The epidemiology of type 2 diabetes and its current measurement.

Best Pract and Res Cl En 1999;13:197–220.

30. Felner EI, Klitz W, Ham M, Lazaro AM, Stastny P, Dupont B, et al. Genetic interaction

among three genomic regions creates distinct contributions to early- and late-onset type 1

diabetes mellitus. Pediatr Diabetes 2005;6:213–20.

31. Durruty P, Ruiz F, de los Ros MG. Age at diagnosis and seasonal variation in the onset of

insulin-dependent diabetes in Chile (southern hemisphere). Diabetologia 1979;17:357–60.

32. Elamin A, Omer MI, Zein K, Tuvemo T. Epidemiology of childhood type I diabetes in

Sudan, 1987–1990. Diabetes Care 1992;15:1556–9.

33. Silink M. Childhood diabetes: a global perspective. Horm Res 2002;57:1–5.

34. Karvonen M, Viik-Kajander M, Moltchanova E, Libman I, LaPorte R, Tuomilehto J.

Incidence of childhood type 1 diabetes worldwide. Diabetes Mondiale (DiaMond) Project

Group. Diabetes Care 2000;23:1516–26.

35. Tuomilehto J, Podar T, Tuomilehto-Wolf E, Virtala E. Evidence for importance of gender

and birth cohort for risk of IDDM in offspring of IDDM parents. Diabetologia

1995;38:975–82.

36. Guo S-W, Tuomilehto J. Preferential transmission of type 1 diabetes from parents to

offspring: fact or artifact? Genet Epidemiol 2002;23:323–34.

37. Tillil H, Kbberling J. Age-corrected empirical genetic risk estimates for first-degree

relatives of IDDM patients. Diabetes 1987;36:93–9.

38. Graves EJ, Gillum BS. Detailed diagnoses and procedures, National Hospital Discharge

Survey, 1995. Vital Health Stat 13 1997:1–146.

39. Kitabchi AE, Umpierrez GE, Murphy MB, Barrett EJ, Kreisberg RA, Malone JI, et al.

Management of hyperglycemic crises in patients with diabetes. Diabetes Care

2001;24:131–53.

40. Campbell IW, Duncan LJP, Innes JA, MacCuish AC, Munro JF. Abdominal pain in

diabetic metabolic decompensation. J Amer Med Assoc1975;233:166–8.

41. Nugent B. Hyperosmolar hyperglycemic state. Emerg Med Clin N Am 2005;23:629–48,

vii.

162

42. Trence DL, Hirsch IB. Hyperglycemic crises in diabetes mellitus type 2. Endocrinol

Metab Clin N Am 2001;30:817–31.

43. Williams R, Airey M, Baxter H, Forrester J, Kennedy-Martin T, Girach A. Epidemiology

of diabetic retinopathy and macular oedema: a systematic review. Eye 2004;18:963–83.

44. Mohamed Q, Gillies MC, Wong TY. Management of diabetic retinopathy: a systematic

review. J Amer Med Assoc 2007;298:902–16.

45. Davis MD, Fisher MR, Gangnon RE, Barton F, Aiello LM, Chew EY, et al. Risk factors

for high-risk proliferative diabetic retinopathy and severe visual loss: early Treatment

Diabetic Retinopathy Study Report #18. Invest Ophth Vis Sci 1998;39:233–52.

46. Photocoagulation for diabetic macular edema. Early Treatment Diabetic Retinopathy

Study report number 1. Early Treatment Diabetic Retinopathy Study research group. Arch

Ophthalmol 1985;103:1796–806.

47. Klein R, Klein BE, Moss SE, Davis MD, DeMets DL. Glycosylated hemoglobin predicts

the incidence and progression of diabetic retinopathy. JAMA 1988; 260: 2864-71.

48. Klein R, Klein BE, Moss SE. Relation of glycemic control to diabetic microvascular

complications in diabetes mellitus. Ann Intern Med 1996; 124: 90-6.

49. Sheetz MJ, King GL. Molecular understanding of hyperglycemia's adverse effects for

diabetic complications. JAMA 2002; 288: 2579-88.

50. Gabbay KH. The sorbitol pathway and the complications of diabetes. N Engl J Med. 1973;

288: 831-6.

51. Patel V, Rassam S, Newsom R, Wiek J, Kohner E. Retinal blood flow in diabetic

retinopathy. BMJ 1992; 305: 678-83.

52. Larkins RG, Dunlop ME, Johnson EI. Norman MacAlister Gregg Lecture. The

pathogenesis of diabetic retinopathy. Aust N Z J Ophthalmol 1996; 24: 97-104.

53. Chaturvedi N. Modulation of the renin-angiotensin system and retinopathy. Heart 2000;

84 Suppl 1: i29-31:discussion i50.

54. Makimattila S, Summanen P, Matinlauri I, et al. Serum total renin, an independent

marker of the activity and severity of retinopathy in patients with IDDM. Br J

Ophthalmol 1998; 82: 939-44.

55. Wagner J, Jan Danser AH, Derkx FH, et al. Demonstration of renin mRNA,

angiotensinogen mRNA, and angiotensin converting enzyme mRNA expression in the

human eye: evidence for an intraocular renin-angiotensin system. Br J Ophthalmol 1996;

80: 159-63.

163

56. Phipps JA, Jobling AI, Greferath U, Fletcher EL, Vessey KA. Alternative pathways in the

development of diabetic retinopathy: the renin-angiotensin and kallikrein-kinin systems.

Clin Exp Optom 2012; 95: 282-9.

57. Watanabe D, Suzuma K, Matsui S, et al. Erythropoietin as a retinal angiogenic factor in

proliferative diabetic retinopathy. N Engl J Med. 2005; 353: 782-92.

58. Sebag J, McMeel JW. Diabetic retinopathy. Pathogenesis and the role of retina-derived

growth factor in angiogenesis. Surv. Ophthalmol. 1986; 30: 377-84.

59. Yamagishi S, Matsui T, Nakamura K, Takeuchi M, Imaizumi T. Pigment epithelium-

derived factor (PEDF) prevents diabetes- or advanced glycation end products (AGE)-

elicited retinal leukostasis. Microvasc. Res. 2006; 72: 86-90.

60. Gariano RF, Gardner TW. Retinal angiogenesis in development and disease. Nature 2005;

438: 960-6.

61. Morello CM. Etiology and natural history of diabetic retinopathy: an overview. Am J

Health Syst Pharm 2007;64:S3–7.

62. Resnikoff S, Pascolini D, Etya'ale D, Kocur I, Pararajasegaram R, Pokharel, GP, et al.

Global data on visual impairment in the year 2002. B World Health Organ 2004;82:844–

51.

63. Kempen JH, O'Colmain BJ, Leske MC, Haffner SM, Klein R, Moss SE, et al. The

prevalence of diabetic retinopathy among adults in the United States. Arch Ophthalmol

2004;122:552–63.

64. Roy MS, Klein R, O'Colmain BJ, Klein BE, Moss SE, Kempen JH. The prevalence of

diabetic retinopathy among adult type 1 diabetic persons in the United States. Arch

Ophthalmol 2004;122:546–51.

65. Younis N, Broadbent DM, Vora JP, Harding SP. Incidence of sight-threatening

retinopathy in patients with type 2 diabetes in the Liverpool Diabetic Eye Study: a cohort

study. The Lancet 2003;361:195–200.

66. Stratton IM, Kohner EM, Aldington SJ, Turner RC, Holman RR, Manley SE, et al.

UKPDS 50: risk factors for incidence and progression of retinopathy in type II diabetes

over 6 years from diagnosis. Diabetologia 2001;44:156–63.

67. Mitchell P, Smith W, Wang JJ, Attebo K. Prevalence of diabetic retinopathy in an older

community: the Blue Mountains Eye Study. Ophthalmology 1998;105:406–11.

68. McKay R, McCarty CA, Taylor HR. Diabetic retinopathy in Victoria, Australia: the

Visual Impairment Project. Brit J Ophthalmol 2000;84:865–70.

164

69. Mitchell P, Moffitt PS, Beaumont P. Prevalence of vision-threatening diabetic retinopathy

in Newcastle, Australia. Tohoku J Exp Med 1983;14:379–83.

70. Mitchell P. Development and progression of diabetic eye disease in Newcastle (1977–

1984): rates and risk factors. Aust NZ J Ophthalmol 1985;13:39–44.

71. Tapp RJ, Zimmet PZ, Harper CA, de Courten MP, Balkau B, McCarty DJ, et al. Diabetes

care in an Australian population: frequency of screening examinations for eye and foot

complications of diabetes. Diabetes Care 2004;27:688–93.

72. Colagiuri S. National Diabetes Strategy and Implementation Plan. Canberra: Diabetes

Australia; 1998.

73. Ferris FL 3rd. How effective are treatments for diabetic retinopathy? J Amer Med Assoc

1993;269:1290–1.

74. Goldberg M, Fine S. Symposium on the treatment of diabetic retinopathy. Airlie House,

Warrenton, Virginia: Department of Health, Education and Welfare, 1968.

75. Davis MD, Hubbard LD, Trautman J, Klein R. Conference on insulin pump therapy in

diabetes. Multicenter study effect on microvascular disease. Studies of retinopathy.

Methodology for assessment and classification with fundus photographs. Diabetes

1985;34:42–9.

76. Kohner EM, Porta M. Screening for diabetic retinopathy in Europe: a field guide-book.

Denmark: WHO; 1992.

77. Murray R, Metcalf SM, Lewis PM, Mein JK, McAllister IL. Sustaining remote-area

programs: retinal camera use by Aboriginal health workers and nurses in a Kimberley

partnership. Med J Aust 2005;182:520–3.

78. Hammond C, Shackleton J, Flanagan DW, Herrtage J, Wade J. Comparison between an

ophthalmic optician and ophthalmologist in screening for diabetic retinopathy. Eye

1996:10:107–12.

79. Buxton M, Sculpher MJ, Ferguson BA, Humphreys JE, Altman JF, Spiegelhalter DJ, et al.

Screening for treatable diabetic retinopathy: a comparison of different methods. Diabet

Med 1991;8:371–7.

80. Lienert R. Inter-observer comparisons of ophthalmoscopic assessment of diabetic

retinopathy. Aust NZ J Ophthalmol 1989;17:363–8.

81. National Institute for Clinical Excellence. Diabetic retinopathy—Early management and

screening. London; 2001.

165

82. Olson JA, Strachan FM, Hipwell JH, Goatman KA, McHardy KC, Forrester JV, et al. A

comparative evaluation of digital imaging, retinal photography and optometrist

examination in screening for diabetic retinopathy. Diabet Med 2003;20:528–34.

83. Lin D, Blumenkranz M, Brothers R, Grosvenor D. The sensitivity and specificity of

single-field nonmydriatic monochromatic digital fundus photography with remote image

interpretation for diabetic retinopathy screening: a comparison with ophthalmoscopy and

standardized mydriatic color photography. Am J Ophthalmol 2002;134:204–13.

84. Fransen S, Leonard-Martin TC, Feuer WJ, Hildebrand PL. Clinical evaluation of patients

with diabetic retinopathy: accuracy of the Inoveon diabetic retinopathy-3DT system.

Ophthalmology 2002;109:595–601.

85. Liew G, Mitchell P, Wang JJ, Wong TY. Fundoscopy: to dilate or not to dilate? Brit Med

J 2006;332:179.

86. Inan U, Ozturk F, Ermis SS. Pharmacologic pupil dilation in diabetic patients. Retina

2003;23:254–6.

87. Miura M, Hong YJ, Yasuno Y, Muramatsu D, Iwasaki T, Goto H. Three-dimensional

vascular imaging of proliferative diabetic retinopathy by Doppler optical coherence

tomography. Am J Ophthalmol 2015; 159: 528-38 e3.

88. Silva PS, Cavallerano JD, Tolls D, et al. Potential efficiency benefits of nonmydriatic

ultrawide field retinal imaging in an ocular telehealth diabetic retinopathy program.

Diabetes care 2014; 37: 50-5.

89. Wessel MM, Aaker GD, Parlitsis G, Cho M, D'Amico DJ, Kiss S. Ultra-wide-field

angiography improves the detection and classification of diabetic retinopathy. Retina

2012; 32: 785-91.

90. Kim JE, Chung M. Adaptive optics for retinal imaging: current status. Retina 2013; 33:

1483-6.

91. Lombardo M, Parravano M, Serrao S, Ducoli P, Stirpe M, Lombardo G. Analysis of

retinal capillaries in patients with type 1 diabetes and nonproliferative diabetic

retinopathy using adaptive optics imaging. Retina 2013; 33: 1630-9.

92. Lombardo M, Parravano M, Lombardo G, et al. Adaptive optics imaging of parafoveal

cones in type 1 diabetes. Retina 2014; 34: 546-57.

93. Nelson DA, Krupsky S, Pollack A, et al. Special report: Noninvasive multi-parameter

functional optical imaging of the eye. Ophthalmic Surg Lasers Imaging 2005; 36: 57-66.

166

94. Burgansky-Eliash Z, Nelson DA, Bar-Tal OP, Lowenstein A, Grinvald A, Barak A.

Reduced retinal blood flow velocity in diabetic retinopathy. Retina 2010; 30: 765-73.

95. Field MG, Elner VM, Puro DG, et al. Rapid, noninvasive detection of diabetes-induced

retinal metabolic stress. Arch Ophthalmology 2008; 126: 934-8.

96. National Health and Medical Research Council. Guidelines for the Management of

Diabetic Retinopathy 2008. http://www.nhmrc.gov.au/publications/synopses/di15syn.htm.

97. National Health and Medical Research Council. Clinical Practice Guideline: Management

of Diabetic Retinopathy. 1997.

http://www.nhmrc.gov.au/publications/synopses/cp53covr.htm.

98. Australian Institute of Health and Welfare (AIHW). The tenth biennial health report of

the Australian Institute of Health and Welfare. 2006.

99. Grimshaw JM, Russell IT. Achieving health gain through clinical guidelines II: ensuring

guidelines change medical practice. Qual Health Care 1994;3:45–52.

100. Talks SJ, Tsaloumas M, Mission GP, Gibson JM. Angle closure glaucoma and diagnostic

mydriasis. The Lancet 1993;342:1493–4.

101. Baeza M, Orozco-Beltran D, Gil-Guillen VF, Pedrera V, Ribera MC, Pertusa S, et al.

Screening for sight threatening diabetic retinopathy using non-mydriatic retinal camera in

a primary care setting: to dilate or not to dilate? Int J Clin Pract 2009;63:433–8.

102. Dickson PR, McCarty CA, Keeffe JE, Baxter R, Harper CA, Taylor HR. Diabetic

retinopathy: examination practices and referral patterns of general practitioners. Med J

Aust 1996;164:341–4.

103. Royal Australian College of General Practitioners. Management of diabetes mellitus.

104. Therapeutic Guideline—Endocrinology. Management of diabetes mellitus.

105. National Prescribing Service Limited Guideline. Management of diabetes mellitus.

106. American Diabetes Association. Management of diabetes mellitus.

107. Australian Government Department of Health and Ageing. Australian health Ministers'

Conference: The National Eye Health Framework: action to promote eye health and

prevent avoidable blindness and vision loss. 2004.

108. Ting DSW, Ng J, Morlet N, Yuen J, Clark A, Taylor H, et al. Australian Optometrists

National Survey of Diabetic Retinopathy Management. Royal Australian New Zealand

College of Ophthalmology (RANZCO) Annual General Meeting and Scientific

Conference. Brisbane, Australia.; 2010.

167

109. Askew D, Schluter PJ, Spurling G, et al. Diabetic retinopathy screening in general

practice: a pilot study. Aust Fam Physician 2009;38:650–6.

110. Photocoagulation for diabetic macular edema: Early Treatment Diabetic Retinopathy

Study report number 1: Early Treatment Diabetic Retinopathy Study research group.

Arch Ophthalmol 1985;103:1796–806.

111. Photocoagulation treatment of proliferative diabetic retinopathy. Clinical application of

Diabetic Retinopathy Study (DRS) findings, Diabetic Retinopathy Study report number 8:

Diabetic Retinopathy Study research group. Ophthalmology 1981;88:583–600.

112. Ting D, Ng J, Morlet N, Yuen J, Clark A, Taylor HR, et al. Diabetic retinopathy

management by Australian optometrists. Clin Experiment Ophthalmol 2011;39:230–5.

113. Arun CS, Al-Bermani A, Stannard K, Taylor R. Long-term impact of retinal screening on

significant diabetes-related visual impairment in the working age population. Diabet Med

2009;26:489–92.

114. James M, Turner DA, Broadbent DM, Vora J, Harding SP. Cost effectiveness analysis of

screening for sight threatening diabetic eye disease. Brit Med J 2000;320:1627–31.

115. Garvican L, Scanlon PH. A pilot quality assurance scheme for diabetic retinopathy risk

reduction programmes. Diabet Med 2004;21:1066–74.

116. Chylack Jr LT, Wolfe JK, Singer DM, Leske MC, Bullimore MA, Bailey IL, et al. The

Lens Opacities Classification System III. Arch Ophthalmol 1993;111:831–6.

117. Bragge P, Gruen RL, Chau M, Forbes A, Taylor HR. Screening for presence or absence

of diabetic retinopathy: a meta-analysis. Arch Ophthalmol 2011;129:435–44.

118. Fleming A, Goatman K, Philip S, Prescott G, Sharp P, Olson J. Automated grading for

diabetic retinopathy: a large-scale audit using arbitration by clinical experts. Brit J

Ophthalmol 2010;94:1606–10.

119. Ting DSW, Tay-Kearney ML, Yogesan K. A light and portable novel device for diabetic

retinopathy screening. Clin Experiment Ophthalmol 2012;40;40-46.

120. Ting DSW, Tay-Kearney ML, Constable IJ, Liam L, Yogesan K. Retinal video recording:

a new way to image and diagnose diabetic retinopathy. Ophthalmology 2011; 118:1588-

93.

121. Wright SE, McKay R, Taylor KI, Keeffe JE, McCarty CA. Changes in attitudes and

practices of optometrists in their management of diabetic retinopathy after the release of

NHMRC guidelines. National Health and Medical Research Council. Clin Exp

Ophthalmol 2001;29:121–4.

168

122. Moss SE, Klein R, Klein BE. The 14-year incidence of visual loss in a diabetic

population. Ophthalmology 1998;105:998–1003.

123. Klein R, Klein BE, Moss SE, Davis MD, DeMets DL. The Wisconsin epidemiologic

study of diabetic retinopathy. III. Prevalence and risk of diabetic retinopathy when age at

diagnosis is 30 or more years. Arch Ophthalmol 1984;102:527–32.

124. Klein R, Klein BE, Moss SE, Davis MD, DeMets DL. The Wisconsin epidemiologic

study of diabetic retinopathy. II. Prevalence and risk of diabetic retinopathy when age at

diagnosis is less than 30 years. Arch Ophthalmol 1984;102:520–6.

125. Javitt JC, Aiello LP, Chiang Y, Ferris FL 3rd, Canner JK, Greenfield S. Preventive eye

care in people with diabetes is cost-saving to the federal government: implications for

health-care reform. Diabetes Care 1994;17:909–17.

126. Javitt JC, Aiello LP, Bassi LJ, Chiang YP, Canner JK. Detecting and treating retinopathy

in patients with type I diabetes mellitus: savings associated with improved

implementation of current guidelines. American Academy of Ophthalmology.

Ophthalmology 1991;98:1565–73.

127. Javitt JC, Aiello LP. Cost-effectiveness of detecting and treating diabetic retinopathy.

Ann Intern Med 1996;124:164–9.

128. Bursell SE, Cavallerano JD, Cavallerano AA, Clermont AC, Birkmire-Peters D, Aiello

LP, et al. Stereo nonmydriatic digital-video color retinal imaging compared with Early

Treatment Diabetic Retinopathy Study seven standard field 35-mm stereo color photos

for determining level of diabetic retinopathy. Ophthalmology 2001;108:572–85.

129. Conlin PR, Fisch BM, Orcutt JC, Hetrick BJ, Darkins AW. Framework for a national

teleretinal imaging program to screen for diabetic retinopathy in Veterans Health

Administration patients. J Rehabil Res Dev 2006;43:741–8.

130. Cavallerano AA, Cavallerano JD, Katalinic P, Blake B, Rynne M, Conlin PR, et al. A

telemedicine program for diabetic retinopathy in a Veterans Affairs Medical Center: the

Joslin Vision Network Eye Health Care Model. Am J Ophthalmol 2005;139:597–604.

131. National Health Service (NHS). National Screening Programme for Diabetic Retinopathy.

132. Kumar S, Tay-Kearney ML, Constable IJ, Yogesan K. Internet based ophthalmology

service: impact assessment. Brit J Ophthalmol 2005;89:1382–3.

133. Kumar S, Tay-Kearney ML, Chaves F, Constable IJ, Yogesan K. Remote ophthalmology

services: cost comparison of telemedicine and alternative service delivery options. J

Telemed Telecare 2006;12:10–22.

169

134. Williams GA, Scott IU, Haller JA, Maguire AM, Marcus D, McDonald HR. Single-field

fundus photography for diabetic retinopathy screening: a report by the American

Academy of Ophthalmology. Ophthalmology 2004;111:1055–62.

135. Aptel F, Denis P, Rouberol F, Thivolet C. Screening of diabetic retinopathy: effect of

field number and mydriasis on sensitivity and specificity of digital fundus photography.

Diabetes Metab 2008;34:290–3.

136. Harding SP, Broadbent DM, Neoh C, White MC, Vora J. Sensitivity and specificity of

photography and direct ophthalmoscopy in screening for sight threatening eye disease:

the Liverpool Diabetic Eye Study. Brit Med J 1995;311:1131–5.

137. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas 1960;20:37–

46.

138. Legler U, Jonas JB. Frequency of spontaneous pulsations of the central retinal vein in glaucoma. J Glaucoma 2009;18:210–2.

Original Article

Diabetic retinopathy management byAustralian optometristsDaniel SW Ting MBBS(Hons),1,2 Jonathon Q Ng MBBS PhD,1 Nigel Morlet FRANZCO,1 Joshua Yuen MBBSMPH,1 Antony Clark MBBS(Hons),1 Hugh R Taylor AC FRANZCO,3 Jill Keeffe OAM PhD4 and David B Preen PhD1

1Eye and Vision Epidemiology Research Group, Centre for Health Services Research, School of Population Health, University ofWestern Australia, Perth, 2Centre of Ophthalmology and Visual Science, Lions Eye Institute, The University of Western Australia, Perth,Western Australia, 3Melbourne School of Indigenous Health, The University of Melbourne, Melbourne and 4Centre for Eye ResearchAustralia, School of Population Health, University of Melbourne, Melbourne, Victoria, Australia

ABSTRACT

Background: To survey the current diabetic retinopa-thy screening and management practices of Austra-lian optometrists following the release of the 1997National Health Medical Research Council DiabeticRetinopathy Management Guidelines.

Design: Cross-sectional national survey, primary caresetting.

Participants: 1000 Australian optometrists across dif-ferent states.

Methods: A self-administered questionnaire was sentto 1000 optometrists across all states during 2007/2008.

Main outcome measures: Use of retinal camera,screening practices/attitudes and behaviour in dia-betic retinopathy management.

Results: 568 optometrists (57%) responded to thesurvey. Patients’ unpreparedness to drive post dila-tion (51%) and the fear of angle closure glaucoma(13%) were the two main barriers to optometristsnot performing dilated ophthalmoscopy. Those whohad strong desire to screen for diabetic retinopathywere more likely to use a retinal camera (p < 0.005).Use of a retinal camera was significantly associatedwith an increased confidence in detecting clinicalsigns of diabetic retinopathy including macularoedema (P < 0.001). Optometrists who read the

guidelines at least once were 2.5-times (P < 0.001)more likely to have confidence in detecting macularoedema than those who had never read theguidelines. Although they may be confident in diag-nosis, and may use retinal cameras for screening,nearly 60% of optometrists would not refer patientswith macular oedema to an ophthalmologist.

Conclusions: Despite their self-reported desire forinvolvement in diabetic retinopathy, the manage-ment of macular oedema by Australian optometristsneeds improvement. The use of retinal cameras andpromotion of the 2008 NHMRC guidelines should beencouraged to improve overall optometric diabeticretinopathy management, particularly with macularoedema.

Key words: diabetes, diabetic retinopathy, screening,survey.

INTRODUCTION

The prevalence of diabetes mellitus is growingrapidly worldwide.1 In 2000, there were approxi-mately 941 000 Australians living with diabetes, andit is estimated that by 2030, this will rise to 1.6million.1 Diabetic retinopathy (DR) occurs in 25% ofpatients with diabetes in Australia.2 As 98% ofvisual impairment secondary to DR can be preventedby timely treatment, early detection is crucial.3

As the primary eye care providers, the optom-etrists and general practitioners play an enormous

� Correspondence: Dr Daniel SW Ting, Ophthalmology Department, Royal Perth Hospital, GPO Box X2213, Perth 6001, Western Australia. Email:

[email protected]

Received 29 June 2010; accepted 12 September 2010.

Clinical and Experimental Ophthalmology 2011; 39: 230–235 doi: 10.1111/j.1442-9071.2010.02446.x

© 2010 The AuthorsClinical and Experimental Ophthalmology © 2010 Royal Australian and New Zealand College of Ophthalmologists

role in DR screening in the community. As part ofroutine DR screening, the National Health andMedical Research Council (NHMRC) guidelines4

recommended that all examiners should assesspatients’ best-corrected visual acuity and performdilated fundus examination at time of diagnosis ofdiabetes. Alternately, the dilated fundus examina-tion may be replaced by retinal photography. Addi-tional information such as HbA1c (glycosylatedhaemoglobin), blood pressure profile, lipid profile,smoking status and other diabetes-related complica-tions may also help in determining the urgency forreferrals.

Australian optometrists have previously beensurveyed in 1999 and 20015,6 following the releaseof the original 1997 NHMRC guidelines on DRmanagement.7 A revised version of these guidelineswas released in late 2008.4 However, to date nopublished studies have examined the long-termimpact of these guidelines on DR screeningand management practices among Australianoptometrists.

Our aim was to identify any changes in DR screen-ing and management practices that have occurredover the last decade following the release of thesenational guidelines. This will provide informationthat will guide the implementation of the revisedguidelines, as well as establishing updated data forfuture evaluation.

METHODS

We conducted a cross-sectional survey of currentlypractising Australian optometrists. A random sampleof 1000 optometrists was selected from the Optom-etrists Association of Australia membership database(4414 members). A self-administered two-pagequestionnaire, an information pamphlet about theobjectives of this study and a postage-paid returnenvelope were mailed to each selected optometrist inNovember 2007. A repeat mail-out of surveys to non-respondents was conducted after 3 months to maxi-mize responses. The University of Western AustraliaHuman Research Ethics Committee approved thisstudy.

The questionnaire used for this study was adaptedfrom two previous surveys conducted by McCartyet al.5,6 to allow temporal comparison regarding DRmanagement practices by Australian optometrists.The survey instrument comprised questions relatingto general professional and practice details, and DRscreening attitudes and practices (e.g. perceived bar-riers and estimated frequency of performing dilatedfundoscopy on diabetic patients, confidence indetecting sight-threatening DR, desire to participatein community screening and perceived need forfurther education on DR).

Optometrists were surveyed about their manage-ment practices using 12 hypothetical clinical scen-arios. The first seven scenarios involved patients ofdifferent ages (7, 18 and 60 years of age), varyingdiabetic treatment (diet alone, oral hypoglycaemicagent or insulin) and who had no DR detected attheir first visit. The last five scenarios focused uponDR management following the detection of microa-neurysms, retinal haemorrhages, new vessels forma-tion and macular oedema. Optometrists were givenfive choices of performing dilated fundoscopy in lessthan 6 months, 1 year, 2 years, 5 years or promptreferral to an ophthalmologist.

All participants remained anonymous throughoutdata collection and analyses. Responses were analy-sed using Stata 10.0 (StataCorp, College Station, TX,USA) with significance set at P < 0.05 for allanalyses. Descriptive statistics were calculated for allcontinuous variables. Relationships between cat-egorical variables were explored using Pearsonc2-tests. Multivariate logistic regression models wereused where appropriate to explore outcomes of inter-est (such as the use of a retinal camera and confi-dence of detecting DR signs and diabetic macularoedema) while controlling for possible confoundingfactors of practice location, place of training andyears of practice.

RESULTS

A total of 568 (57%) optometrists currently practisingin Australia responded to the survey. Demographiccharacteristics of the respondent optometrists areshown in Table 1. Our sample size was 13% of thetotal optometrists practising in Australia (totaloptometrists = 4414), and the percentage of state andurban/rural distribution of the respondent optom-etrists was reasonably comparable to the data pub-lished by the Australian Institute of Health andWelfare (AIHW) in 2006.8 More than 80% reportedreceiving a copy of the 1997 NHMRC guideline, andonly 65% reported having read them at least once.

Of the ophthalmic equipment used by optom-etrists to examine patients with diabetes, the directophthalmoscope was most frequently used (72%),followed by slit-lamp biomicroscopy (65%), binocu-lar indirect ophthalmoscope (56%) and retinalcamera (51%). Almost 15% of optometrists neverperformed direct ophthalmoscopy whereas 55% ofoptometrists used a retinal camera in their practiceson more than half of their diabetic patients.

Table 2 shows the perceived barriers to optom-etrists not performing dilated ophthalmoscopy.Patients’ unpreparedness to drive (51%) and a fear ofprecipitating angle closure glaucoma (13%) postdilation were the two leading reported ‘moderate’ to‘major’ barriers. Only a small number of optometrists

Diabetic retinopathy management 231


reported a lack of confidence in detecting changes(2%) and uncertainty about DR management (1%) asmoderate to major barriers to performing dilatedfundoscopy.

Table 3 summarizes reported current practices,examinations procedures and routine enquiry of riskfactors by optometrists. About 90% of optometristsreported either ‘often’ or ‘almost always’ performingdilated fundoscopy on patients with known diabe-tes, and only 23% would routinely perform dilatedfundoscopy on patients without any history of dia-betes or glaucoma. Two-thirds always questionedabout a positive diagnosis of diabetes in patientsolder than 40 years. As part of routine follow up forpatients with diabetes, 95% of optometrists ‘often’ or‘almost always’ enquired about blood glucose level,as well as factors such as assessment of blood pres-sure, cholesterol level, smoking status and impor-tance of risk factors control to prevent diabetescomplications (Table 3).

Almost all optometrists reported being ‘often’ or‘always’ confident in detecting microaneurysms(93%) and retinal haemorrhages (97%), but fewerwere confident in detecting new vessel formation(85%). More than 50% of the optometrists were‘often’ or ‘always’ unsure in detecting macularoedema.

Table 1. Demographics of the optometrists responding to asurvey on diabetic retinopathy screening

Total number (%)

State or Territory of practice – 566New South Wales 195 (34%)Victoria 160 (28%)Queensland 97 (17%)Western Australia 44 (7%)South Australia 42 (7%)Tasmania 20 (4%)ACT 7 (1%)Northern Territory 1 (<1%)

Number of years of practice0–10 42 (7%)11–20 169 (30%)21–30 260 (46%)>30 97 (17%)

Location of practicesMetropolitan 359 (64.1%)Rural 170 (30.4%)Metropolitan and rural 31 (5.5%)

Location of trainingAustralia 520 (92%)UK 25 (4%)South Africa 9 (2%)New Zealand 8 (1%)USA 2 (<1%)Canada 1 (<1%)

Table 2. Barriers to optometrists performing dilated fundoscopy

Barriers Not a barrier Minor barrier Moderate barrier Strong barrier

Patients’ unpreparedness to drive 68 (12%) 205 (37%) 179 (32%) 107 (19%)Worry of angle closure glaucoma 305 (55%) 185 (33%) 48 (9%) 22 (4%)Time consuming 297 (53%) 195 (35%) 61 (11%) 9 (2%)Lack of dilating drops 539 (97%) 9 (2%) 2 (<1%) 8 (1%)Lack of ophthalmoscopes 545 (98%) 3 (1%) 3 (1%) 6 (1%)Lack of confidence in detecting changes 468 (84%) 76 (14%) 11 (2%) 1 (<1%)Unsure of diabetic retinopathy management 514 (92%) 37 (7%) 6 (1%) 1 (<1%)

Table 3. Current optometrists’ management and attitudes to diabetes and diabetic retinopathy

Screening questions and examinations Almost never Sometimes Half the time Often Almost always

On patients >40 years1 Routine questioning about diagnosis of diabetes 20 (4%) 56 (10%) 21 (4%) 80 (14%) 389 (69%)2 Routine dilated fundoscopy without history of diabetes

or glaucoma102 (18%) 258 (46%) 73 (13%) 56 (10%) 74 (13%)

3 Routine of dilated fundoscopy with history of diabetes 13 (2%) 27 (5%) 22 (4%) 58 (10%) 448 (79%)Frequency of risk factors enquiries

Blood sugar control 2 (<1%) 3 (<1%) 21 (4%) 95 (17%) 446 (79%)Blood pressure control 10 (2%) 15 (3%) 77 (14%) 185 (33%) 279 (49%)Blood cholesterol control 14 (3%) 53 (9%) 121 (21%) 167 (30%) 211 (37%)Smoking status 48 (9%) 116 (21%) 161 (28%) 140 (25%) 102 (18%)Advice regarding complications 11 (2%) 33 (6%) 98 (17%) 221 (39%) 202 (36%)

232 Ting et al.


Most optometrists (95%) used a recall system tofollow up patients with diabetes for examination.However, only half reported that more than 70% oftheir patients would return to see them within thesuggested time frame. When patients were referredto ophthalmologists, the majority (83%) of referringoptometrists reported that more than 70% of theirpatients would see their ophthalmologist within thesuggested time frame. The percentage of optometristsexpressing ‘moderate’ to ‘strong’ desire to screen for,and receive further education regarding DR was 78%and 72%, respectively.

Changes in reported managementsince 19995

The percentage of optometrists performing dilatedfundoscopy on diabetic patients has increased sig-nificantly from 75% in 1999 to 89% in 2009 (P <0.001). In addition, the confidence in detecting sight-threatening DR changes (new vessels elsewhere andmacular oedema) improved significantly from 1999to 2009 (new vessels elsewhere: 75–85%, P < 0.01;macular oedema: 33–47%, P < 0.001). Significantly,more optometrists reported using a recall system in2009 (95%) compared with 1999 (83%). From 1999to 2009, there were no significant changes in thepotential perceived barriers such as fear of inducingangle closure glaucoma post dilation (1999: 17%;2009: 13%), lack of confidence in detecting DRchanges (1999: 4%; 2009: 2%), uncertainty about DRmanagement (1999: 2%; 2009: 1%) and the desire toscreen for DR (1999: 84%; 2009: 78%). In contrast,there was significantly less desire for further educa-

tion about DR diagnosis and management from 1999to 2009 (84% to 72%, P < 0.0001).

Responses to the hypotheticalclinical scenarios (Table 4)

Only 11% of optometrists reported that they wouldperform dilated fundoscopy on a 7-year-old childnewly diagnosed with diabetes in 5 years, and 17%would refer this child to an ophthalmologist.Optometrists would still recommend a 12-monthexamination for diabetic patients with good glycae-mic control with no signs of DR compared with the2-year follow up recommended by the NHMRCguidelines (Table 3). Nonetheless, the overall per-centage of optometrists following the NHMRC-recommended management practices (dilatedfundoscopy within 1 or 2 years of diagnosis, or refer-ral to an ophthalmologist) was greater than observedin 1999. In 2009, the majority of optometristsreported that they would refer patients with severenon-proliferative DR (NPDR) (90%) and prolifera-tive DR (PDR) (98%) to an ophthalmologist.However, only 42% would refer patients withmacular oedema to an ophthalmologist for promptinvestigation and treatment.

From multivariate logistic regression analyses,optometrists who had a strong desire to screen forDR were almost twice as likely to ‘often’ or‘almost always’ use a retinal camera to examinepatients with diabetes after controlling for previoustraining location, duration and location ofcurrent practice (OR = 1.98, 95% CI = 1.27–3.10,P < 0.005).

Table 4. Optometrists’ management of hypothetical clinical scenarios and specific signs of DR

Case scenarios Referral toophthalmologists

Review in5 years

Review in2 years

Review in1 year

Review in<6 months

If no signs of DR at baseline examination7-yo – newly diagnosed diabetic 95 (17%) 63 (11%)† 125 (22%) 183 (33%) 96 (17%)18-yo – newly diagnosed with DM 40 (7%) 16 (3%) 141 (25%)† 255 (45%) 113 (20%)60-yo with good HbA1c control – diet alone 5 (1%) 1 (<1%) 211 (37%)† 290 (51%) 58 (10%)60-yo, 10-year diabetes, commenced on OHA 9 (2%) 0 (0%) 86 (15%)† 374 (66%) 98 (17%)60-yo, 10-year diabetes with good HbA1c on OHA 10 (2%) 1 (<1%) 72 (13%)† 417 (74%) 65 (12%)60-yo, 10-year diabetes with good HbA1c on insulin 21 (4%) 0 (0%) 31 (6%)† 409 (72%) 105 (19%)60-yo, poorly controlled HbA1c despite insulin 79 (14%) 0 (0%) 2 (<1%) 127 (22%)† 360 (63%)

DR signsOccasional microaneurysms on peripheries 43 (8%) 14 (3%) 297 (53%) 167 (30%)† 44 (8%)Macular oedema (not clinically significant) 234 (41%)† 6 (1%) 43 (8%) 138 (25%) 139 (25%)Peripheral microaneurysms and retinal haemorrhages 224 (40%) 0 (0%) 81 (14%) 169 (30%) 93 (16%)†

Extensive microaneurysms, retinal haemorrhagesand occasional cotton wool spots all located peripherally

509 (90%)† 0 (0%) 4 (1%) 18 (3%) 36 (6%)

New vessel formation 557 (98%)† 0 (0%) 0 (0%) 1 (<1%) 10 (2%)

†Recommended time frame suggested by National Health and Medical Research Council guidelines in 1997. DM, diabetes mellitus;DR, diabetic retinopathy; HbA1c, glycosylated haemoglobin; OHA, oral hypoglycaemic agents; yo, year-old.



When controlled for reading the guidelines, pre-vious training location, duration and location ofpractice, optometrists who ‘often’ and ‘always’ useda retinal camera were more confident in detectingretinal diabetic changes such as microaneurysms(OR = 5.29, 95% CI = 2.22–12.40, P < 0.001), newvessels formation elsewhere (OR = 4.62, 95% CI =2.56–8.34, P < 0.001) and macular oedema (OR =2.49, 95% CI = 1.51–4.12, P < 0.001).

Reading the guidelines at least once was alsoassociated with increased confidence in detectingmacular oedema (OR = 2.49, 95% CI = 1.51–4.12,P < 0.001). However reading the guidelines did notimprove referrals for patients with macular oedema(OR = 1.04, 95% CI = 0.65–1.65, P = 0.88). Likewiseconfidence in detecting macular oedema after con-trolling for other factors such as previous traininglocation, duration and location of practice and use ofa retinal camera was not associated with optometricreferrals of patients with macular oedema to anophthalmologist (OR = 0.99, 95% CI = 0.71–1.39,P = 0.97).

DISCUSSION

Our results indicate that DR management practicesof Australian optometrists have improved since therelease of NHMRC guidelines in 1997.7 Comparedwith the two national surveys conducted in 1999 and2001,5,6 more optometrists now perform dilated fun-doscopy on diabetic patients, use recall notices andhad greater confidence in detecting and managingDR changes in their patients. Additionally, nearly80% of optometrists had ‘moderate’ to ‘strong’ desireto screen for DR. This could significantly reduce thewaiting period of diabetic patients to see an ophthal-mologist in the public setting and may triage theurgency of ophthalmic review by severity, especiallyfor patients with sight-threatening DR. We alsofound an approximate 10% rise in the use of recallnotices over the last decade. This helps to ensureregular eye screening of patients, as a patientreminder system was reported to be an effective wayof enforcing and increasing the patients’ adherenceto clinical guidelines.9

Both the 1997 and 2008 NHMRC guidelines6,7 rec-ommend that screening of diabetic children shouldstart at the time of puberty, with the screening inter-val determined by the clinical findings. Those withmoderate to severe NPDR, PDR or macular oedemawarrant prompt referrals to an ophthalmologist.We found that optometrists generally reviewed dia-betic patients with no signs of DR more frequentlythan recommended.7 The optometrists should beencouraged to read the guidelines more frequentlyin order to review the patients with diabetes in an

appropriate time frame to reduce their unnecessaryfinancial burden.

The responses we obtained regarding the manage-ment of diabetic macular oedema by Australianoptometrists were of concern. More than 50% ofoptometrists reported that they lacked confidence indetecting macular oedema and only 40% wouldrefer patients with macular oedema to anophthalmologist.

Although both the use of a retinal camera andreading the guidelines were significantly associatedwith increased confidence in detecting macularoedema, they were not associated with appropriatereferral of patients with macular oedema to anophthalmologist. Confidence in detecting macularoedema was also not associated with referrals to anophthalmologist.

In other words, the optometrists were not likelyto refer patients with macular oedema to an oph-thalmologist despite having read the guidelines,being confident in detecting macular oedema andusing a retinal camera. Given macular oedema is amajor cause of significant visual impairment,optometrists need to improve their management(confidence to detect and referrals) of this conditionto ensure prompt laser treatment for patients withdiabetic maculopathy. Although early stages ofmacular oedema may be difficult to detect with-out indirect ophthalmoscopy, any reduction invisual acuity should raise suspicion and prompt areferral.

Concern about use of dilating drops inducingangle closure glaucoma seems unwarranted as it is arare event (1 in 20 000).10 Promoting the use of non-mydriatic fundus cameras may help. We found therewas a strong association between the frequency ofretinal camera use and the desire to screen, as well asconfidence in detecting DR changes. Others havefound non-mydriatic retinal camera fundus photog-raphy yielded a reasonable sensitivity (95%) andspecificity (99%).11

The present study provides an overview of DRmanagement by Australian optometrists, which hasimproved over the last decade following the releaseof 1997 NHMRC guideline.7 Given that macularoedema causes significant visual impairment inpatients with diabetes, further education about thedetection and referral of subjects with macularoedema is important. The use of retinal camerasand promotion of the new 2008 NHMRC guide-lines7 should be encouraged to improve the overalloptometric DR management and reduce the inci-dence of this preventable blinding disease in thefuture. A further survey may help assess the impactof the new 2008 NHMRC guidelines, especially themanagement of diabetic macular oedema by Aus-tralian optometrists.

234 Ting et al.


REFERENCES

1. Wild S, Roglic G, Green A, Sicree R, King H. Globalprevalence of diabetes: estimates for the year 2000and projections for 2030. Diabetes Care 2004; 27: 1047–53.

2. Tapp RJ, Shaw JE, Harper CA et al. The prevalence ofand factors associated with diabetic retinopathy in theAustralian population. Diabetes Care 2003; 26: 1731–7.

3. Ferris FL 3rd. How effective are treatments for diabeticretinopathy? JAMA 1993; 269: 1290–1.

4. National Health and Medical Research Council. Guide-lines for the Management of Diabetic Retinopathy[Internet]. 2008. Available from: http://www.nhmrc.gov.au/publications/synopses/di15syn.htm

5. McCarty CA, McKay R, Keeffe JE. Management ofdiabetic retinopathy by Australian optometrists.Working Group on Evaluation of NHMRC RetinopathyGuideline Distribution. National Health and MedicalResearch Council. Aust N Z J Ophthalmol 1999; 27:404–9.

6. McCarty CA, Taylor KI, McKay R, Keeffe JE. Diabeticretinopathy: effects of national guidelines on the

referral, examination and treatment practices of oph-thalmologists and optometrists. Clin Experiment Ophthal-mol 2001; 29: 52–8.

7. National Health and Medical Research Council. Clini-cal practice guideline: management diabetic retino-pathy [Internet]. 1997. Available from: http://www.nhmrc.gov.au/publications/synopses/cp53covr.htm

8. Australian Institute of Health and Welfare. [Internet].2006. Available from: http://www.aihw.gov.au/publications/phe/phe-116-10722/phe-116-10722.pdf

9. Grimshaw JM, Russell IT. Achieving health gainthrough clinical guidelines II: ensuring guidelineschange medical practice. Qual Health Care 1994; 3:45–52.

10. Talks SJ, Tsaloumas M, Mission GP, Gibson JM.Angle closure glaucoma and diagnostic mydriasis.Lancet 1993; 342: 1493–4.

11. Baeza M, Orozco-Beltran D, Gil-Guillen VF et al.Screening for sight threatening diabetic retinopathyusing non-mydriatic retinal camera in a primary caresetting: to dilate or not to dilate? Int J Clin Pract 2009;63: 433–8.



RESEARCH

AUSTRALIAN FAMILY PHYSICIAN VOL. 40, NO. 4, APRIL 2011 233

Daniel TingJonathon NgNigel MorletJoshua YuenAntony ClarkHugh TaylorJill KeefeDavid Preen

Diabetic retinopathyScreening and management by Australian GPs

AimTo describe current diabetic retinopathy (DR) screening and management practices among Australian general practitioners.

MethodA self administered questionnaire on DR management was mailed to 2000 rural and urban GPs across Australia in 2007–2008.

ResultsOnly 29% of the GP respondents had read the National Health and Research Council guidelines at least once and 41% had a ‘moderate’ to ‘strong’ desire to screen for DR. A majority of GPs (74%) reported not routinely examining their diabetic patients for DR. Lack of confidence in detecting DR changes (86.4%) and time constraints (73.4%) were the two major barriers to GPs performing dilated fundoscopy on diabetic patients.

DiscussionGiven that access to optometry is not evenly distributed across the country, and that ophthalmology is under-resourced, GPs are the healthcare providers most able to manage and screen for DR in the community.

Keywords: diabetes mellitus; diabetic retinopathy; mass screening; secondary prevention; general practice

Diabetes mellitus is rising in prevalence within Australia and internationally, with estimates indicating that the global prevalence of diabetes will double by 2030.1 In Australia, the prevalence of diabetes is 8% in adult men and 6.5% in adult women;2 of these, one in four will be diagnosed with diabetic retinopathy (DR).3 Early detection and prompt treatment can prevent 98% of visual impairment.4

Primary healthcare providers such as general practitioners and optometrists are at the ‘front line’ of service provision and play a crucial role in screening for DR in the community. However, a 1994 survey of Victorian GPs found over half had little interest in DR screening and that most routinely examined less than half of their patients with diabetes for DR.5 The National Health and Medical Research Council (NHMRC) released clinical practice guidelines for DR in 1997,6 outlining evidence based DR management practices and encouraging physicians to increase the DR screening rate in order to reduce DR related visual impairment. Nonetheless, a subsequent Victorian survey reported that despite the NHMRC guidelines, 48% of GPs still had little or no desire to screen for DR.7

This study is the first national survey of Australian GPs on DR management. The purpose is to investigate current Australian GPs’ management practices and attitudes towards DR and its management since the release of the 1997 NHMRC guidelines. Given the survey was started in late 2007, it can also serve as a baseline before the release of the most recent NHMRC guidelines in 2008.8

MethodA random sample of 2000 currently practising Australian GPs selected from The Royal Australian College of General Practitioners’ (RACGP) membership database were surveyed. A package consisting of a self administered two page questionnaire, an information leaflet detailing the aim of the study and a reply paid return envelope was posted to selected GPs in December 2007. A repeat mailout of surveys to nonresponders was conducted after 3 months to maximise response.

This study was approved by the University of Western Australia’s Human Research Ethics Committee.

The survey questions related to:

(eg. location of previous training, duration at and location of practices, rural or metropolitan practice)

pressure, cholesterol, smoking status

factors, patient refusal, fear of angle closure glaucoma, lack of confidence in detecting and managing DR and lack of dilating drops and ophthalmoscopes in practices)

optometrists and ophthalmologists for DR screening

visual acuity, dilated or nondilated fundoscopy)

The participants were also asked to respond to five case scenarios regarding the management of DR clinical signs: microaneurysms, retinal haemorrhages, cottonwool spots, new vessel formation and presence of hard exudates near

RESEARCH

234 AUSTRALIAN FAMILY PHYSICIAN VOL. 40, NO. 4, APRIL 2011

Diabetic retinopathy – screening and management by Australian GPs

to an ophthalmologist or optometrist if visual symptoms were present. Nearly 80% of GPs felt their patients would see an ophthalmologist should it be necessary.

Table 4 shows responses to the management of specific DR signs and hypothetical case scenarios. Most GPs (63%) would refer their patients with occasional microaneurysms with normal vision within 1 month to see an ophthalmologist while only 3% were confident to not refer these patients to the ophthalmologist. For patients with hard exudates near the macula and normal vision, 87% of GPs indicated that they would refer to an ophthalmologist within 1 month. Following the detection of peripheral microaneurysms and retinal haemorrhages, 95% of GPs would refer their patients to see an ophthalmologist within the recommended

DR changes (86%) and time constraints (73%) were the primary barriers to performing dilated fundoscopy on diabetic patients for GPs. Additional reported barriers included patient refusal, concern of angle closure glaucoma, lack of dilating drops and uncertainty surrounding DR management (Table 3).

Less than half of GPs (41%) expressed ‘moderate’ to ‘strong’ desire to screen for DR in the community setting. Nearly all GPs (91%) referred diabetic patients to ophthalmologists every 1–2 years, while 68% referred their diabetic patients to optometrists in the first instance. One-fifth of GPs never referred any diabetic patients to see an optometrist, these GPs all preferred to refer their patients to see an ophthalmologist every 1–2 years. A small number of GPs (8%) would only refer their patients

the macula. A further seven hypothetical case scenarios involving patients that varied in age (7, 18 and 60 years of age), diabetic management (diet control, oral hypoglycaemic agents and insulin) and glycaemic control (poorly and well controlled) who had no signs of DR detected at baseline examination were also included.

Analyses were performed using SPSS version 17 (SPSS, Chicago, IL, USA). Descriptive statistics (mean, standard deviation) were calculated for continuous variables. Relationships between categorical variables were explored using chi-square tests. Additionally, a multivariate logistic regression model was used to study the possible factors relating to GPs’ confidence in detecting DR clinical signs, such as their years in practice, previous training location and whether or not they had read the NHMRC guidelines.

ResultsThere were 429 (21%) respondents to the survey (Table 1). Almost half reported having received the 1997 NHMRC guidelines for DR management, however, of these only 29% had read the guidelines at least once. Apart from the NHMRC guidelines, GPs also reported using other resources such as RACGP guidelines,9 Therapeutic guidelines: endocrinology,10 National Prescribing Service Guidelines,11 American Diabetic Association Guidelines,12 various online websites, and diabetes focused peer reviewed journals.

Almost all GPs reviewed their diabetic patients’ blood pressure (98.6%) and HbA1c (92.1%) at least every 6 months (Table 2). Assessment of lipid profile, smoking status and advice on diabetes complications were conducted less frequently by the respondent GPs (Table 2). Nearly 75% of GPs did not routinely examine their diabetic patients for DR; of those, 89% would refer their diabetic patients to see an ophthalmologist within 2 years of initial diabetes diagnosis. More GPs ‘usually’ and ‘always’ performed nondilated (61%) than dilated fundoscopy (13%) to detect DR signs and only 65% of GPs ‘usually’ or ‘always’ checked visual acuity.

Only 21% of GPs responded that they were ‘often’ or ‘almost always’ confident in detecting DR changes. Lack of confidence in detecting

Table 1. Demographics of GP respondents

Demographic information Total (%) Australian GPs (%)20

State or territory of practice

New South Wales 101 (24%) 33.4

Victoria 122 (29%) 25.0

Queensland 74 (18%) 19.1

Western Australia 68 (16%) 9.2

South Australia 40 (9%) 8.5

Tasmania 11 (3%) 2.7

Australian Capital Territory 5 (1%) 1.5

Northern Territory 1 (<1%) 0.7

Years of practice

0–10 47 (11%) No comparison available11–20 61 (14%)

21–30 136 (32%)

>30 185 (43%)

Locality of practice

Metropolitan 279 (66%) 73.8

Rural 143 (34%) 26.2


Australia 343 (81%) 68.4

United Kingdom 39 (9%) –

India 12 (3%) –

New Zealand 6 (1%) –

Other 26 (6%) Overseas 31.6

Note: Not all respondents answered all questions, so not all numbers total to the 429 respondents

RESEARCH



DiscussionIn response to the World Health Assembly resolution on the elimination of avoidable blindness, the National Eye Health Framework was developed following the Australian Health Ministers’ Conference in 2005.13 It identified five potential key areas which may help prevent avoidable blindness and low vision. Two of these related to increasing early detection and improving access to eye healthcare services.13 At the time of the release of the NHMRC guidelines on DR management in 1997,6 the Victorian GP DR survey showed that half of GPs expressed a desire to regularly screen for DR in patients attending their practice.7 Unfortunately, the authors found that since the last survey, even fewer GPs (41%) expressed a desire to screen

General practitioners’ confidence in detecting DR changes was strongly associated with:

(χ2=7.48, p<0.01)

years (χ2=7.71, p<0.01)χ2=3.88,

p<0.05). When controlled for years of practice and previous training location, GPs who had read the guidelines at least once were 2.11 times more likely to report confidence in screening for DR (OR=2.11, SE=0.54, 95% CI: 1.27–3.50, p<0.005). The location of the GP practice (rural or metropolitan) was not associated with reported confidence in detecting DR changes, desire of DR screening or the frequency of referral to ophthalmologists or optometrists.

timeframe. For new vessel formation, nearly all GPs would refer within 1 month to an ophthalmologist.

A majority of GPs (82%) would (inappropriately) refer a child, 7 years of age, with diabetes and no signs of DR for regular eye screening while 18% would refer such patients in 5 years. As shown in Table 3, the majority of the GPs would refer patients of various ages (7, 18 and 60 years); diabetic management (diet control, oral hypoglycaemic and insulin); and glycaemic control (well and poorly controlled) elsewhere for regular eye screening even without any signs of DR rather than perform the review themselves.

General practitioners’ desire to screen for DR in the community was strongly associated with having read the 1997 NHMRC guidelines at least once (χ2=17.64, p<0.001) and with reporting confidence in detecting DR clinical changes (χ2=28.5, p<0.001). In multivariate logistic regression analyses, GPs who reported confidence in detecting DR clinical signs were 3.31 times more likely to have a desire to screen for DR – after controlling for reading the guidelines at least once, years of practice and previous training location (OR=3.31, SE=0.85, 95% CI: 2.00–5.47, p<0.001). However, the frequency of GPs performing visual acuity measurement and dilated fundoscopy as part of the routine eye examination for patients with diabetes was not associated with reading the NHMRC guidelines.

Table 2. Current GP management and attitudes to diabetes and diabetic retinopathy

At diabetes follow up Never and rarely (%) Sometimes (%) Usually and always (%)

Visual acuity measurement 30 (13.5%) 49 (22%) 144 (64.5%)

Fundsocopy (undilated) 54 (25.1%) 29 (13.5%) 132 (61.4%)

Dilated fundoscopy 155 (79.0%) 16 (8.2%) 25 (12.8%)

Diabetic patients examined for diabetic retinopathy None197 (46.4%)

Some119 (28%)

All109 (25.6%)

Frequency with which diabetic patients were reviewed

Yearly or less137 (57.3%)

More than yearly102 (42.7%)

–

Frequency of risk factor management Six monthly or less More than 6 monthly –

HbA1c 387 (92.1%) 33 (7.9%) –

Blood pressure 416 (98.6%) 6 (1.4%) –

Cholesterol 245 (58.2%) 176 (41.8%) –

Smoking 278 (65.9%) 144 (34.1%) –

Advice regarding complications 334 (79.7%) 85 (20.3%) –


Table 3. Barriers to GPs performing dilated fundoscopy

Barrier No barrier or minor barrier (%)

Moderate or major barrier (%)

No confidence in detecting diabetic retinopathy signs

55 (13.6%) 349 (86.4%)

Time consuming 109 (26.7%) 300 (73.3%)

Patients’ refusal to dilation 145 (36.5%) 252 (63.5%)

Worry of inducing angle closure glaucoma 223 (55.8%) 177 (44.2%)

Lack of dilating drops 251 (62.4%) 151 (37.6%)

Unsure of diabetic retinopathy management 328 (82.2%) 71 (17.8%)

Lack of ophthalmoscopes 396 (97.8%) 9 (2.2%)


RESEARCH



GPs.18 Nearly 60% of optometrists would not refer patients with diabetic maculopathy to see an ophthalmologist and 10% would not refer patients with severe nonproliferative DR. Comparing these two surveys suggests that GPs may be the most proficient healthcare providers to manage and screen for DR in the community.

From the 2009 study on optometrists,18 the retinal cameras were shown to have increased optometrists’ confidence to detect DR changes such as microaneurysms, retinal haemorrhages, new vessel formation and macular oedema. It is unknown whether GPs would feel more confident with retinal photographic screening, however, a pilot study of photographic screening by GPs for DR found that they would be willing to expand their roles into DR screening if such infrastructure were readily accessible.19 The study19 also showed that GPs had good diagnostic accuracy (sensitivity 87%; specificity 95%) for DR. Given that this was a pilot study with a relatively small sample size, a larger study using cheaper, portable retinal cameras

increase the DR screening rate in the Australian community.

In this study, a great proportion of GPs (92%) reported that they review their patients’ HbA1c every 6 months or less. Nonetheless, numerous previous data based on Medicare data analysis showed that only 25–80% of GPs measured their patients’ HbA1c level at a 6 monthly interval.15 This highlighted a gap between what GPs think they should do as opposed to what is done in reality on a routine basis. (However, those who completed the survey may have been more motivated to measure HbA1c than others.) Given that good HbA1c control significantly reduces the risk of microvascular and macrovascular complications,16–17 it is important to translate theoretical knowledge on diabetes management into routine clinical practices by checking their patients’ HbA1c every 6 months as per NHMRC guidelines.8

In a similar survey of optometrists, the authors found that the optometric DR management was generally inferior to that of

for DR. This is of concern given that primary healthcare screening provides an excellent opportunity to differentiate those patients who require specialist ophthalmic care from those who can continue to be managed by their GP.

We also found that similar to the previous Victorian study,7 GPs’ perceived the lack of confidence in detecting clinical DR signs was the leading barrier to performing dilated fundoscopy. A fear of inducing angle closure glaucoma (a rare 1:20 000 event postdilation)14 was perceived as another major barrier to performing dilated fundoscopy. Despite the low numbers of GPs who reported confidence in detecting DR clinical signs (21%), based on the hypothetical clinical scenarios in the survey most GPs were generally confident and proficient to manage DR once DR changes were detected. Given that the GPs who reported confidence in detecting DR clinical signs were three times more likely to have strong desire to screen for DR, more education needs to be directed toward detection of DR clinical signs to

Table 4. GP management of hypothetical clinical scenarios and specific signs of diabetic retinopathy

Clinical scenario Appropriate referral (%)

Inappropriate referral (%)

Recommended referral time frame (2008)*

Recommended referral time frame (1997)**

Occasional microaneurysms with normal vision 101 (24.7%) 308 (75.3%) 1 year Nonurgent, routine referral

Hard exudates near macula with normal vision 348 (86.6%) 54 (13.4%) 1 month or less Refer urgently

Peripheral microaneurysms and retinal haemorrhages 384 (94.8%) 21 (5.2%) 6 months or less Refer urgently

Extensive microaneurysms, retinal haemorrhages and cottonwool spots (all peripherally)

407 (99.5%) 2 (0.5%) 3 months or less Refer urgently

New vessel formation 407 (99.8%) 1 (0.2%) 1 month or less Refer urgently

If no signs of DR at baseline examination:

7 year old – newly diagnosed diabetic 338 (80.3%) 83 (19.7%) At puberty Refer in 5 years

18 year old – newly diagnosed diabetic† 351 (84.0%) 67 (16.0%) 1 to 2 years Yearly, no later than 2 yearly

60 year old with good HbA1c control – diet# 284 (67.1%) 139 (32.9%) 2 years Yearly, no later than 2 yearly

60 year old, 10 years diabetes, commenced on OHA† 361 (84.9%) 64 (15.1%) 1 to 2 years Yearly, no later than 2 yearly

60 year old, 10 years diabetes with good HbA1c on OHA#

289 (68.5%) 133 (31.5%) 2 years Yearly, no later than 2 yearly

60 year old, 10 years diabetes with good HbA1c on insulin#

269 (63.4%) 155 (36.6%) 2 years Yearly, no later than 2 yearly

60 year old, poorly controlled HbA1c despite insulin 306 (72.0%) 119 (28.0%) 1 year Yearly, no later than 2 yearly

* Referral time frame recommended by 2008 NHMRC guidelines8 ** Referral time frame recommended by 1997 NHMRC guidelines6

† Should HbA1c is unavailable, all diabetic patients should be referred within 1–2 years# Patients with good HbA1c control and no signs of DR are recommended to undergo 2 yearly fundus examination for DR OHA = oral hypoglycaemic agents; HbA1c = glycosylated haemoglobinNote: Not all respondents answered all questions, so not all numbers total to the 429 respondents

RESEARCH

238 AUSTRALIAN FAMILY PHYSICIAN VOL. 40, NO. 4, APRIL 2011


7. McCarty CA, Taylor KI, Keeffe JE. Management of diabetic retinopathy by general practitioners in Victoria. Clin Exp Ophthal 2001;29:12–6.

8. National Health and Medical Research Council. Guidelines for the management of diabetic retin-opathy. Canberra: Commonwealth of Australia, 2008. Available at www.nhmrc.gov.au/publications/synopses/di15syn.htm [Accessed 17 September 2010].

9. Royal Australian College of General Practitioners. Management of diabetes mellitus 2010/2011. 16th edn. Canberra: Diabetes Australia, 2010. Available at www.racgp.org.au/guidelines/diabetes [Accessed 17 September 2010].

10. Endocrinology Expert Group. Therapeutic guide-lines: endocrinology. Diabetes: diagnosis and principles of management. 3rd edn. Therapeutic Guidelines Limited: North Melbourne, 2004.

11. National Prescribing Service Limited. Diabetes drugs. Available at www.nps.org.au/health_pro-fessionals/drug_and_therapeutic_topics/diabetes_drugs. [Accessed 3 December 2010].

12. American Diabetes Association. Management of diabetes mellitus. Available at http://care.diabetes-journals.org/content/33/Supplement_1 [Accessed 20 September 2010].

13. Commonwealth of Australia. Eye health in Australia – a background paper to the national framework for action to promote eye health and prevent avoidable blindness and vision loss. Canberra: Commonwealth of Australia, 2005.

14. Talks SJ, Tsaloumas M, Mission GP, et al. Angle closure glaucoma and diagnostic mydriasis. Lancet 1993;342:1493–4.

15. National Health and Medical Research Council. Evidence-Practice Gaps Report Volume 1: a review of developments 2004–2007. Available at www.nhmrc.gov.au/_files_nhmrc/file/nics/material_resources/epgr_review_chapter_6.pdf. [Accessed 20 September 2010].

16. Selvin E, Marinopoulos S, Berkenblit G, et al. Meta-analysis: glycosylated hemoglobin and car-diovascular disease in diabetes mellitus. Ann Intern Med 2004;141:421–31.

17. Stettler C, Allemann S, Juni P, et al. Glycemic control and macrovascular disease in types 1 and 2 diabetes mellitus: meta-analysis of randomized trials. Am Heart J 2006;152:27–38.

18. Ting DSW, Yuen J, Clark A, et al. Australian Optometrists National Survey of Diabetic Retinopathy Management. Poster session pre-sented at Annual General Meeting and Scientific Congress, 14–18 November 2009, Brisbane, Australia.

19. Askew D, Schluter PJ, Spurling G, et al. Diabetic retinopathy screening in general practice: a pilot study. Aust Fam Physician 2009;38:650–6.

20. Britt H, Miller GC, Charles J, et al. General practice activity in Australia 2008–09. Canberra: Australian Institute of Health and Welfare, 2009. Available at www.aihw.gov.au/publications/gep/gep-25-11013/gep-25-11013-c03.pdf [Accessed 20 September 2010].

AuthorsDaniel Ting MBBS, is Master of Medical Science candidate, Eye and Vision Epidemiology Research Group, Centre for Health Services Research, University of Western Australia, and Centre of Ophthalmology and Visual Science, Lions Eye Institute, University of Western Australia, Lions Eye Institute, Perth, Western Australia. [email protected] Ng MBBS, PhD, is ophthalmology trainee, Eye and Vision Epidemiology Research Group, Centre for Health Services Research, University of Western AustraliaNigel Morlet FRANZCO, is consultant ophthalmolo-gist, Eye and Vision Epidemiology Research Group, Centre for Health Services Research, University of Western AustraliaJoshua Yuen MBBS, MPH, is ophthalmology trainee, Eye and Vision Epidemiology Research Group, Centre for Health Services Research, University of Western AustraliaAntony Clark MBBS(Hons), is ophthalmology trainee, Eye and Vision Epidemiology Research Group, Centre for Health Services Research, University of Western AustraliaHugh Taylor AC, FRANZCO, is Harold Mitchell Chair of Indigenous Eye Health, School of Indigenous Health, University of Melbourne, VictoriaJill Keefe OAM, PhD, is Head, Centre of Eye Research Australia, School of Population Health, University of Melbourne, VictoriaDavid Preen PhD, is Director, Centre for Health Services Research, University of Western Australia.

Conflict of interest: none declared.

References1. Wild S, Roglic G, Green A, et al. Global prevalence

of diabetes: estimates for the year 2000 and projec-tions for 2030. Diabetes Care 2004;27:1047–53.

2. Dunstan DW, Zimmet PZ, Welborn TA, et al. The rising prevalence of diabetes and impaired glucose tolerance: the Australian Diabetes, Obesity and Lifestyle Study. Diabetes Care 2002;25:829–34.

3. Tapp RJ, Shaw JE, Harper CA, et al. The prevalence of and factors associated with diabetic retinopa-thy in the Australian population. Diabetes Care 2003;26:1731–7.

4. Ferris FL, 3rd. How effective are treatments for dia-betic retinopathy? JAMA 1993;269:1290–1.

5. Dickson PR, McCarty CA, Keeffe JE, et al. Diabetic retinopathy: examination practices and referral patterns of general practitioners. Med J Aust 1996;164:341–4.

6. National Health and Medical Research Council. Clinical practice guidelines: management of diabetic retinopathy. Canberra: Commonwealth of Australia, 1997. Available at www.nhmrc.gov.au/publications/synopses/cp53covr.htm [Accessed 17 September 2010].

would provide more insight into the enthusiasm for, and effectiveness of, retinal photographic screening in primary care.

The updated 2008 NHMRC guidelines suggest that mydriatic retinal photography is the most effective DR screening tool with a sensitivity of at least 80%.8 Cheaper retinal cameras are now available and should be encouraged in primary healthcare, especially for large practices. Retinal photographs taken by staff could be instantly interpreted by the GP to determine the need for referrals to ophthalmologists during general diabetes consultations.

The strength of this study is that it is the first nationwide GP DR survey. The study findings not only revealed the current management practices of Australian GPs on DR management but can be utilised as baseline findings for a future study to assess the impact of the 2008 NHMRC guidelines.8 Additionally, the GPs’ demographics from this study, such as the distribution of respondent GPs and locality of practices, were comparable to the national demographics for GPs (age and gender were not enquired about in this survey).20 On the other hand, the study findings were limited by several factors. Of the GPs practising in Australia (n=12 938), this study had a relatively small sample size (3.3%, n=429) which may affect the generalisability of results.20 In addition, the study did not gather data regarding current use of retinal cameras among GPs, their previous education on eye training and the size of their practice, which could affect GPs’ confidence in screening for DR.

ConclusionIn conclusion, Australian GPs in general reported sound management of sight threatening DR. Given that the access to optometry is not evenly distributed across Australia and that ophthalmology is under-resourced, GPs may be the most proficient healthcare providers to manage and screen for DR in the community. In the absence of retinal photography, dilated ophthalmoscopy is still the most convenient and effective method to examine for DR and therefore, it is a basic clinical examination skill that should be encouraged for all GPs.

Original Article

Light and portable novel device for diabeticretinopathy screeningDaniel SW Ting MBBS(Hons),1,2,3 Mei Ling Tay-Kearney FRANZCO1,3 and Yogesan Kanagasingam PhD2

1Center for Ophthalmology and Visual Sciences, Lions Eye Institute, University of Western Australia, Nedlands,2The Australian E-HealthResearch Centre, Commonwealth Scientific Industrial Research Organization (CSIRO), Floreat, and 3Ophthalmology Department, RoyalPerth Hospital Wellington Street Campus, Perth, Western Australia, Australia

ABSTRACT

Background: To validate the use of an economicalportable multipurpose ophthalmic imaging device,EyeScan (Ophthalmic Imaging System, Sacramento,CA, USA), for diabetic retinopathy screening.

Design: Evaluation of a diagnostic device.

Participants: One hundred thirty-six (272 eyes) wererecruited from diabetic retinopathy screening clinicof Royal Perth Hospital, Western Australia, Australia.

Methods: All patients underwent three-field (opticdisc, macular and temporal view) mydriatic retinaldigital still photography captured by EyeScan andFF450 plus (Carl Zeiss Meditec, North America) andwere subsequently examined by a senior consultantophthalmologist using the slit-lamp biomicroscopy(reference standard). All retinal images were inter-preted by a consultant ophthalmologist and amedical officer.

Main Outcome Measures: The sensitivity, specificityand kappa statistics of EyeScan and FF450 plus withreference to the slit-lamp examination findings by asenior consultant ophthalmologist.

Results: For detection of any grade of diabetic retin-opathy, EyeScan had a sensitivity and specificity of93 and 98%, respectively (ophthalmologist), and 92and 95%, respectively (medical officer). In contrast,

FF450 plus images had a sensitivity and specificity of95 and 99%, respectively (ophthalmologist), and 92and 96%, respectively (medical officer). The overallkappa statistics for diabetic retinopathy grading forEyeScan and FF450 plus were 0.93 and 0.95 forophthalmologist and 0.88 and 0.90 for medicalofficer, respectively.

Conclusions: Given that the EyeScan requiresminimal training to use and has excellent diagnosticaccuracy in screening for diabetic retinopathy, itcould be potentially utilized by the primary eye careproviders to widely screen for diabetic retinopathy inthe community.

Key words: diabetes, diabetic retinopathy, diagnostictechnique.

INTRODUCTION

The worldwide prevalence of diabetes is estimatedto double to 366 million (4.8%) by 2030.1 Peoplewith diabetes should receive regular eye screen-ing, as early detection could prevent diabeticretinopathy-related visual impairment.2–4 Retinalstill photography remains the mainstay screeningtool in various diabetic retinopathy screening pro-grams worldwide, and it has been shown to increasethe primary eye care providers’ confidence and desireto detect diabetic retinopathy in the community.5 Todate, mydriatic retinal photography has been shown

� Correspondence: Correspondence: Dr Daniel SW Ting, Block 10C #10-10, Braddell View, Singapore, 579719. Email: [email protected]

Received 2 June 2011; accepted 12 September 2011.

Competing/conflict of interest: Professor Yogesan Kanagasingam invented the EyeScan (Ophthalmic Imaging System, CA, USA) but he is

not/has not received any direct/indirect funding from the manufacturer.

Funding sources: Diabetes Australia Research Trust and Royal Perth Hospital have provided research funding to this project. The sponsor or

funding organization had no role in the design or conduct of this research.

Clinical and Experimental Ophthalmology 2012; 40: e40–e46 doi: 10.1111/j.1442-9071.2011.02732.x


to be the most effective means to detect diabeticretinopathy in a screening setting.6,7

The rising prevalence of diabetes will requireimplementation of more community screening pro-grams. This will lead to a rise in health-care costs8

and maintenance of an effective quality assurancesystem.9 The cost of a retinal camera is and willcontinue to be an important factor in the health-care cost equation. Numerous retinal cameras(mydriatic and non-mydriatic) with various func-tions, including anterior and posterior segmentimaging and fluorescein angiography, are currentlyavailable on the market. However, many are notinexpensive.

The purpose of our study was to validate the effi-cacy of the EyeScan (Ophthalmic Imaging System,CA, USA) machine to screen for diabetic retinopathyin the community. The EyeScan is a flash camera thatweighs around 0.9 kg. Apart from retinal imaging,this device also captures the anterior segments of theeye, such as the cornea and the lens, and this isespecially important in determining the causes ofsome ‘ungradable’ retinal still images secondary tocataracts or pterygium. It has a maximum imagecapture rate of three images per second and a 5.3-megapixel Sensor. It is able to capture the photoswith near infrared or visible light and has a field ofview of 35 degrees. This machine was approved bythe US Food and Drug Administration in 2010 and iscurrently available in United States, Australia andother European countries. Given that this is anextremely portable device, which can be carriedaround in a suitcase, it will be a suitable imagingdevice in routine, mobile and teleretinal diabetic ret-inopathy screening clinics in both metropolitan andrural areas.

METHODS

Design

This single-centre study aimed to validate andcompare a new diagnostic device, EyeScan, with thecurrently accepted diagnostic device FF450 plus(Carl Zeiss Meditec, Inc., North America), with ref-erence to slit-lamp examination by a consultantophthalmologist.

Study sample

We enrolled 136 consecutive patients (272 eyes) fromthe diabetic retinopathy screening clinic of RoyalPerth Hospital, Western Australia, into our study.All patients signed an informed consent forparticipation. This study was approved by the RoyalPerth Hospital Ethics Committee.

Patients’ characteristics and diabeteshistory

We collected information on patients’ demographics(e.g. age and ethnicity), current and past ocularhistory, diabetes history [e.g., type, duration, gly-cated hemoglobin level, macro- and microvascularcomplications (cerebrovascular accidents, ischaemicheart disease, peripheral vascular disease, nephropa-thy and neuropathy)] and other associated cardiovas-cular risk factors (e.g. smoking status, blood pressureand lipid profile).

Screening process

Upon arrival at the screening clinic, all patientsreceived pupil-dilating drops (2.5% phenylephrineand 0.5% tropicamide) in both eyes. They under-went three sets of retinal examinations in thefollowing order: (i) non-stereo colour retinal stillphotography (FF450 plus); (ii) non-stereo colourretinal still photography (EyeScan) and (iii) slit-lamp biomicroscopy examination with a 78-dioptrelens by a senior consultant ophthalmologist.

Retinal still photography using the EyeScan andFF450 plus was performed by a medical officer and aretinal photographer, respectively. In order to assessthe usability of the EyeScan device, a medical officerwith no previous experience in ocular imaging wasrecruited and compared to the retinal photographerwho has had 10 years experience in performingretinal still photography for diabetic retinopathyscreening. Three retinal fields (optic disc, macula andtemporal views) were captured using both devices,and the images were subsequently de-identified,randomized and interpreted by a consultant ophthal-mologist and a medical officer (who has graded morethan 1000 colour fundus photos of patients with dia-betes) on a 27-in. iMac (Apple, Cupertino, CA, USA)with a display resolution of 2560 ¥ 1440 pixels in adimly lit room.

The retinal digital still images of the EyeScan andFF450 plus were all downloaded in Joint Photo-graphic Experts Group format. The colour resolutionof the still images of EyeScan and FF450 plus were640 ¥ 480 and 2392 ¥ 1944 pixels, respectively. Thefield angles of EyeScan and FF450 plus were 35 and30 degrees, respectively, centring on optic disc,macular and temporal views.

The retinal images were graded on the basis of thepresence or absence of diabetic retinopathy signs(microaneurysms, retinal haemorrhages, hard exu-dates, cotton wool spots, venous beading, intraretinalmicrovascular abnormalities, new vessel formationand preretinal/vitreous haemorrhage) using the Inter-national Clinical Diabetic Retinopathy Severity Scale(Table 1). The retinal photographs were classified as

A screening device for diabetic retinopathy e41


‘unacceptable’, ‘average’ or ‘excellent’ depending ontheir quality; the retinal photograph was graded as‘unacceptable’ if more than one-third of it was‘blurred’ or ‘uninterpretable’.


We calculated the sensitivity and specificity of thetwo imaging devices (EyeScan and FF450 plus) indetecting and grading diabetic retinopathy withreference to slit-lamp examination. In addition,Cohen’s kappa coefficient was utilized as a measureof agreement for diabetic retinopathy signs andgrading using the two types of imaging devices. Thetechnical failure rate was defined as the fractionof the ‘unacceptable’ retinal images captured byboth devices; such images were excluded from thecalculation for sensitivity, specificity and kappacoefficient. All data were analysed using Statisti-cal Package for Social Sciences version 17 (SPSS,Chicago, IL, USA).

RESULTS

Patients’ demographics and clinicalcharacteristics

A total of 136 patients (272 eyes) participated in ourstudy. The mean � standard deviation age of partici-pants was 53.9 � 15.3 years, duration of diabeteswas 13.9 � 9.9 years, and glycated hemoglobin was8.0 � 1.7%. Among the recruited patients, 74%(n = 101) were Whites, 17% (n = 23) were Asians,and 9% (n = 12) were from other ethnic groups.Of these, 96 patients (71%) had type 2 diabetes.Figure 1 shows the colour fundus images capturedby EyeScan and FF450 plus.

The best-corrected visual acuity of 240 eyes (88%)was 6/6 or 6/9, and that of 23 eyes (9%) was between6/12 and 6/36, and that of nine eyes (3%) was 6/60or less. Of the consecutively recruited eyes, nearly35% had diabetic retinopathy ranging from mildnon-proliferative diabetic retinopathy to prolife-rative diabetic retinopathy (Table 2). Nearly 15%(n = 37) of eyes had previously received panretinalphotocoagulation, and cataracts were diagnosed in28 eyes (10.3%) on the basis of slit-lamp biomicros-copy examination using Lens Opacities Classifica-tion III.11 Almost 45% (n = 118) of the patients hadnever undergone any diabetic retinopathy screening.Of the self-reported diabetes-related complications,diabetic neuropathy (23%, n = 62) and nephropa-thy (22%, n = 60) were the leading complications(Table 3).

Main outcome measure

Compared with the slit-lamp biomicroscopy exami-nation, EyeScan, in detecting any grade of diabeticretinopathy, had a sensitivity and specificity of 93%(95% confidence interval [CI]: 84.9–97.1) and 98.2%(95% CI: 94.3–99.5), respectively, when graded bythe ophthalmologist; however, they were 91.7%(95% CI: 83.2–96.3) and 94.7% (95% CI: 89.9–97.4),respectively, for the medical officer (Table 4). In con-trast, FF450 plus images graded by the ophthalmolo-gist had a sensitivity and specificity of 95.1% (95%CI: 87–98.4) and 98.8% (95% CI: 95.4–99.8), res-pectively, whereas for the medical officer, he had a

Table 1. International Clinical Diabetic Retinopathy SeverityScales10


None No abnormalitiesMild NPDR Microaneurysms onlyModerate NPDR More than just microaneurysms but less

than severe NPDRSevere NPDR Any of the following:

i. Extensive (>20) intraretinal haemor-rhages in each of four quadrants

ii. Definite venous beading in 2+ quadrantsiii. Prominent IRMA in 1+ quadrantAND no signs of PDR

PDR One or more of the following:i. Neovascularizationii. Vitreous/preretinal haemorrhage

IRMA, intraretinal microvascular abnormalities; NPDR, non-proliferative diabetic retinopathy; PDR, proliferative diabeticretinopathy.

Table 2. Diabetic retinopathy grading of the study patientsbased on slit-lamp biomicroscopy examination

Diabetic retinopathy severity Eyes (n) %

Normal 181 66.5Mild non-proliferative diabetic retinopathy 51 18.8Moderate non-proliferative diabetic retinopathy 26 9.6Severe non-proliferative diabetic retinopathy 8 2.9Proliferative diabetic retinopathy 6 2.2Maculopathy 6 2.2

Table 3. The self-reported diabetes micro- and macrovascularcomplications of the enrolled study population

Diabetes complications Patientsn (%)

MicrovascularNeuropathy 31 (23)Nephropathy 30 (22)

MacrovascularIschaemic heart disease 23 (17)Peripheral vascular disease 13 (10)Cerebrovascular disease 9 (7)

e42 Ting et al.


sensitivity and specificity of 91.9% (95% CI: 83.4–96.4) and 95.9% (95% CI: 91.5–98.2), respectively.For the detection of sight-threatening diabetic retin-opathy (severe non-proliferative diabetic retinopa-thy and proliferative diabetic retinopathy), thesensitivity and specificity of images from bothdevices (EyeScan and FF450 plus) graded by bothreaders increased to 100%.

The technical failure rate for EyeScan and FF450plus were 8.5 and 7%, respectively, and they werenot statistically significant (c2 = 0.23, d.f. = 1, P =0.63). Of the failed retinal photographs captured bythe EyeScan, 39% (n = 9) were photographs of eyes

with cataracts and 9% (n = 2) with dark fundi; in52% (n = 12) of the photographs, failure was due tointolerance to bright flash. On the other hand, the‘uninterpretable’ Zeiss retinal images were second-ary to cataracts (42.1%, n = 8), dark fundi (10.5%,n = 2) and intolerance to bright flash (47.4%, n = 9).

The overall kappa statistics for diabetic retinopa-thy grading for EyeScan and FF450 plus were 0.93and 0.95 for ophthalmologist and 0.88 and 0.90 formedical officer, respectively (Table 4). The kappacoefficients for all diabetic retinopathy signs, exceptmacular oedema, based on the analysis of EyeScanand FF450 plus images by both readers, with refer-

EyeScan: normal fundus

EyeScan: subhyaloid haemorrhage

FF450 plus: normal fundus

FF450 plus: subhyaloid haemorrhage

Figure 1. Same fundus imagescaptured by EyeScan and FF450plus.

Table 4. The sensitivity, specificity and kappa correlations of overall diabetic retinopathy grading by a consultant ophthalmologist anda medical officer from colour fundus photographs of EyeScan and FF450 with reference to slit-lamp biomicroscopy examination by aconsultant ophthalmologist

Graders Sensitivity Specificity Kappa correlation(95% CI) (95% CI)

OphthalmologistEyeScan 93% 98.20% 0.93

(84.9–97.1) (94.3–99.5)Zeiss 95.10% 98.80% 0.95

(87–98.4) (95.4–99.8)Medical officer

EyeScan 91.70% 94.70% 0.88(83.2–96.3) (89.9–97.4)

Zeiss 91.90% 95.90% 0.9(83.4–96.4) (91.5–98.2)

CI, confidence interval.



ence to the slit-lamp biomicroscopy examination,were more than 0.8 (Table 5). The kappa coefficientsfor the ophthalmologist in detecting diabetic macul-opathy using EyeScan and FF450 plus were 0.70 and0.74, respectively, whereas for the medical officer,they were 0.71 and 0.76, respectively.

DISCUSSION

In our study, we utilized the International ClinicalDiabetic Retinopathy Severity Scale (Table 1)10 as thegrading system as it is much more simplified, withless severity levels and diagnostic criteria, as com-pared with the Early Treatment Diabetic RetinopathyStudy classification system (Table 6).12 In addition,the slit-lamp examination was chosen as the referencestandard, given that it has been shown to be com-pared favourably with the seven-field stereoscopic30 degrees Early Treatment Diabetic RetinopathyStudy.13 In addition, it is easy to perform, less timeconsuming and more tolerable especially when allpatients had to undergo two sets of retinal stillimaging on EyeScan and FF450 plus. Should thequality of the retinal still images were compromiseddue to the presence of cataracts or pterygiums(encroaching the visual axis), the slit-lamp examina-tion by an ophthalmologist would be more accuratethan the retinal still images captured using the seven-field Early Treatment Diabetic Retinopathy Study.

In this study, we have shown that with referenceto the slit-lamp examination, both readers had com-parable sensitivity and specificity in grading diabeticretinopathy from retinal photographs captured usingthe EyeScan machine or FF450 plus (Table 4).Both devices had a sensitivity and specificity of100% in detecting sight-threatening diabetic retino-pathy changes. Additionally, the kappa statistics ofEyeScan and FF450 plus for the detection of all dia-betic retinopathy signs, except for macular oedema,were more than 0.8 (Table 5). Because the imagesobtained from EyeScan showed excellent sensitivity,specificity and kappa coefficients in diagnosing dia-betic retinopathy, it could be used as reliably as cur-rently used cameras for the screening of diabeticretinopathy.

The colour fundus images from both devicesgraded by the ophthalmologist and medical officerhad kappa coefficients of less than 0.8 for detectingdiabetic maculopathy (Table 5). Because retinal pho-tographs provide only two-dimensional views of theretina, colour fundus images do not afford easy iden-tification of any retinal thickening or macularoedema. In this study, we diagnosed diabetic macu-lopathy on the basis of the presence of hard exudates,microaneurysms and retinal haemorrhages close tothe macular area. Presence of these lesions in themacula region should always prompt an urgentreferral.

Table 5. Kappa statistics for retinal photography using EyeScan and FF450 plus in comparison with the gold standard slit-lampbiomicroscopy examination by an ophthalmologist and a medical officer

Retinal findings Kappa statistics

Retinal photography Retinal photographyEyeScan FF450 plus

OphthalmologistMicroaneurysms 0.94 0.97Retinal haemorrhages 0.87 0.95Cotton wool spots 1 1Venous beading 1 0.9Intraretinal microvascular abnormalities 0.93 0.93New vessels formation 1 1Vitreous haemorrhage 1 1Hard exudates 0.97 1Macular oedema 0.7 0.74Cupped optic disc 1 1

Professional graderMicroaneurysms 0.88 0.88Retinal haemorrhages 0.88 0.91Cotton wool spots 1 1Venous beading 0.89 0.89Intraretinal microvascular abnormalities 0.76 0.76New vessels formation 1 1Vitreous haemorrhage 1 1Hard exudates 0.94 0.94Macular oedema 0.71 0.76Cupped optic disc 0.87 0.89

e44 Ting et al.


In order to evaluate the usability of EyeScan, amedical officer with no previous experience in ocularimaging was recruited as the operator of the device.He received a single day’s training on the deviceprior to the screening. The technical failure rate ofthe FF450 plus operated by an experienced retinalphotographer (10 years of experience on retinal stillphotography for diabetic retinopathy screening) andthe EyeScan were similar (8.5% vs. 7%, P > 0.05).These figures show that EyeScan may be used bynon-experienced personnel with minimal training.Further studies will be of great value in evaluatingthe user-friendliness of the EyeScan for both medicaland non-medical personnel in the primary health-care and teleophthalmology setting.

In our study, assessments by both readers hadexcellent sensitivity, specificity and kappa coeffi-cients in detecting and grading diabetic retinopathyusing the colour fundus images captured by bothdevices (Tables 4,5). These results indicate that

non-specialist personnels, such as primary carephysicians and allied health personnels, can betrained to screen for diabetic retinopathy in theprimary health-care setting. The EyeScan will be asuitable device to be utilized in a teleophthalmol-ogy setting as it requires small storage capacity forits lower resolution files, and thus, the availablebandwidth would better handle the transfer andarchives of the images form community to central-ized screening centre. Further research is requiredto evaluate the cost effectiveness of using Eyescanin a community and teleophthalmology settingusing primary eye care providers, such as optom-etrists, general practitioners, orthoptists and diabe-tes nurses, to screen for diabetic retinopathy. Theophthalmologist will not be able to service the pro-jected increase in numbers of patients with dia-betes. Optimizing the use of people resources,embracing new and affordable technology will benecessary to effectively screen these patients andtherefore reduce the impact of diabetic retinopathy-related visual impairment.

The strength of our study was that all recruitedpatients underwent colour fundus photo-imagingby using both devices (EyeScan and FF450 plus),and this allowed a head-to-head comparisonbetween the two devices. In addition, we have uti-lized two different statistical methods (sensitivity/specificity and Cohen’s kappa statistics) and tworeaders (an ophthalmologist and a medical officer)to increase the reliability and validity of our studyfindings. In contrast, our study carries a few limi-tations, one of which was that the slit-lamp exami-nation (reference standard) was performed by asingle senior consultant ophthalmologist. In addi-tion, all patients successively underwent two sets ofretinal still photography within a short period oftime, and this could have contributed to the tech-nical failure that had occurred as a consequence ofpatients’ intolerance to bright light from both thedevices.

REFERENCES

1. Wild S, Roglic G, Green A, Sicree R, King H. Globalprevalence of diabetes: estimates for the year 2000 andprojections for 2030. Diabetes Care 2004; 27: 1047–53.

2. Ferris FL 3rd. How effective are treatments for diabeticretinopathy? JAMA 1993; 269: 1290–1.

3. Photocoagulation for Diabetic Macular Edema: EarlyTreatment Diabetic Retinopathy Study Report Num-ber 1 Early Treatment Diabetic Retinopathy StudyResearch Group. Arch Ophthalmol 1985; 103: 1796–806.

4. Photocoagulation Treatment of Proliferative DiabeticRetinopathy. Clinical Application of Diabetic Retin-opathy Study (DRS) Findings, DRS Report Number 8.The Diabetic Retinopathy Study Research Group. Oph-thalmology 1981; 88: 583–600.

Table 6. Classification of diabetic retinopathy into retinopathystages (Wisconsin level)12

Diabetic retinopathystage

Retinal findings

Minimal NPDR MA and one or more of the following:retinal haem, Hex, CWS, but notmeeting the criteria for moderateNPDR

Moderate NPDR H/Ma > std photo 2A in at least onequadrant and one of more of: CWS,VB, IRMA, but not meeting severeNPDR

Severe NPDR Any of:H/Ma > std photo 2A in all four

quadrants,IRMA > std photo 8A in one or

more quadrants,VB in two or more quadrants

PDR Any of:NVE or NVD < std photo 10A,

vitreous/preretinal haemNVE < 1/2 DA without NVD

High-risk PDR Any of:NVD > 1/4 to 1/3 disc area,or with vitreous/preretinal haem,or NVE > 1/2 DA with vitreous/

preretinal haemAdvanced PDR High-risk PDR with tractional

detachment involving macula orvitreous haem obscuring ability tograde NVD and NVE

CWS, cotton wool spots; DA, disc area; Haem, haemorrhages;H/Ma, haemorrhages and microaneurysms; Hex, hard exudates;IRMA, intraretinal microvascular abnormalities; MA, microaneu-rysms; NPDR, non-proliferative diabetic retinopathy; NVD, newvessels disc; NVE, new vessels elsewhere; PDR, proliferative dia-betic retinopathy; std, standard; VB, venous beading.



5. Ting DSW, Ng J, Morlet N et al. Diabetic retinopathymanagement by Australian optometrists. Clin Experi-ment Ophthalmol 2011; 39: 230–5.

6. Arun CS, Al-Bermani A, Stannard K, Taylor R. Long-term impact of retinal screening on significantdiabetes-related visual impairment in the working agepopulation. Diabet Med 2009; 26: 489–92.

7. Hutchinson A, McIntosh A, Peters J et al. Effectivenessof screening and monitoring tests for diabetic retinopa-thy – a systematic review. Diabet Med 2000; 17: 495–506.

8. James M, Turner DA, Broadbent DM, et al. Cost effec-tiveness analysis of screening for sight threatening dia-betic eye disease. BMJ 2000; 320: 1627–31.

9. Garvican L, Scanlon PH. A pilot quality assurancescheme for diabetic retinopathy risk reductionprogrammes. Diabet Med 2004; 21: 1066–74.

10. Wilkinson CP, Ferris FL 3rd, Klein RE et al. Proposedinternational clinical diabetic retinopathy and diabetic

macular edema disease severity scales. Ophthalmology2003; 110: 1677–82.

11. Chylack LT Jr, Wolfe JK, Singer DM et al. The LensOpacities Classification System III. Arch Ophthalmol1993; 111: 831–6.

12. Early Treatment Diabetic Retinopathy Study ResearchGroup. Grading diabetic retinopathy from stereoscopiccolor fundus photographs – an extension of the modi-fied Airlie House classification. ETDRS report number10. Early Treatment Diabetic Retinopathy StudyResearch Group. Ophthalmology 1991; 98: 786–806.

13. Scanlon PH, Malhotra R, Greenwood RH et al. Com-parison of two reference standards in validating twofield mydriatic digital photography as a method ofscreening for diabetic retinopathy. Br J Ophthalmol2003; 87: 1258–63.

e46 Ting et al.


Diabetic retinopathyscreening: can theviewing monitorinfluence thereading and gradingoutcomes

DSW Ting1 ;2 ;3, ML Tay-Kearney2 ;3, J Vignarajan1

and Y Kanagasingam1

Abstract

Purpose To evaluate the accuracy of

different viewing monitors for image reading

and grading of diabetic retinopathy (DR).

Design Single-centre, experimental case

series—evaluation of reading devices for

DR screening.

Method A total of 100 sets of three-field

(optic disc, macula, and temporal views)

colour retinal still images (50 normal and

50 with DR) captured by FF 450 plus

(Carl Zeiss) were interpreted on 27-inch

iMac, 15-inch MacBook Pro, and 9.7-inch

iPad. All images were interpreted by a retinal

specialist and a medical officer. We calculated

the sensitivity and specificity of 15-inch

MacBook Pro and 9.7-inch iPad in detection

of DR signs and grades with reference to the

reading outcomes obtained using a 27-inch

iMac reading monitor.

Results In detection of any grade of DR,

the 15-inch MacBook Pro had sensitivity and

specificity of 96% (95% confidence interval

(CI): 85.1–99.3) and 96% (95% CI: 85.1–99.3),

respectively, for retinal specialist and 91.5%

(95% CI: 78.7–97.2) and 94.3% (95% CI:

83.3–98.5), respectively, for medical officer,

whereas for 9.7-inch iPad, they were 91.8%

(95% CI: 79.5–97.4) and 94.1% (95% CI:

82.8–98.5), respectively, for retinal specialist

and 91.3% (95% CI: 78.3–97.1) and 92.6%

(95% CI: 81.3–97.6), respectively, for medical

officer.

Conclusion The 15-inch MacBook Pro and

9.7-inch iPad had excellent sensitivity and

specificity in detecting DR and hence, both

screen sizes can be utilized to effectively

interpret colour retinal still images for DR

remotely in a routine, mobile or tele-

ophthalmology setting. Future studies could

explore the use of more economical devices

with smaller viewing resolutions to reduce

cost implementation of DR screening services.

Eye (2012) 26, 1511–1516; doi:10.1038/eye.2012.180;

published online 12 October 2012

Keywords: diabetic retinopathy; screen

resolution; screen size; screening

Introduction

Diabetic retinopathy screening programs have

been implemented worldwide to enable early

detection of diabetic retinopathy, which if

treated appropriately, will minimize severe

visual impairment.1 By 2030, it is estimated that

the total number of people with diabetes will

rise to 366 million.2 Because of the rising

prevalence of diabetes, the current screening

services in developing and developed countries

will be faced with increasing costs of

implementation and maintenance of a screening

programme for the people with diabetes.3 It is

therefore prudent that stakeholders continue

to look for different ways of servicing the

increasing diabetic population, and at the same

time minimizing the economical impact of

screening programs within the community.

To date, various studies have evaluated

various parameters, which may affect the

sensitivity, specificity, and cost effectiveness

in screening diabetic retinopathy, including

numbers of retinal fields,4 colour or

monochromatic fundus photographs,5

mydriatic status,6,7 photographers and

readers with different medical qualifications,7

automated grading system,8 use of an

economical retinal camera,9 and retinal video

recording technique.10 However, none of the

1The Australian E-HealthResearch Centre,Commonwealth ScientificIndustrial ResearchOrganization (CSIRO),Floreat, Western Australia,Australia

2Center for Ophthalmologyand Visual Sciences, LionsEye Institute, University ofWestern Australia,Nedlands, WesternAustralia, Australia

3OphthalmologyDepartment, Royal PerthHospital, Perth, WesternAustralia, Australia

Correspondence:DSW Ting, Center ofOphthalmology and VisualScience, University ofWestern Australia, LionsEye Institute, 2 VerdunStreet, Nedlands, WesternAustralia 6009, AustraliaTel: þ61 8 9224 2167;Fax: þ 61 8 9224 3511.E-mail: [email protected]

Received: 15 November2011Accepted in revised form:15 April 2012Published online:12 October 2012

CL

INIC

AL

ST

UD

Y

Eye (2012) 26, 1511–1516& 2012 Macmillan Publishers Limited All rights reserved 0950-222X/12

www.nature.com/eye

http://dx.doi.org/10.1038/eye.2012.180

mailto:[email protected]


http://www.nature.com/EYE

studies have compared the use of different viewing

monitors for the reading and grading of diabetic

retinopathy from the digital colour fundus photographs.

To date, the use of small viewing monitors in screening

for diabetic retinopathy has become an emerging trend

among the ophthalmologists or professional graders,

as they can utilize them remotely without being

physically present at the reading centre. Because of the

growing popularity of these mobile and portable

technologies, the purpose of our study is to evaluate the

efficacy of using different portable and mobile devices

with varying viewing monitors (15-inch MacBook Pro

and 9.7-inch iPad) to detect subtle diabetic retinopathy

changes (microaneurysms and dots haemorrhages) and

diagnose the severity level. This also helps to determine

the suitability of using smaller and more affordable

portable devices to interpret the colour retinal images for

grading of diabetic retinopathy.

Materials and methods

Design and data acquisition

This is a single-centre case series to evaluate different

screening resolutions to interpret retinal colour still

images for diabetic retinopathy screening. A total of

100 sets of non-stereo mydriatic three-field (optic disc,

macula and temporal views) 35 degrees colour retinal

still images consisting of 50 normal and 50 with diabetic

retinopathy were selected into our study. The quality of

all recruited retinal images were at least ‘acceptable’

(more than two-third of retinal images were ‘interpretable’)

based on the reading outcomes using a 27-inch iMac

(the standard viewing screen in our reading centre).

All images were captured using FF 450 plus (Carl Zeiss,

Inc., Oberkochen, Germany) by an experienced retinal

photographer at the Diabetic Retinopathy Screening

Clinic of Royal Perth Hospital and downloaded in Joint

Photographic Experts Group (JPEG) format. The colour

resolution of all images was fixed at 2588� 1958 pixels.

This study has been approved by the Royal Perth

Hospital Human Research Ethics committee.

Data interpretation

All images were de-identified, randomized, and

interpreted by two readers (a retinal specialist and a

medical officer) in a dark room using the standardized

Apple software—iPhoto (Apple, Cupertino, CA, USA) on

three monitors with different sizes—27-inch iMac (Apple),

15-inch MacBook Pro (Apple), and 9.7-inch iPad (Apple).

All images were interpreted on the specific three

monitors with calibrated brightness at 100%; target white

point at D65; and target gamma at 2.2 using a software

named Display Calibrator Assistant. The specification of

the reading devices were listed in Table 1. The quality

of retinal images was rated as ‘acceptable’ or

‘uninterpretable’ by the readers. The diabetic retinopathy

severity level was graded based on presence/absence of

microaneurysms, retinal haemorrhages, cotton wool

spots, venous beading, intraretinal microvascular

abnormalities, new vessels formation, and hard exudates

using the International Clinical Diabetic Retinopathy

Severity Scales11 (Table 2).


To allow for a power of 95%, desired precision of 0.10,

expected sensitivity and specificity of 90%, the total

number of eyes required for each device was 71

(prevalence was set at 0.50, as the selected samples

consisted of 50% normal and 50% abnormal retinal

colour still images).


All data were analysed using SPSS version 17 (SPSS,

Chicago, IL, USA). The sensitivity, specificity, and Kappa

correlation coefficient of 15-inch MacBook Pro and 9.7-

inch iPad in detecting diabetic retinopathy lesions and

grading were calculated with reference to the findings on

the 27-inch iMac (as the reference standard). Moreover,

the Kappa coefficient was performed on the diabetic

retinopathy grading detected on 27-inch iMac for both

readers and diabetic retinopathy lesions detected by

Table 1 Specifications and prices of 27-inch iMac, 15-inch MacBook Pro, and 9.7-inch iPad (the indicated prices are obtainedin US dollars)

27-inch iMac 15-inch MacBook Pro 9.7-inch iPad (Wi-Fiþ 3G)

Monitor screen (inch) 27 15 9.7Screen resolution (pixels) 1920� 1080 1440� 900 1024� 768

ATI Radeon HD NVIDIA GeForceGraphics processor 4670 320M A4 1GhzWeight (kg) 13.8 2.54 0.73Price (USD) $1699 $1799 $629

Monitor screens for diabetic retinopathy screeningDSW Ting et al

1512

Eye

15-inch MacBook Pro and 9.7-inch iPad with reference

to 27-inch iMac.

Results

The mean age (±SD) of the recruited participants was

51.3±13.8 years with HbA1c of 8.4±1.6% and duration

of diabetes of 12.1±8.7 years. Of the retinal images,

50 (50%) had no diabetic retinopathy, 16 (16%) had mild

non-proliferative diabetic retinopathy (NPDR), 25 (25%)

had moderate NPDR, 7 (7%) had severe NPDR, and 2 (2%)

had proliferative diabetic retinopathy. All retinal images

were rated as ‘acceptable’ by the retinal specialist and

medical officer on 15-inch MacBook Pro and

9.7-inch iPad.

For 27-inch iMac (the ‘reference standard’ of our

study), both retinal specialist and medical officer had a

Kappa correlation of 0.88 in detecting the overall diabetic

retinopathy grading. In detection of any grade of diabetic

retinopathy on 15-inch MacBook Pro, the retinal

specialist had sensitivity and specificity of 96% and 96%,

respectively, whereas the medical officer had 91.5% and

94.3%, respectively, with reference to the 27-inch iMac

(Table 3). On the other hand, the sensitivity and

specificity in detecting any grade of diabetic retinopathy

on 9.7-inch iPad for retinal specialist were 91.8% and

94.1%, respectively, whereas for medical officer, they

were 91.3 and 92.6% respectively. For sight-threatening

diabetic retinopathy, the retinal specialist had 100%

sensitivity and specificity on both reading devices,

whereas for medical officer, the sensitivity, and specificity

were 100% and 97.7%, respectively, on both devices.

The iPad had lower sensitivity and specificity

(retinal specialist: 89.1% and 96.3%, respectively; medical

officer: 87.5% and 98.1%, respectively) in detecting

microaneurysms by both readers compared with

MacBook Pro (sensitivity— retinal specialist: 100% and

96.3%, respectively; medical officer: 95.8% and 100%,

respectively) (Table 4). Both devices had comparable

sensitivity and specificity in detecting retinal

haemorrhages by both readers (Table 4).

For retinal specialist, the Kappa coefficient for 15-inch

MacBook Pro and 9.7-inch iPad in detection of any grade

of diabetic retinopathy were 0.94 and 0.89, respectively,

whereas for medical officer, they were 0.89 and 0.88,

respectively, with reference to 27-inch iMac. The Kappa

Table 2 International clinical diabetic retinopathy severityscales9


None No abnormalitiesMild NPDR Microaneurysms onlyModerate NPDR More than just microaneurysms


Severe NPDR Any of the following:i. Extensive (420) intraretinal

haemorrhages in each of 4 quadrantsii. Definite venous beading in 2þ quadrantsiii. Prominent IRMA in 1þ quadrant and no

signs of proliferative DR

Proliferative DR One or more of the following:i. Neovascularization

ii. Vitreous/pretinal haemorrhage

Abbreviations: NPDR, non-proliferative diabetic retinopathy; DR, diabetic

retinopathy.

Table 3 The sensitivity, specificity, and Kappa coefficient of 15-inch MacBook Pro and 9.7-inch iPad in detecting diabetic retinopathygrading by a retinal specialist and a medical officer with reference to 27-inch iMac

Any grade of diabetic retinopathy Sensitivity (95% CI) Specificity (95% CI) Kappa

Retinal specialist15-inch MacBook Pro 96% (85.1–99.3) 96% (85.1–99.3) 0.949.7-inch iPad 91.8% (79.5–97.4) 94.1% (82.8–98.5) 0.89

Medical officer15-inch MacBook Pro 91.5% (78.7–97.2) 94.3% (83.3–98.5) 0.899.7-inch iPad 91.3% (78.3–97.1) 92.6% (81.3–97.6) 0.88

Sight-threatening diabetic retinopathyRetinal specialist

15-inch MacBook Pro 100% 100% 1.009.7-inch iPad 100% 100% 1.00

Medical officer15-inch MacBook Pro 100% 97.7% 0.929.7-inch iPad 100% 97.7% 0.92

Sight-threatening diabetic retinopathy comprises of severe non-proliferative diabetic retinopathy and proliferative diabetic retinopathy.


1513

Eye

coefficient in detecting all diabetic retinopathy signs

(microaneurysms, retinal haemorrhages, cotton wool

spots, new vessels formation, and hard exudates) by both

readers were more than 0.80 (Table 5).

Discussion

The success of a screening process relies on multiple

factors, including the photographers’ factor, patients’

factor, and readers’ factor. In the presence of an

experienced photographer, patients with good ocular

media, and experienced readers, the influence of viewing

monitors also has a role in determining the diagnostic

accuracy of retinal images grading. To evaluate the

effectiveness of 15-inch Macbook Pro and 9.7-inch iPad in

detecting diabetic retinopathy lesions and grading, we

compared the retinal findings of each of the devices with

the respective findings on 27-inch iMac. In our study, the

retinal specialist and the medical officer as the trained

reader had extremely strong inter-observer agreement

(Kappa¼ 0.88) in grading diabetic retinopathy on the

27-inch iMac. Both readers had excellent sensitivity and

specificity in diagnosing diabetic retinopathy using

15-inch MacBook Pro (retinal specialist: 96%, 96%,

respectively; medical officer: 91.5%, 94.3%, respectively)

and 9.7-inch iPad (retinal specialist: 91.8%, 94.1%,

respectively; medical officer: 91.3%, 92.6%, respectively)

(Table 3). In addition, the Kappa correlation between

15-inch MacBook Pro and 9.7-inch iPad vs 27-inch

iMac in detection of diabetic retinopathy changes

(microaneurysms, retinal haemorrhages, cotton wool

spots, neovascularization, and hard exudates) and

diabetic retinopathy grading were excellent (greater than

0.8). These results indicated that the specialist (retinal

specialist) and non-specialist (medical officer) screeners

could effectively interpret and diagnose diabetic

retinopathy from the colour retinal still images

using a 15-inch or a 9.7-inch reading screen.

In detection of sight-threatening diabetic retinopathy

(severe NPDR), the medical officer had 100% sensitivity

and 97.7% specificity on both devices (MacBook Pro and

iPad). The discrepancy of the specificity between the

retinal specialist and the medical officer were due to two

false positives, which had been graded by the medical

officer as severe NPDR instead of moderate NPDR.

Nevertheless, a screener especially the non-

ophthalmologist personnel, such as the optometrists and

general practitioners, should always be suspicious and

have lower threshold in referring patients with uncertain

diabetic retinopathy lesions detected on the retinal still

images, even if this will result in some ‘unnecessary’

referrals to the specialists.

The native image resolution of 2588� 1958 pixels

exceeded any of the compared displays spatial

capabilities. This image resolution of 2588� 1958 pixels

was set by the fundus camera (Zeiss FF 450 plus) and all

images were interpreted using a common software—

iPhoto (Apple). The image size exceeded the compared

displays spatial resolution and therefore, all the images

were set to 100% to fit the full screen during the viewing

and interpretation process. However, the full image may

still be navigated on different screens using the iPhoto

display programme with ease.

Table 4 The sensitivity, specificity and Kappa coefficient of15-inch MacBook Pro and 9.7-inch iPad in detecting micro-aneurysms and retinal haemorrhages by a retinal specialist and amedical officer with reference to 27-inch iMac

Microaneurysms Sensitivity(95% CI)

Specificity(95% CI)

Retinal specialist15-inch Macbook Pro 100% (90.4–100) 96.3% (86.1–99.4)9.7-inch iPad 89.1% (75.6–95.9) 96.3% (86.2–99.4)

Medical officer15-inch Macbook Pro 95.8% (84.6–99.3) 100% (91.4–100)9.7-inch iPad 87.5% (74.1–94.8) 98.1% (88.4–99.9)

Retinal haemorrhagesRetinal specialist

15-inch Macbook Pro 93.8% (77.8–98.9) 98.5% (91–99.9)9.7-inch iPad 93.8% (77.8–98.9) 95.6% (86.8–98.9)

Medical officer15-inch Macbook Pro 93.9% (78.4–98.9) 98.5% (90.9–99.9)9.7-inch iPad 93.9% (78.4–98.9) 95.5% (86.6–98.8)

Table 5 The Kappa correlation of diabetic retinopathy changesinterpreted by a retinal specialist and a medical officer on15-inch MacBook Pro and 9.7-inch iPad with reference to theretinal findings detected on 27-inch iMac

Retinal findings Kappa statistics

15-inchMacBook Pro

9.7-inchiPad

Retinal specialistMicroaneurysms 0.96 0.86Retinal haemorrhages 0.93 0.89Cotton wool spots 1.00 1.00New vessels formation 1.00 1.00Hard exudates 0.96 0.96Cupped optic disc 1.00 1.00

Medical officerMicroaneurysms 0.92 0.86Retinal haemorrhages 0.93 0.91Cotton wool spots 1.00 1.00New vessels formation 1.00 1.00Hard exudates 0.96 0.96Cupped optic disc 1.00 1.00


1514

Eye

In our study, we utilized the 27-inch iMac as the

reference standard of our study to avoid any diagnostic

error secondary to small screen size and low screen

resolution. We adopted the mydriatic 50 degrees three-

field retinal still photography in our screening clinic

given that its sensitivity in detecting any grade of

diabetic retinopathy has been shown to be more than

90%. Compared with the gold standard seven-field

30 degrees stereoscopic views, it is more time saving and

practical to be implemented in the routine screening

setting. For the displaying programme, the ‘iPhoto’ was

utilized instead of the more specialized programs such as

‘Visupac’ or ‘IMAGEnet system’, as the latter programs

often will need to be purchased. On the other hand, it is

more economical to use ‘iPhoto’ programme as it is a

free software, which is included in the Mac computers.

In our study, the iPad had slightly lower sensitivity,

specificity, and Kappa correlation with the 27-inch iMac

compared with the 15-inch MacBook Pro (Table 4). Both

devices, however, had excellent diagnostic accuracy in

detecting sight-threatening diabetic retinopathy lesions

(retinal haemorrhages, cotton wool spots, and new

vessels) and diabetic retinopathy grading by both readers

(Table 3). In a screening setting, it is rather critical

to detect and refer patients with sight-threatening

diabetic retinopathy changes, such as multiple retinal

haemorrhages, cotton wool spots, and neovascularization,

such that pan-retinal photocoagulation could be

applied without delay to prevent severe visual

impairment. Therefore, an occasional missed

microaneurysm often will not result in severe visual

impairment and this is consistent with the findings

of our study given that both readers had 100%

sensitivity in diagnosing sight-threatening diabetic

retinopathy on both devices. Given the Kappa correlation

between iPad and 27-inch iMac, graded by retinal

specialist and medical officer, in detection of

microaneurysms was within the excellent range

(kappa¼ 0.86), it will be feasible for the specialist and

non-specialist readers to utilize a small reading screen

(eg, 9.7-inch iPad) with a spatial resolution of 1024� 768

pixels to effectively screen for diabetic retinopathy in the

community. A further study will be of great value to

explore the efficacy of other cheaper PC, tablet computers

(eg, Galaxy Tab (Samsung, Samsung Town, Seoul,

South Korea)) and cell/smart phones (eg, 3.5-inch iPhone

(Apple) and 4-inch Galaxy S (Samsung)) with smaller

reading screen sizes to screen for diabetic retinopathy

in a routine, mobile, or tele-ophthalmology setting.

The strength of our study was being one of the recent

studies that evaluated the effect of using devices with

varying monitor resolution to screen for diabetic

retinopathy. Moreover, we utilized two statistical

methods (sensitivity/specificity and Kappa coefficient)

and two readers (retinal specialist and medical officer) to

justify the diagnostic accuracy of each monitor size by

the specialist and non-specialist personnel. On the other

hand, one of the weaknesses of our study was that the

colour resolution of the three display monitors was

different. Despite having a similar colour depth for all

screens, the colour gamut, the range, and set of colours

that they can produce, were not the same for all three

monitors. The iMac can display much wider colour

gamut than the other two displays used in this study.

The Macbook Pro and iPad displays have much less

display colour gamut than iMac. The colour gamut

can influence the accuracy of the colours and may

show undersaturated colours and hence, potentially

affecting the interpretation of fundus images.

In addition, our results may be potentially subject to a

selection bias, as we only selected the good quality colour

retinal images into our study. It is unknown if the colour

retinal images with suboptimal quality due to media

opacity or dark fundi will affect the sensitivity and

specificity in detecting diabetic retinopathy changes

using the 15-inch or 9.7-inch reading screens. Thus, it will

be of great significance to recruit all patients with

diabetes consecutively from the screening clinic in the

future study to evaluate the overall efficacy of different

monitor resolutions in detecting diabetic retinopathy

lesions from the colour retinal images with good and

suboptimal quality.

Summary

What was known beforeK Many factors can affect the sensitivity and specificity of

diabetic retinopathy screening, including the numbers offiled, mydriatic status, readers, or photographers withdifferent medical qualifications, use of an economicalretinal camera and other screening techniques.

K None of the study has evaluated the influence of theviewing monitor in screening diabetic retinopathy.

What this study adds

K Our study is one of the few studies that investigatethe influence of different viewing monitors, which canpotentially affect the diagnostic sensitivity and specificity.

K The results show that the grading outcome using a15-inch and 9.7-inch reading device shared a comparablesensitivity and specificity in interpreting retinal colourstill images for diabetic retinopathy screening.

K Given the increasing popularity of tablet computersworldwide, the graders (eg, ophthalmologists orprofessional graders) can safely use them in screeningdiabetic retinopathy without affecting the diagnosticaccuracy.

Conflict of interest

The authors declare no conflict of interest.


1515

Eye

Acknowledgements

Diabetes Australia Research Trust and Royal Perth

Hospital have provided research funding to this project.

The sponsor or funding organization had no role in

the design or conduct of this research.

Author contributions

Daniel SW Ting contributed to the study conception

and design, data acquisition, data analysis, data

interpretation, and drafting the article. M L Tay-Kearney

contributed to the study conception and design,

revision of the important intellectual content, and final

approval of the version to be published. J Vignarajan

contributed to the data acquisition and data analysis.

Y Kanagasingam contributed to the study conception

and design, revision of the important intellectual content,

and final approval of the version to be published.

References

1 Ferris FL III. How effective are treatments for diabeticretinopathy? JAMA 1993; 269: 1290.

2 Wild S, Roglic G, Green A, Sicree R, King H. Globalprevalence of diabetes: estimates for the year 2000 andprojections for 2030. Diabetes Care 2004; 27: 1047–1053.

3 James M, Turner DA, Broadbent DM, Vora J, Harding SP.Cost effectiveness analysis of screening for sight threateningdiabetic eye disease. BMJ 2000; 320: 1627–1631.

4 Scanlon PH, Malhotra R, Greenwood RH, Aldington SJ,Foy C, Flatman M et al. Comparison of two referencestandards in validating two field mydriatic digitalphotography as a method of screening for diabeticretinopathy. Br J Ophthalmol 2003; 87: 1258–1263.

5 Lin D, Blumenkranz M, Brothers R, Grosvenor DM.The sensitivity and specificity of single-field nonmydriaticmonochromatic digital fundus photography with remoteimage interpretation for diabetic retinopathy screening: acomparison with ophthalmoscopy and standardizedmydriatic colour photography. Am J Ophthalmol 2002; 134:204–213.

6 Hutchinson A, McIntosh A, Peters J, O’Keeffe C, Khunti K,Baker R et al. Effectiveness of screening and monitoring testsfor diabetic retinopathy—a systematic review. Diabet Med2000; 17: 495–506.

7 Bragge P, Gruen RL, Chau M, Forbes A, Taylor HR.Screening for presence or absence of diabetic retinopathy:a meta-analysis. Arch Ophthalmol 2011; 129: 435–444.

8 Ting DSW, Tay-Kearney ML, Yogesan K. A light andportable novel device for diabetic retinopathy screening.Clin Experiment Ophthalmol 2011; 40: e40–e46.

9 Ting DSW, Tay-Kearney ML, Constable IJ, Lim L, Preen DB,Kanagasingam Y. Retinal video recording: a new way toimage and diagnose diabetic retinopathy. Ophthalmology2011; 118: 1588–1593.

10 Wilkinson CP, Ferris III FL, Klein RE, Lee PP, Agardh CD,Davis M et al. Proposed international clinical diabeticretinopathy and diabetic macular edema disease severityscales. Ophthalmology 2003; 110: 1677–1682.

11 Olson JA, Strachan FM, Hipwell JH, Goatman KA,McHardy KC, Forrester JV et al. A comparative evaluationof digital imaging, retinal photography and optometristexamination in screening for diabetic retinopathy. DiabetMed 2003; 20: 528–534.


1516

Eye

Retinal Video RecordingA New Way to Image and Diagnose Diabetic Retinopathy

Daniel S. W. Ting, MBBS(Hons),1,2,3 Mei Ling Tay-Kearney, FRANZCO,1,3 Ian Constable, FRANZCO,1

Liam Lim, FRANZCO,3 David B. Preen, PhD,4 Yogesan Kanagasingam, PhD2

Purpose: To validate the use of retinal video recording for diabetic retinopathy screening by comparing withstandard retinal photography and slit-lamp examination.

Design: Evaluation of a new diagnostic technique.Participants: One hundred patients.Methods: All fundus images were captured using standard retinal still photography (FF 450 plus; Carl Zeiss)

and retinal video (EyeScan; Ophthalmic Imaging System, Sacramento, CA), followed by a gold standard slit-lampbiomicroscopy examination. All videos and still images were de-identified, randomized, and interpreted by 2senior consultant ophthalmologists (M.L.T-K. and L.L.). Kappa statistics, sensitivity, and specificity for all thediabetic retinopathy signs and grades were calculated with reference to slit-lamp examination results as the goldstandard.

Main Outcome Measures: Sensitivity and specificity of video recording for detecting diabetic retinopathysigns and grades.

Results: The mean age (�standard deviation [SD]) of participants was 52.8�15.1 years, mean duration ofdiabetes (�SD) was 13.7�9.7 years, and the mean glycosylated hemoglobin level was 8.0�1.7%. Comparedwith the gold standard slit-lamp examination results, the sensitivity and specificity of video recording fordetecting the presence of any diabetic retinopathy was 93.8% and 99.2%, respectively (ophthalmologist 1), and93.3% and 95.2%, respectively (ophthalmologist 2). In contrast, the sensitivity and specificity of retinal photog-raphy was 91.8% and 98.4%, respectively (ophthalmologist 1), and 92.1% and 96.8%, respectively (ophthal-mologist 2), for detection of any diabetic retinopathy. Both imaging methods had 100% sensitivity and specificityin detecting sight-threatening diabetic retinopathy. For overall diabetic retinopathy grading by both ophthalmol-ogists, the measurements of agreement (Cohen’s � coefficient) between the overall grading obtained from theretinal video versus slit-lamp examination and retinal photography versus slit-lamp examination were more than0.90. Technical failure rate for retinal video recording and retinal photography were 7.0% and 5.5%, respectively.

Conclusions: This study demonstrated that retinal video recording was equally as effective as retinalphotography in the subjects evaluated in this study. It is a novel alternative diabetic retinopathy screeningtechnique that not only offers primary eye care providers the opportunity to view numerous retinal fields withina short period but also is easy to learn by nonexperienced personnel with minimal training.

Financial Disclosure(s): Proprietary or commercial disclosure may be found after the references.Ophthalmology 2011;118:1588–1593 © 2011 by the American Academy of Ophthalmology.

easdtaSalaPpt

Diabetes is one of the world’s fastest growing chronicdiseases and a leading cause of acquired vision loss.1 Ac-cording to the World Health Organization, the total numberof people with diabetes for all age groups will more thandouble from 171 million (2.8%) in 2000 to 366 million(4.8%) by 2030.2 On average, 25% of diabetic patients haveidentifiable signs of diabetic retinopathy,3 and the preva-lence of diabetic retinopathy can increase in up to 75% inpatients who have had diabetes for 20 years or longer.4,5

Nevertheless, early detection and prompt treatment has beenreported to be able to prevent up to 98% of diabetes-relatedvisual impairment.6

In the United States, federal savings of $624 million and
400 000 person-years of sight could be achieved annually if s
1588 © 2011 by the American Academy of OphthalmologyPublished by Elsevier Inc.

veryone with diabetes underwent regular diabetic retinop-thy screening and received treatment according to theeverity of their condition.7–9 A national teleretinal imagingiabetic retinopathy screening program was set up betweenhe Veterans Health Administration, Joslin Vision Network,nd the Department of Defense and the Veterans Integratedervice Network to improve the access to diabetic retinop-thy screening for the United States population.10–12 Simi-arly, the National Health Service in the United Kingdomlso has set up a National Diabetic Retinopathy Screeningrogram with the aim to achieve a 100% screening rate foratients with diabetes.13 In Australia, the use of teleoph-halmology in diabetic retinopathy screening also has been
hown to be cost effective and cheaper than any of the
ISSN 0161-6420/11/$–see front matterdoi:10.1016/j.ophtha.2011.04.009

PDigcmain

RAr(sioparuDfur

piw(vhvrvi

IVAatdvcgbspt

twie

gMeGcs

Ting et al � Retinal Video Recording for Diabetic Retinopathy Screening

alternative options, with a minimum number of patients peryear of 128.14,15

To date, all teleretinal screening services worldwide useretinal photography (mydriatic or nonmydriatic) as the dia-betic retinopathy screening tool. For detection of diabeticretinopathy, the Early Treatment Diabetic RetinopathyStudy 7 standard field 35-mm stereoscopic color fundusphotographs using the modified Airlie-House classificationcurrently is the gold standard photographic technique.16

However, the use of single-field mydriatic 45° retinal pho-tography with sensitivity of more than 80% has been shownto be adequate and effective (level 1 evidence)17,18 in ascreening setting, and mydriatic 3-field retinal photographycould increase the sensitivity further to 97%.19

A good-quality retinal image is highly dependent on theoperators’ skills in performing retinal photography, patientcompliance with instructions, and their tolerance to thebright light. As a result, retinal photography often is per-formed by trained allied health professionals who may notbe available in a remote community. Moreover, even whenperformed by experienced retinal photographers, the re-ported technical failure rate of mydriatic retinal photogra-phy has been shown to be as high as 12%.20,21

This study examined a new technique using retinal videorecording to screen for diabetic retinopathy. Screening wasperformed by a simple fundus camera (EyeScan; Ophthal-mic Imaging System, Sacramento, CA) with a digital colorvideo recording function. Unlike traditional retinal photog-raphy, retinal video recording is easier and quicker to learn.This technique also provides a view of the retina thatsimulates what is seen with a slit-lamp examination. Eachretinal video consists of the optic disc, macula, and temporalviews of an eye lasting for 30 to 45 seconds. This is slightlyquicker than the 3-field color retinal still photography,which usually takes up to 90 to 120 seconds. However, thevariation of time taken to capture a retinal video or retinalstill photography often is subject to the experience of aphotographer.

Patients and Methods

Design

This was a single-center masked study to determine the validityof a new diagnostic technique, retinal video recording, forassessing the presence and stage of diabetic retinopathy bycomparison with currently accepted diagnostic practice usingretinal still photography.

Study Sample

A total of 100 patients with diabetes (200 eyes) were recruited forthis study from the Ophthalmology Clinic of Royal Perth Hospital,a public teaching hospital in Western Australia. No patient wasexcluded based on age, gender, socioeconomic status, or clinicalcharacteristics. All participants provided a signed informed con-sent for participation, and the study was approved by the Royal
Perth Hospital Ethics Committee. c
atient Demographics and Diabetes Historyata collected included patient demographics (age and ethnic-

ty), best-corrected visual acuity, ocular history (e.g., cataracts,laucoma, laser therapy), self-reported diabetic and cardiovas-ular history (type and duration of diabetes, glycosylated he-oglobin levels, smoking status, blood pressure, lipid profile),

s well as diabetic complications (cerebrovascular accidents,schemic heart disease, peripheral vascular disease, retinopathy,ephropathy, and neuropathy).

etinal Video and Retinal Photography Protocolll patients underwent 3 separate tests for diagnosis of diabetic

etinopathy: (1) retinal color digital video recording, (2) 3-fieldoptic disc, macula, and temporal views) nonstereo color retinaltill photography, followed by (3) slit-lamp biomicroscopy exam-nation (with a 78-diopter lens) by a senior consultant ophthalmol-gist (M.L.T-K. and L.L.) with a special interest in diabetes. Onresentation at the clinic, patients received 2.5% phenylephrinend 0.5% tropicamide in both eyes to maximize pupil dilation. Theetinal video recording and retinal still photography were capturedsing the EyeScan device and the FF450 plus (Carl Zeiss, Inc., Saniego, CA), respectively. The retinal video recording was per-

ormed by a medical officer who received a full day of training insing the device. Retinal photography was performed by an expe-ienced photographer using FF450 plus.

All retinal video recordings commenced at the optic disc androceeded to the macula and temporal regions. To obtain continu-ty of retinal information between the regions, the retinal cameraas tilted at a consistent pace from left to right for the right eye

optic disc, macula, and temporal retina) for at least 5 seconds andice versa for the left eye (Videos 1, 2, and 3; available atttp://aaojournal.org). The video file format is in a standard audioideo interleave format, and the size of an uncompressed rawetinal video is approximately 1 gigabyte. Moreover, all retinalideos can be viewed with any media player that reads audio videonterleave format.

nterpretation of Retinal Photography and Retinalideosll retinal still images and videos were deidentified, randomized,

nd then interpreted by 2 senior consultant ophthalmologists (oph-halmologist 1, a consultant specialist with special interest iniabetes; ophthalmologist 2, a retinal specialist). The quality of theideos and retinal photography images were classified as unac-eptable, average, or excellent. The retinal video recordings wereraded as unacceptable by the ophthalmologists if they werelurred, out of focus, dark, had insufficient views (fewer than 5econds on any view), or a combination thereof. The retinalhotographs were graded as unacceptable if more than one third ofhe photograph was blurred.

Diabetic retinopathy was diagnosed based on the presence ofhe following: microaneurysms, dot or blot hemorrhages, cottonool spots, intraretinal microvascular abnormalities, venous bead-

ng, neovascularization, preretinal or vitreous hemorrhage, hardxudates, or macular edema.

Diabetic retinopathy and diabetic macular edema severity wereraded using the International Clinical Diabetic Retinopathy andacular Edema Disease Severity Scale (Table 1).22 This scale was

stablished by the American Academy of Ophthalmology in thelobal Diabetic Retinopathy Project with the aim to improve the

ommunication between the medical and nonmedical personnel,uch as ophthalmologists, diabetic specialists, primary care physi-
ians, and retinal photographers, by promoting the development of
1589

http://aaojournal.org

awencddr(

cachnrrr

C

Fvrm[worsrr9(tvrimbmrwm

p0

ME

D

Ophthalmology Volume 118, Number 8, August 2011

a common clinical severity scale for diabetic retinopathy.22 All theretinal photographs and retinal videos were viewed using the samemonitor (iMac 27 inches; Apple, Cupertino, CA) and VLC mediaplayer 1.1.4 (Mac; Apple).

Statistical Analyses

The main outcome measures investigated were the sensitivity andspecificity of the 2 methods of imaging (retinal video recordingand retinal photography) in detecting diabetic retinopathy gradingwith reference to the currently accepted gold standard, slit-lampbiomicroscopy examination. Technical failure rate, defined as thefraction of the unacceptable images or videos, was determined forboth photographic methods.

As a measurement of agreement, Cohen’s � coefficient23 wasobtained for diabetic retinopathy signs and grading between retinalvideo recording and retinal photography versus the gold standardslit-lamp biomicroscopy examination were calculated. All datawere analyzed using SPSS software version 17 (SPSS, Inc., Chi-cago, IL).

Results

A total of 100 patients (200 eyes) were enrolled in this study. Asummary of the patients’ demographic characteristics and diabeteshistory is shown in Table 2. The mean�standard deviation age ofparticipants was 52.8�15.1 years, the mean�standard deviationduration of diabetes was 13.7�9.7 years, and mean�standarddeviation glycosylated hemoglobin was 8.0�1.7%. Most of thepatients were white (75.0%; n � 75) and had type II diabetes(70.0%; n � 70).

Patient Clinical Characteristics

Most eyes (88%; n � 176) had best-corrected visual acuity of 6/6

Table 1. International Clinical Diabetic Retinopathy SeverityScales and International Clinical Diabetic Macular Edema

Disease Severity Scales22

Grades Retinal Findings

None No abnormalitiesMild NPDR Microaneurysms onlyModerate NPDR More than just microaneurysms, but

less than severe NPDRSevere NPDR Any of the following: (1) extensive

(�20) intraretinal hemorrhages ineach of 4 quadrants, (2) definitevenous beading in 2� quadrants,(3) prominent IRMA in 1�quadrant; AND no signs of PDR

PDR One or more of the following: (1)neovascularization and (2)vitreous/preretinal hemorrhage

DME apparently absent No apparent retinal thickening orhard exudates in posterior pole

DME apparently present Some apparent retinal thickening orhard exudates in posterior pole

DME � diabetic macular edema; IRMA � intraretinal microvascularabnormalities; NPDR � nonproliferative diabetic retinopathy; PDR �proliferative diabetic retinopathy.

or 6/9, 22 eyes (11%) had best-corrected visual acuity 6/12 to 6/36,H

1590

nd 2 eyes (1%) had best-corrected visual acuity 6/60. Cataractsere diagnosed in 18 eyes (9%) based on slit-lamp biomicroscopy

xamination findings. Of the 200 eyes, 70 eyes (35%) were diag-osed with diabetic retinopathy by clinical slit-lamp biomicros-opy. Of those diagnosed, 36 (18%) had mild nonproliferativeiabetic retinopathy, 23 (11.5%) had moderate nonproliferativeiabetic retinopathy, 2 (1%) had severe nonproliferative diabeticetinopathy, 5 (2.5%) had proliferative diabetic retinopathy, and 42%) had diabetic maculopathy.

Of the diabetes-related complications, more patients had mi-rovascular complications than macrovascular complications. Di-betic neuropathy (23%; n � 23) was the most common compli-ation, followed by diabetic nephropathy (22%; n � 22), ischemiceart disease (16%; n � 16), peripheral vascular disease (10%;� 10), and cerebrovascular disease (6%; n � 6). Of the 100

ecruited patients, 40% had never undergone any previous diabeticetinopathy screening, whereas 10% reported having previouslyeceived retinal laser therapy.

ross-source Agreement of Diagnostic Methods

or ophthalmologist 1, the sensitivity and specificity of the retinalideo imaging technique in detection of any grade of diabeticetinopathy, when compared with the gold standard slit-lamp bio-icroscopy examination, were 93.8% (95% confidence interval

CI], 84.2–98.0) and 99.2% (95% CI, 94.7–99.9), respectively,hereas retinal still photography had a sensitivity and specificityf 91.8% (95% CI, 81.1–96.9) and 98.4% (95% CI, 93.8–99.7),espectively (Table 3). For ophthalmologist 2, the sensitivity andpecificity in detection of any grade of diabetic retinopathy, witheference to gold standard slit-lamp examination, for retinal videoecording were 93.3% (95% CI, 83.0–97.8) and 95.2% (95% CI,1.6–99.0), whereas for retinal still photography, they were 92.1%95% CI, 81.7–97.0) and 96.8% (95% CI, 91.6–99.0), respec-ively. In detection of sight-threatening diabetic retinopathy (se-ere nonproliferative diabetic retinopathy, proliferative diabeticetinopathy), the sensitivity and specificity of retinal video record-ngs and retinal photography compared with the gold standardeasurement for both ophthalmologists were 100%. Moreover,

oth imaging techniques had 100% sensitivity in detecting diabeticaculopathy by the ophthalmologists, but the specificity of

etinal videos and retinal photography for ophthalmologist 1as 98.9% and 97.9%, respectively, whereas those for ophthal-ologists 2 were 98.9% and 99.4%, respectively.

The � coefficients for retinal video recordings and retinalhotography for all diabetic retinopathy grades were more than.90 (Table 3) based on the gold standard slit-lamp examination. In

Table 2. Patient Characteristics and Their Diabetes History

ean age�SD (yrs) 52.8 � 15.1thnicity, n (%)White 75 (75%)Asian 15 (15%)Indigenous 6 (6%)Other 4 (4%)

iabetes historyMean duration of diabetes�SD (yrs) 13.7 � 9.7Mean HbA1c�SD (%) 8.0 � 1.7Type 1 (insulin dependent) 60 (30%)Type 2

Non–insulin dependent 94 (47%)Insulin dependent 46 (23%)

bA1c � glycosylated hemoglobin; SD � standard deviation.

r7

s2igib

D

Tvsicctsdrbc

stsrifpfsgv

porps

stem,


detection of diabetic maculopathy, the � coefficient for retinalvideo recording and retinal still photography were 0.79 and 0.66(ophthalmologist 1) and 0.66 and 0.66 (ophthalmologist 2), respec-tively. Additionally, Cohen’s � coefficient for retinal video record-ings and retinal photography for all diabetic retinopathy signs(microaneurysms, retinal hemorrhages, venous beading, intrareti-nal microvascular abnormalities, new vessel formation, preretinalhemorrhage, hard exudates) and optic disc abnormalities all were0.80 or more, except for macular edema (Table 4). Of the captured200 eyes, the technical failure rate (the fraction of still images orretinal videos rated as unacceptable) by both ophthalmologists for

Table 3. Sensitivity, Specificity, and � CorrelatVideo Recording with Reference to

Sensitivity (95ConfidenceInterval)

Ophthalmologist 1Retinal still photography 91.8% (81.1–96Retinal video recording 93.8% (84.2–98

Ophthalmologist 2Retinal still photography 92.1% (81.7–97Retinal video recording 93.3% (83.0–97

Retinal photography was carried out using an FF450 Plout using the EyeScan device (Ophthalmic Imaging Sy

Table 4. � Statistics for Retinal Videos and RetinalPhotography by Both Consultant Ophthalmologists Comparedwith the Gold Standard Slit-Lamp Biomicroscopy Examination

Retinal findings

� Statistics

Retinal VideoRecording

Retinal StillPhotography

Ophthalmologist 1Microaneurysms 0.93 0.95Retinal hemorrhages 0.93 0.94Cotton wool spots 1.00 1.00Venous beading 0.91 0.89IRMA 1.00 0.85Preretinal hemorrhages 1.00 1.00New vessels formation 1.00 1.00Hard exudates 1.00 1.00Cupped optic disc 1.00 1.00Macular edema 0.80 0.66

Ophthalmologist 2Microaneurysms 0.9 0.93Retinal hemorrhages 0.9 0.92Cotton wool spots 1.00 1.00Venous beading 0.86 0.86IRMA 0.8 0.83Preretinal hemorrhages 1.00 1.00New vessels formation 1.00 1.00Hard exudates 1.00 1.00Cupped optic disc 1.00 1.00Macular edema 0.66 0.66

IRMA � intraretinal microvascular abnormalities.Retinal photography was carried out using an FF450 Plus (Carl Zeiss, Inc.),and retinal video recording was carried out using the EyeScan device

d(Ophthalmic Imaging System, Sacramento, CA).

etinal videos and retinal still photography were 7.5% (n � 15) and% (n � 14), respectively.

Of the 15 failed retinal videos, 7 eyes had grade III nuclearclerotic cataract based on the Lens Opacities Classification III,24

eyes were from a darkly pigmented patient, and 6 eyes werentolerant to bright light. In contrast, the 14 failed retinal photo-raphs were the result of cataracts (5 eyes), blurriness of themages secondary to eye movement (4 eyes), and intolerance toright flash (5 eyes).

iscussion

his study demonstrated the possibility of using retinalideo recording as an alternative diabetic retinopathycreening technique, given that the sensitivity and specific-ty of retinal video recording were comparable with those ofolor retinal photography with reference to slit-lamp biomi-roscopy examination in determining the diabetic retinopa-hy grading (Table 3). Moreover, retinal video also pos-essed a comparable � coefficient to retinal photography inetecting diabetic retinopathy signs such as microaneu-ysms, retinal hemorrhages, cotton wool spots, venouseading, and intraretinal microvascular abnormalities whenompared with the slit-lamp examination.

Whereas both imaging devices demonstrated excellentensitivity and specificity in detecting diabetic maculopa-hy, the sample size of patients with macular edema in thistudy (n � 4) was small. Given that the retinal videos andetinal photography only offer 2-dimensional retinal views,t will be impossible to detect any signs of retinal thickeningrom a retinal photograph or video alone. Nevertheless, theresence of hard exudates or microaneurysms close to theovea and an unexplained drop in a patient’s visual acuityhould prompt an urgent referral to an ophthalmologist,iven that diabetic maculopathy could cause significantisual impairment in patients with diabetes.

In this study, the retinal video recording technique waserformed by a nonexperienced medical officer (no previ-us ocular imaging experience), whereas the retinal photog-aphy was performed by an experienced retinal photogra-her (an orthoptist who had more than 10 years of retinaltill photography for diabetic retinopathy experience, but

oefficient for Retinal Photography and RetinalLamp Biomicroscopy Examination

Specificity (95%ConfidenceInterval)

�Correlation

98.4% (93.8–99.7) 0.9399.2% (94.7–99.9) 0.95

96.8% (91.6–99.0) 0.9195.2% (89.4–98.0) 0.90

arl Zeiss, Inc.), and retinal video recording was carriedSacramento, CA).

ion CSlit-

%

.9)

.0)

.0)

.8)

us (C

oes not have any ophthalmic certification). The difference

1591

nrawirta

vruatbtewqar

R

1

1

Ophthalmology Volume 118, Number 8, August 2011

of technical failure rate between the video recording and theretinal photography was not statistically significant (7.5%vs. 7%; chi-square, 0.04; degrees of freedom, 1; P � 0.85),suggesting that the retinal video recording is a potentiallyeasy-to-operate technique. Thus, nonexperienced personnelsuch as general practitioners, nurses, allied health workers,or any volunteer could be trained to screen for diabeticretinopathy within a short period (1- to 2-day training ses-sions), and this may increase the diabetic retinopathyscreening rates in the community, particularly in developingcountries. Nonetheless, the ease of use for this techniquewas based on the technical failure rate and subjective ex-perience of a sole user in this study, and hence, furtherstudies will need to recruit both experienced and nonexpe-rienced personnel to assess and compare widely the usabil-ity of this technique of image acquisition with standardretinal still photography in the setting of diabetic retinopa-thy screening.

Of the 15 failed retinal videos, 6 eyes were the resultof intolerance to bright light. In this study, the retinalvideo recordings were obtained after retinal photography,which might have resulted in the irritation or exhaustionof patients’ eyes during the retinal video recording pro-cess. Given that the retinal video recording technique hasbeen used only for a short period, the failed videos alsoin part may be the result of the operator’s technique.Hence, the technical failure rate may be diminished fur-ther with improvement of the operator’s technique overtime. Of the failed retinal photographs, they were sec-ondary to eye movement (4 eyes) and intolerance tobright flash (5 eyes). Retinal photography always re-quires patient compliance to fixate the eye during theprocedure to avoid image blurriness, whereas retinalvideo recordings could tolerate slightly more eye move-ment. This is particularly useful for patients who cannotfixate their eyes to the external or internal fixator of theretinal camera.

Compared with retinal photography, retinal video record-ing carries some disadvantages. It does not possess a flashfunction to increase briefly the light intensity. As a result,the light intensity required throughout the entire retinalvideo recordings will be slightly higher, but less than theintensity of the flash, for the period of recording, especiallyin patients with dark fundi. Additionally, the videos savedas audio video interleave format need relatively high storagecapacity, because 1 second in the video takes up approxi-mately 20 megabytes. Depending on patient compliance andtolerance, an average retinal video recording will take atleast 30 seconds, and thus, a large storage capacity isrequired to accommodate the size of the videos. To incor-porate the retinal video recording as a future teleretinalscreening tool, future research will need to be conducted onthe development of web-based teleophthalmology software.The optimal or maximum compression level for retinalvideos is needed without compromising their quality anddiagnostic accuracy in detecting diabetic retinopathychanges to maximize the storage capacity and data trans-mission speed.

This study used a high-end computer with a large reading
monitor (iMac 27-inch screen; Apple) to avoid any diag-
1592

ostic error resulting from a small screen and low screenesolution. Given that the iMac 27-inch computer is a rel-tively sophisticated and expensive device, further studiesill be valuable to evaluate the use of other more econom-

cal devices (with smaller screen sizes and lower screenesolution) to interpret diabetic retinopathy changes fromhe retinal videos, particularly for areas with finite resourcesnd financial constraints.

In conclusion, this study demonstrated that the retinalideo recording was equally as effective as retinal photog-aphy in detecting diabetic retinopathy on the subjects eval-ated in this study. It is a novel alternative diabetic retinop-thy screening technique that is easy to learn with minimalraining by nonexperienced personnel. Although it will note a substitute for the comprehensive ophthalmic examina-ion for diabetic patients, it offers primary eye care provid-rs the opportunity to view a greater field of retinal viewithin a short period. Nonetheless, more research is re-uired to explore its user friendliness, cost effectiveness,nd clinical effectiveness in a community setting of diabeticetinopathy screening.

eferences

1. Moss SE, Klein R, Klein BE. The 14-year incidence of visual loss ina diabetic population. Ophthalmology 1998;105:998–1003.

2. Wild S, Roglic G, Green A, et al. Global prevalence ofdiabetes: estimates for the year 2000 and projections for 2030.Diabetes Care 2004;27:1047–53.

3. Tapp RJ, Shaw JE, Harper CA, et al, AusDiab Study Group. Theprevalence of and factors associated with diabetic retinopathy in theAustralian population. Diabetes Care 2003;26:1731–7.

4. Klein R, Klein BE, Moss SE, et al. The Wisconsin Epidemi-ologic Study of Diabetic Retinopathy. III. Prevalence and riskof diabetic retinopathy when age at diagnosis is 30 or moreyears. Arch Ophthalmol 1984;102:527–32.

5. Klein R, Klein BE, Moss SE, et al. The Wisconsin Epidemi-ologic Study of Diabetic Retinopathy. II. Prevalence and riskof diabetic retinopathy when age at diagnosis is less than 30years. Arch Ophthalmol 1984;102:520–6.

6. Ferris FL III. How effective are treatments for diabetic reti-nopathy? JAMA 1993;269:1290–1.

7. Javitt JC, Aiello LP, Chiang Y, et al. Preventive eye care in peoplewith diabetes is cost-saving to the federal government: implicationsfor health-care reform. Diabetes Care 1994;17:909–17.

8. Javitt JC, Aiello LP, Bassi LJ, et al. Detecting and treatingretinopathy in patients with type I diabetes mellitus: savingsassociated with improved implementation of current guide-lines. American Academy of Ophthalmology. Ophthalmology1991;98:1565–73; discussion 1574.

9. Javitt JC, Aiello LP. Cost-effectiveness of detecting andtreating diabetic retinopathy. Ann Intern Med 1996;124:164 –9.

0. Bursell SE, Cavallerano JD, Cavallerano AA, et al, JoslinVision Network Research Team. Stereo nonmydriatic digital-video color retinal imaging compared with Early TreatmentDiabetic Retinopathy Study seven standard field 35-mm stereocolor photos for determining level of diabetic retinopathy.Ophthalmology 2001;108:572–85.

1. Conlin PR, Fisch BM, Orcutt JC, et al. Framework for anational teleretinal imaging program to screen for diabeticretinopathy in Veterans Health Administration patients. J Re-
habil Res Dev 2006;43:741–8.

1

1

2

2

2

2

2


12. Cavallerano AA, Cavallerano JD, Katalinic P, et al, Joslin VisionNetwork Research Team. A telemedicine program for diabeticretinopathy in a Veterans Affairs Medical Center—the JoslinVision Network Eye Health Care Model. Am J Ophtthalmol2005;139:597–604.

13. National Screening Programme for Diabetic Retinopathy. Availableat: http://www.retinalscreening.nhs.uk/userFiles/File/EyeScreeningForDiabetes.pdf. Accessed March 16, 2011.

14. Kumar S, Tay-Kearney ML, Constable IJ, Yogesan K. Internetbased ophthalmology service: impact assessment [letter]. Br JOphthalmol 2005;89:1382–3.

15. Kumar S, Tay-Kearney ML, Chaves F, et al. Remote ophthal-mology services: cost comparison of telemedicine and alternativeservice delivery options. J Telemed Telecare 2006;12:19–22.

16. Early Treatment Diabetic Retinopathy Study ResearchGroup. Grading diabetic retinopathy from stereoscopiccolor fundus photographs—an extension of the modifiedAirlie House classification. ETDRS report number 10. Oph-thalmology 1991;98:786 – 806.

17. Williams GA, Scott IU, Haller JA, et al, Ophthalmic Tech-nology Assessment Committee Retina Panel. Single-field fun-dus photography for diabetic retinopathy screening: a reportby the American Academy of Ophthalmology. Ophthalmol-
ogy 2004;111:1055–62.
Footnotes and Financial Disclosures

University of Western Australia, Crawley, Australia.

PM

FTY

Ss

CDVA

8. Hutchinson A, McIntosh A, Peters J, et al. Effectiveness ofscreening and monitoring tests for diabetic retinopathy—asystematic review. Diabet Med 2000;17:495–506.

9. Aptel F, Denis P, Rouberol F, Thivolet C. Screening of dia-betic retinopathy: effect of field number and mydriasis onsensitivity and specificity of digital fundus photography. Di-abetes Metab 2008;34:290–3.

0. Harding SP, Broadbent DM, Neoh C, et al. Sensitivity andspecificity of photography and direct ophthalmoscopy inscreening for sight threatening eye disease: the LiverpoolDiabetic Eye Study. BMJ 1995;311:1131–5.

1. Olson JA, Strachan FM, Hipwell JH, et al. A comparativeevaluation of digital imaging, retinal photography and optom-etrist examination in screening for diabetic retinopathy. DiabetMed 2003;20:528–34.

2. Wilkinson CP, Ferris FL III, Klein RE, et al, Global Dia-betic Retinopathy Project Group. Proposed internationalclinical diabetic retinopathy and diabetic macular edemadisease severity scales. Ophthalmology 2003;110:1677– 82.

3. Cohen J. A coefficient of agreement for nominal scales. EducPsychol Meas 1960;20:37–46.

4. Chylack LT Jr, Wolfe JK, Singer DM, et al, LongitudinalStudy of Cataract Study Group. The Lens Opacities Classifi-
cation System III. Arch Ophthalmol 1993;111:831–6.
Originally received: September 30, 2010.Final revision: March 16, 2011.Accepted: April 5, 2011.Available online: June 19, 2011. Manuscript no. 2010-1357.1 Center for Ophthalmology and Visual Sciences, Lions Eye Institute,University of Western Australia, Nedlands, Australia.2 The Australian E-Health Research Center, Commonwealth ScientificIndustrial Research Organization (CSIRO), Floreat, Australia.3 Ophthalmology Department, Royal Perth Hospital, Perth, Australia.4 Center for Health Services Research, School of Population Health, The

resented as a poster at: American Academy of Ophthalmology Annualeeting, October 2010, Chicago, Illinois.

inancial Disclosure(s):he author(s) have made the following disclosure(s):ogesan Kanagasingam - Inventor - EyeScan device

upported by Diabetes Australia and Royal Perth Hospital Medical Re-earch Foundation, Perth, Western Australia.

orrespondence:aniel Shu Wei S. Ting, MBBS(Hons), Center for Ophthalmology andisual Sciences, Lions Eye Institute, 2 Verdun Street, Nedlands, Western
ustralia, Australia 6009. E-mail: [email protected].
1593

http://www.retinalscreening.nhs.uk/userFiles/File/EyeScreeningForDiabetes.pdf

http://www.retinalscreening.nhs.uk/userFiles/File/EyeScreeningForDiabetes.pdf


Retinal videorecordings atdifferentcompression levels:a novel video-basedimaging technologyfor diabeticretinopathyscreening

DSW Ting1 ;2 ;3, ML Tay-Kearney1 ;3,

I Constable1, J Vignarajan2 and

Y Kanagasingam2

Abstract

Background To evaluate the optimal

compression level of retinal color digital

video recordings, a novel video-based

imaging technology, in screening for diabetic

retinopathy (DR).

Design Evaluation of a diagnostic

technique.

Methods A total of 36 retinal videos,

captured using EyeScan (Ophthalmic

Imaging System), were compressed from

original uncompressed file size of 1 GB

(gigabyte) to four different compression

levels—100 MB (megabyte) (Group 1); 30 MB

(Group 2); 20 MB (Group 3); and 5 MB

(Group 4). The videos were subsequently

interpreted by an ophthalmologist and a

resident using the International Clinical

Diabetic Retinopathy Severity Scales.

Main outcome measures The sensitivity,

specificity and j coefficient for DR grading

detected by were calculated for each

compression level (Groups 1–4), with

reference to the original uncompressed

retinal videos.

Results Groups 1, 2, and 3 graded by both

readers had sensitivity and specificity 490%

in detecting DR, whereas for group 4, the

sensitivity and specificity were 70.6% and

94.7% for ophthalmologist and 80.0% and

72.2% medical officer, respectively. The j

correlation in detecting DR for groups 1, 2,

and 3 were 40.95, whereas for Group 4, the j

was 0.76 and 0.66 for ophthalmologist and

medical officer, respectively.

Conclusion Retinal video recording is a

novel and effective DR screening technique

with high sensitivity, specificity and j

correlation. With its compressibility, this is a

potential effective technique that can be

widely implemented in a routine, mobile,

and tele-ophthalmology setting for DR

screening services.

Eye (2013) 27, 848–853; doi:10.1038/eye.2013.53;

published online 10 May 2013

Keywords: retinal video; compression; diabetic

retinopathy; screening

Introduction

Diabetes mellitus is a metabolic disorder

characterized by hyperglycemia secondary to

either impaired insulin secretion or insulin

resistance. The chronic uncontrolled

hyperglycemia can give rise to macrovascular

(cerebrovascular accident, ischemic heart

disease, and peripheral vascular disease) and

microvascular (retinopathy, nephropathy, and

neuropathy) complications. Diabetic

retinopathy (DR) is one of the commonest

microvascular complications of diabetes. Nearly

100% type 1 diabetic patients and 460% type 2

diabetic patients will have at least some

retinopathy after 20 years of diabetes.1,2

Therefore, it is crucial for primary eye care

providers to regularly screen people with

diabetes for DR, as early detection can prevent

severe visual impairment.3

1Center for Ophthalmologyand Visual Sciences, LionsEye Institute, University ofWestern Australia,Nedlands, WesternAustralia, Australia

2The Australian E-HealthResearch Center,Commonwealth ScientificIndustrial ResearchOrganization (CSIRO),Floreat, Western Australia,Australia

3Department ofOphthalmology, Royal PerthHospital, Perth, WesternAustralia, Australia

Correspondence:DSW Ting, Center ofOphthalmology and VisualScience, Lions Eye Institute,University of WesternAustralia, 2 Verdun Street,Perth, 6009, WesternAustralia, Australia.Tel: +6585282252;Fax: +6562209567;E-mail: [email protected]

Received: 2 March 2012Accepted in revised form:24 June 2012Published online: 10 May2013

CL

INIC

AL

ST

UD

Y

Eye (2013) 27, 848–853& 2013 Macmillan Publishers Limited All rights reserved 0950-222X/13

www.nature.com/eye

http://dx.doi.org/10.1038/eye.2013.53



http://www.nature.com/eye

Traditionally, retinal still photography has been the

gold standard DR screening tool in the primary

health-care setting. However, the video-based imaging

technology using retinal video recording has been

recently proposed to be a novel DR screening method.4

This technique is quick to perform, easy to learn

(single-day training) and does not require any previous

ophthalmic imaging experience. Compared with the

retinal still photography, the retinal video not only

provides a greater field within shorter period of time but

also mimics what is seen with a slit lamp examination.

Nevertheless, this technique was limited by its large

video file size, which requires high storage capacity.4

On average, a retinal video of 30 s takes up B500 MB

and thus, will be impractical to archive and transmit

such large video files during routine DR screenings.

The objective of our study is to investigate and

determine the optimal compression level for retinal

videos to screen for DR. The effect of retinal still image

compression for DR screening has been studied

previously5–7, but none of the which has investigated the

use of retinal videos compression technique.

Compression of retinal videos will help to reduce the

need for a large storage capacity and to increase the data

transmission speed for DR screening in a routine, mobile

and tele-ophthalmology setting.

Materials and methods

Sample population

A total of 36 retinal videos (18 normal and 18 with DR

changes) at four different compression levels—100 MB

(Group 1), 30 MB (Group 2), 20 MB (Group 3), and 5 MB

(Group 4), captured by retinal video recording using OIS

EyeScan (Ophthalmic Imaging System, Sacramento, CA,

USA)—were selected for our study. Given that three-field

(optic disc, macula, and temporal views) retinal still

photogaphy has been the standard of care for patients

with diabetes in our center, the retinal video recordings

were also captured in a similar manner for the recruited

patients in our study. All study subjects were enrolled

from the DR Screening Clinic of Royal Perth Hospital and

this study has been approved by the Royal Perth

Hospital Human Research Ethics Committee.


To allow for a power of 95%, desired precision of 0.10,

expected sensitivity and specificity of 96%, the total

number of eyes required for each compression level was

31 (prevalence was set at 0.50 as selected samples

consisted of 50% normal and 50% abnormal retinal

digital videos).

Conversion process

A digital video can exist in various different file formats

such as moving picture experts group (MPEG), audio

video interleave (AVI), windows media video (WMV),

QuickTime Movie File (MOVIE), and, so on. It consists of

a series of bitmap digital images displayed in a rapid

succession at a constant rate. The size of a video is

determined by bit rates (BRs) and time (T). BR, defined as

the number of bits processed or conveyed per unit time,

represents the amount of information stored per unit of

time of a recording. It is determined by the frame rate

(FR), frame size (FS) and color depth (CD) of a video. The

FR is defined as the number of displayed digital images

per unit time and it is often measured in a second (frames

per second; FPS). The FS is the total number of pixels in

terms of width (W) and height (H) of an image. The CD

represents the amount of bits that form a single pixel. By

altering one of these properties, one can modify the size

of a digital video using various video compression

software packages and codecs, which are readily

available on the internet.

For our study, the uncompressed raw retinal videos

were compressed to four different levels (Groups 1–3, 4)

from the original file size using a video converter

software, Xilisoft Video Converter Ultimate 6.0 (Xilisoft

Corporation, British Virgin Island), which utilizes a

standard video codec H.264 (Table 1). The average size of

Table 1 The file size of a retinal video with different compression levels by reducing its bit rate while keeping other parametersconstant (frame rate, frame size, and zoom)

Groups Bit rate (kbps) Approx file size for 60 S Compression levelPercentage of file size from

its original size

Original uncompressed raw video 165,000 1 GB — 100%1 15,000 100 MB 90% 10%2 5000 30 MB 97% 3%3 3000 20 MB 98% 2%4 512 5 MB 99% 1%

Abbreviations: GB, gigabytes; MB, megabytes.

The setting of other parameters: 1) frame rate: 17 frames per second; (2) frame size: 640� 480 pixels; (3) Zoom: full.

Retinal Video compression for diabetic retinopathy screeningDSW Ting et al

849

Eye

an original uncompressed 30-s video was 500 MB. In our

study, we reduced the 500-MB video file to four different

levels (Group 1: 100 MB, Group 2: 30 MB, Group 3:

20 MB, and Group 4: 5 MB) by changing the BRs. The

FR and FS were set at 17 FPS and 640� 480 pixels,

respectively, as these settings have been preset by the

fundus camera (OIS EyeScan).

Reference standard

We utilized the raw uncompressed raw retinal videos as

the reference of our study as from our previous study,4 the

sensitivity and specificity of an uncompressed raw retinal

videos in detecting any grade of DR by a retinal specialist

and a consultant ophthalmologist with special interest in

diabetes, compared with the slit lamp examination by a

senior consultant ophthalmologist (with 410 years

experience in screening DR), were 490%.

Data interpretation

The converted videos were randomized and sent to two

different readers (a consultant ophthalmologist and a

medical officer) who were blinded to the compression

level of the videos. Each reader interpreted a total of 144

retinal videos (36 retinal videos at four different

compression levels) using a standard monitor screen

(iMac 27’, Apple, Cupertino, CA, USA) with a VLC

media player 1.1.4 (Apple) in a dimly lit room. The

International Clinical DR Severity Scales8 was utilized to

interpret and grade the DR severity by determining the

presence/absence of lesions including microaneuryms,

retinal hemorrhages, cotton wool spots, venous beading,

intraretinal microvascular abnormalities, new vessels

formation, vitreous hemorrhage, preretinal hemorrhage,

and hard exudates. In addition, the retinal videos were

rated as ‘acceptable’, ‘pixelated but interpretable’ and

‘unacceptable’ by the readers.

Data analysis

The data analysis was performed using SPSS version 17

(SPSS, Chicago, IL, USA). The main outcome measures

were the sensitivity, specificity, and k coefficient between

the uncompressed raw retinal videos and compressed

retinal videos in diagnosing any level of DR. In addition,

the k correlation was performed for other variables, such

as (1) the microaneurysms and retinal hemorrhages

between the uncompressed and compressed retinal

videos and; (2) the DR grading between the consultant

ophthalmologist and medical officer using the

uncompressed retinal videos. Values of k of 0.8 and

above were considered as excellent agreement between

two groups for our study.9

Results

All retinal videos in Group 1 (100 MB), 2 (30 MB),

and 3 (20 MB) were rated as ‘acceptable’ by the

ophthalmologist and medical officer (Table 2). For Group

4 (10 MB), only 11% and 3% of the retinal videos were

rated as ‘acceptable’ by the ophthalmologist and medical

officer, respectively. Of the 18 retinal videos with DR,

22% (n¼ 4) had mild non-proliferative DR (NPDR), 61%

(n¼ 11) had moderate NPDR, 11% (n¼ 2) had severe

NPDR, and 6% (n¼ 1) had proliferative DR.

The conversion time for the retinal videos to different

compression levels are shown in Table 3. Using the

uncompressed videos as gold standard, Group 1, 2, and 3

graded by both readers had excellent sensitivity and

specificity of 490% in detecting DR (Table 4). On the

other hand, Group 4 had the lowest sensitivity and

specificity in detecting DR changes (ophthalmologist 1—

Table 2 The quality of retinal videos (with/without diabeticretinopathy changes) of different compression levels rated by anophthalmologist and a medical officer

UninterpretablePixelated butinterpretable Acceptable

OphthalmologistGroup 1(100 MB)

0 (0%) 0 (0%) 36 (100%)

Group 2(30 MB)

0 (0%) 0 (0%) 36 (100%)

Group 3(20 MB)

0 (0%) 0 (0%) 36 (100%)

Group 4(5 MB)

3 (8.3%) 29 (80.6%) 4 (11.1%)

Medical officerGroup 1(100 MB)

0 (0%) 0 (0%) 36 (100%)

Group 2(30 MB)

0 (0%) 0 (0%) 36 (100%)

Group 3(20 MB)

0% 0% 36 (100%)

Group 4(5 MB)

3 (8.3%) 32 (88.9%) 1 (2.8%)

Abbreviation: MB, megabytes.

Table 3 The average conversion timing of an uncompressedraw retinal video (1 GB) to different compression levels

Compression level Timings (s)

Group 1 (100 MB) 25Group 2 (30 MB) 18Group 3 (20 MB) 17Group 4 (5 MB) 16


The compression was perfomed using a 32-bit Windows XP, 2.5 RAM,

Intel Xeon X5550 Processor 2.67 GHz with NVIDIA Quadro FX 580

graphics card (Timings may vary on different computers with different

performance).


850

Eye

sensitivity: 70.6%, specificity: 94.7%; medical officer:

sensitivity: 80.0%; and specificity: 72.2%).

Also, the k between the gold standard uncompressed

videos and Groups 1, 2, and 3 to detect DR-related

changes for both readers were 40.96, whereas for Group

4, the k for the ophthalmologist and medical officer were

0.76 and 0.66, respectively. On the other hand, the kbetween the ophthalmologist and medical officer for raw

uncompressed videos, Group 1, Group 2, Group 3, and

Group 4 were 0.84, 0.84, 0.80, 0.76, and 0.43, respectively.

Similarly, the k correlation of the microaneurysms, retinal

hemorrhages, detected by both readers for Groups 1, 2,

and 3 with reference to the gold standard video files

(1 GB) were 40.9 (Table 5). For Group 4, the k for the

detection of microaneurysms by the ophthalmologist and

medical officer were both 0.87, whereas for retinal

hemorrhages, they were 0.88 and 0.87, respectively. The kcorrelation of new vessel formation and subhyaloid

hemorrhage was 1.0 in groups 1–4 graded by both

ophthalmologist and medical officer, with reference to

the uncompressed videos.

Discussion

Our study showed that the retinal videos could be

significantly compressed down to 20 MB of its original

file size with excellent sensitivity (ophthalmologist:

94.4%; medical officer: 100%) and specificity

(ophthalmologist: 100%; medical officer: 93.8%) in

detecting DR changes. The file size of a compressed 30-s

retinal video consisting of optic disc, macula, and temporal

views is B20 MB. For an uncompressed color fundus

photo captured by FF450 plus (Carl Zeiss Meditec, Inc.,

Dublin, CA, USA) in the tagged image file format (TIFF) or

BIT MAP format, each photo takes up B13 MB. In other

words, a total of 39 MB is required for a three-field color

retinal still photography. In comparison, a compressed

video will have a lower storage capacity compared with

the three uncompressed raw color fundus photos.

For Group 4 (5 MB), most of the retinal videos were

rated as ‘pixelated but interpretable’ by both readers

(Table 2). Nonetheless, they did not possess comparable

sensitivity and specificity with other three groups

(Table 4). In addition, the feedback from the

ophthalmologist and medical officer were similar.

Although reading the video recordings from groups 1–3

was comfortable, interpreting the ‘pixelated’ videos was

‘extremely tiring’ and ‘time consuming’, as more time

was required to differentiate a true lesion from the

normal retina and most of them were very ‘disjointed’

and ‘blurred’. Hence, this compression level will not be

ideal in the setting of diabetic retinopathy screening.

To the authors’ knowledge, no data were published on

retinal digital video recording for DR screening. Given that

this is one of the first studies that evaluate the effectiveness

of retinal video recordings at different compression levels,

we selected the uncompressed raw retinal videos, which

are considered to be of ‘good’ quality. Our results indicated

that a retinal video can be compressed down to 20 MB

without compromising its quality and diagnostic accuracy.

This study will need to be further expanded on retinal

videos with varying quality, especially ones with

compromised quality due to media opacities and dark

fundi, to assess the effects of video compression.

The medical officer in this study had interpreted

41000 color fundus photos of diabetic patients before

this study. The difference in the detection of DR between

the ophthalmologist and medical officer was minimal

(Table 4). These results indicated that the compressed

retinal videos could be potentially interpreted by trained

Table 4 The sensitivity, specificity, and k correlation ofdifferent compression levels for retinal videos in detecting DRgrading by an ophthalmologist and a medical officer withreference to uncompressed raw retinal videos (1 GB)

Sensitivity (%)(95% CI)

Specificity (%)(95% CI)

kcorrelation

OphthalmologistGroup 1(100 MB)

100 (78.1–100) 100 (78.1–100) 1.00

Group 2 (30 MB) 100 (77.1–100) 94.7 (71.9–99.7) 0.96Group 3 (20 MB) 94.4 (70.6–99.7) 100 (78.1–100) 0.96Group 4 (5 MB) 70.6 (44.0–88.6) 94.7 (71.9–99.7) 0.76

Medical officerGroup 1(100 MB)

100 (80.0–100) 100 (75.9–100) 1.00

Group 2 (30 MB) 100 (80.0–100) 100 (75.90100) 1.00Group 3 (20 MB) 100 (80.0–100) 93.8 (67.7–99.7) 0.96Group 4 (5 MB) 80.0 (51.4–94.7) 72.2 (46.4–89.3) 0.66

Abbreviations: CI, confidence interval; DR, diabetic relinopathy; GB,

gigabytes; MB, megabytes.

Table 5 The k correlation of the microaneurysms and retinalhemorrhages detected by both ophthalmologist and medicalofficer at different compression levels with reference to theuncompressed raw video files (1 GB)

Ophthalmologist Microaneurysms Retinal hemorrhages

Group 1 (100 MB) 1.00 1.00Group 2 (30 MB) 1.00 1.00Group 3 (20 MB) 1.00 1.00Group 4 (5 MB) 0.87 0.87

Medical officerGroup 1 (100 MB) 1.00 1.00Group 2 (30 MB) 1.00 1.00Group 3 (20 MB) 0.94 1.00Group 4 (5 MB) 0.88 0.87



851

Eye

non-ophthalmologist personnel and thus, the consultant

specialist input can be redirected to more useful areas

such as provision of consultations and surgical

intervention for patients with sight-threatening DR.

Only one retinal video was graded differently to

the uncompressed recording by the consultant

ophthalmologist in Groups 2 and 3 and medical officer in

Group 3. Owing to the relatively small sample size, this

has significantly reduced the sensitivity and specificity of

detecting DR grading by the consultant ophthalmologist

(specificity of 94.6% in Group 2 and sensitivity of 94.4% in

Group 3) and medical officer (specificity of 93.8% in

Group 3) (Table 4). Despite fulfilling the sample size

estimation, we felt that this was still one of the weaknesses

of our study and thus, further research with a larger

sample size will be of great value to explore the diagnostic

accuracy of video compression at 30 and 20 MB, and other

compression levels such as 15 and 10 MB.

In detecting DR changes for Groups 3 (20 MB) and 4

(5 MB), the sensitivity of the medical officer in detecting

DR changes was higher than the ophthalmologist, but in

contrast, the ophthalmologist had a higher specificity

than the medical officer (Table 4). The difference in

diagnostic sensitivity and specificity indicated that the

medical officer had a lower threshold in diagnosing a

patient with DR compared with the ophthalmologist,

especially in cases where he was uncertain about a

suspicious lesion. Although false-positive diagnoses

may result in more ‘unnecessary’ referrals to an

ophthalmologist, it is critical to note that a screener

should always refer the patients with a suspicious lesion

to an ophthalmologist to avoid any misdiagnosis of a

sight-threatening condition.

The Xilisoft Video Converter Ultimate 6.0 utilizes a

standard video codec H.264 which is compatible for both

Windows and Macintosh. The retinal video recording

device (OIS EyeScan) does not possess a built-in retinal

video compression program, which performs different

compression levels for the retinal videos. Given that the

OIS EyeScan is one of the first fundus cameras to perform

color retinal video recording, future research could

be conducted for OIS EyeScan or other devices to

incorporate a built-in retinal video compression program

to further shorten the process of converting a retinal video.

Depending on each compression level, the conversion

time for a retinal video from an uncompressed format

(500 MB) requires 16–25 s using a 32-bit Windows XP (2.5

RAM, Intel Xeon X5550 Processor 2.67 GHz with NVIDIA

Quadro FX 580 graphics card) (Table 3). This timing may

vary slightly on different computers with different

performance. This is a rapid conversion process as nearly

250 retinal videos can be converted within an hour. As

the retinal videos can be readily compressible to be

interpreted and stored, the retinal video recording

technique can be potentially utilized as a tele-

ophthalmology screening tool. It will be of great value to

conduct further research to evaluate the cost and clinical

effectiveness of performing tele-retinal video recording

for DR screening in the remote underserved areas.

In conclusion, a retinal video recording can be

compressed from 500 to 20 MB while maintaining its

quality, sensitivity, and specificity of DR grading. As

network bandwidth is an issue in most of the rural and

remote locations, the retinal video transmission speed

could be increased by reducing the file size without

significant loss of diagnostic information using efficient

compression techniques. Given that, the retinal videos

are easily compressible while retaining excellent

diagnostic accuracy, retinal video recording may be used

as an alternative technique in DR screening centers. As

this technique is still in its infancy, more research is

required to improve the usability of this technique.

Summary

What was known before

K Retinal video recording has been recently proposed to bea novel diabetic retinopathy screening method. It is quickto perform and yields excellent sensitivity and specificityin detection of any grade of diabetic retinopathy.

What this study adds

K This study shows that the retinal video recording can becompressed down to 20 MB without compromising itssensitivity and specificity in detection of any grade ofdiabetic retinopathy. It enables rapid transmission of datavia internet and also much less requirement for datastorage. Hence, retinal video recording can be potentiallyutilized as a routine, mobile, and tele-ophthalmologyscreening method.

Conflict of interest

The authors declare no conflict of interest.

Acknowledgements

Diabetes Australia Research Trust and Royal Perth

Hospital Medical Research Foundation provided the

research funding for this study. The sponsor or funding

organization had no role in the design or conduct of this

research.

Authors contributions

DSW Ting contributed to the study conception and

design, data acquisition, data analysis, data

interpretation and drafting the article. ML Tay-Kearney

contributed to the study conception and design,


852

Eye

revision of the important intellectual content, and final

approval of the version to be published. J Vignarajan

contributed to the data acquisition and data analysis.

Y Kanagasingam contributed to the study conception

and design, revision of the important intellectual content,

and final approval of the version to be published.

References

1 Williams R, Airey M, Baxter H, Forrester J, Kennedy-Martin T,Girach A et al. Epidemiology of diabetic retinopathy andmacular oedema: a systematic review. Eye 2004; 18: 963–983.

2 Mohamed Q, Gillies MC, Wong TY. Management of diabeticretinopathy: a systematic review. JAMA 2007; 298: 902–916.

3 Ferris 3rd, FL. How effective are treatments for diabeticretinopathy? JAMA 1993; 269: 1290–1291.

4 Ting DSW, Tay-Kearney ML, Constable IJ, Lim L, Preen DB,Kanagasingam Y et al. Retinal video recording: a new way to

image and diagnose diabetic retinopathy. Ophthalmology

2011; 118: 1588–1593.5 Newsom RS, Clover A, Costen MT, Sadler J, Newton J, Luff

AJ et al. Effect of digital image compression on screening for

diabetic retinopathy. Br J Ophthalmol 2001; 85: 799–802.6 Eikelboom RH, Yogesan K, Barry CJ, Constable IJ,

Tay-Kearney ML, Jitskaia L et al. Methods and limits of

digital image compression of retinal images for telemedicine.

Invest Ophthalmol Vis Sci 2000; 41: 1916–1924.7 Conrath J, Erginay A, Giorgi R, Lecleire-Collet A, Vicaut E,

Klein JC et al. Evaluation of the effect of JPEG and JPEG2000

image compression on the detection of diabetic retinopathy.

Eye 2007; 21: 487–493.8 Wilkinson CP, Ferris 3rd FL, Klein RE, Lee PP, Agardh CD,

Davis M et al. Proposed international clinical diabetic

retinopathy and diabetic macular edema disease severity

scales. Ophthalmology 2003; 110: 1677–1682.9 Altman DG. Practical Statistics for Medical Reserach.

Chapman and Hall: London, UK, 1991.


853

Eye

Documents

Diabetic Retinopathy Screening: Current Screening Practices and … · 1" " Diabetic Retinopathy Screening: Current Screening Practices and Novel Retinal Imaging Modalities By Daniel