Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
Welcome to the Centre for Biostatistics Showcase
Weds 3rd May 2017
Programme
13:40 Prof Richard Emsley “The Clinical Trials Unit”
13:50 Dr Jamie Sergeant “The Manchester Musculoskeletal Biomedical Research Unit”
14:00 Dr Sarah Cotterill “Working with the NIHR CLAHRC Greater Manchester”
14:10 “What we do” Session 1:
Matthew Gittins, Liz Howarth, Jack Wilkinson, Calvin Heal
14:30 Tea and Coffee
Programme
14:50 Prof Andy Vail “The Biostatistics Collaboration Unit”
15:00 Dr David Reeves “The National School of Primary Care Research”
15:10 Prof Carlo Berzuini “Causal inference and its use in biomedical research”
15:20 Dr Matthias Pierce “The North West Hub for Trials Methodology Research”
15:30 Sarah Rhodes “The Research Design Service”
15:40 “What we do” Session 2:
Antonia Marsden, Fiona Holland, Stephen Pye, Hui Guo
16:00 End
Manchester Academic Health Science
Centre Trials Co-ordination Unit (CTU)
Professor Richard Emsley Deputy Director, MAHSC CTU Centre for Biostatistics, School of Health Sciences
What is the MAHSC CTU?
• Hosted by Christie NHS Trust since 2000
– Relocating to University by end 2017
• Funding from:
– MAHSC partners
– National Institute for Health Research CTU Support Funding
– University of Manchester (£1.2m)
– Health Innovation Manchester (£370k)
• New management team and structure
UK Clinical Trials Units
• Specialist units with a specific remit to design, conduct, analyse and publish clinical trials & other well-designed studies
• Currently 50 registered CTUs (46 Full/4 provisional)
www.ukcrc-ctu.org.uk
Pre Trial
• Statistics & Methodology
• Feasibility
• Budget
• Timely submission to Fun and constitution of IDSMC & TSC
• Liaison with Sponsor
• Trial supplies
Set-up
• Protocol development
• Coordinate meetings of oversight committees
• Liaise with sponsor
• Management of trial supplies
• CRF and database development
• Trial approval – ethical, institutional
• Establish/maintain Essential Documents centrally
• Site initiation and ongoing communication
Recruitment and follow-up • Primary contact
• Safety monitoring/ reporting
• Progress reports, amendment notification
• Monitor accrual
• Database support, data query resolution
• Interim and final analysis
• Reports for oversight committees
• Notification of the end of the trial
• Archiving arrangements
What does a CTU do?
Biostatistics input into CTU
• Trial design options
• Sample size/power
• Statistical analysis plans
• Statistical analysis
• Report during the trial
• Our model is to cost a senior and junior
statistician in each proposal
• Contact us as early as possible!
Areas of statistical expertise
1. Early phase oncology trials
– New Senior Lecturer in Clinical Trial Statistics
2. Efficacy and mechanisms evaluation
– Causal inference methods
– Precision medicine
3. Health Technology Assessment
Find out more at:
http://www.mahsc.ac.uk/
The NIHR Manchester Musculoskeletal
Biomedical Research Unit (BRU)
…and other stories
Jamie Sergeant
Lecturer in Biostatistics
Contents
• About me
• NIHR Manchester Musculoskeletal Biomedical
Research Unit (BRU)
• NIHR Manchester Biomedical Research Centre
(BRC)
• Biostatistics teaching and learning
…all with a focus on collaboration
About me
• Training in Maths and Statistics
• Qualified teacher
• Experience in risk prediction:
– breast cancer
– musculoskeletal diseases
• Interest in statistics:
– teaching and learning
– communication
NIHR Manchester Musculoskeletal
Biomedical Research Unit (BRU)
• 2012-2017
• £4.9m NIHR funding
• “Treating arthritis: right first time”
• Understanding why some patients
respond to certain treatments and
do not
BRU example: Predicting non-response to
methotrexate in rheumatoid arthritis
• Methotrexate is cheap and
effective for many…but not all
• Idea: predict who is unlikely to
respond to methotrexate
– earlier access to alternative drugs
– prevent disease progression
• Data: Rheumatoid Arthritis
Medication Study (RAMS)
BRU example: Predicting non-response to
methotrexate in rheumatoid arthritis
• Methods: multivariable
regression prediction model
• Results:
– easily-measurable predictors used
to predict non-response
– patient anxiety of particular interest
• Next: additional genetic and
biological predictors
NIHR Manchester Biomedical Research
Centre (BRC)
• 2017-2022
• £28.5m NIHR funding
• 7 themes
• 3 cross cutting themes
BRC themes
• Advanced radiotherapy
• Cancer precision medicine
• Cancer prevention and early detection
• Dermatology
• Hearing health
• Musculoskeletal diseases
• Respiratory diseases
BRC cross cutting themes
• Rapid translational incubator
• Biomarker platforms
• Informatics and data sciences
Biostatistics teaching and learning
• Context:
– Students have statistics learning needs
– Staff have statistics teaching needs
• Actions:
– Training course aimed at non-statisticians teaching
statistics: “Strategies for teaching statistics”
– Pilot of “Stats Buddy” offering specialist support to
non-specialist staff
…and clinical training for non-clinicians
CLAHRC Greater Manchester 23
NIHR CLAHRC Greater
Manchester Sarah Cotterill
Sarah Rhodes
Eva Batistatou
Nia Coupe
Centre for Biostatistics Showcase – 3 May 2017
CLAHRC Greater Manchester 25 CLAHRC Greater Manchester 25
OSCARSS
• Organising Support for CARers of Stroke Survivors
• Cluster randomised controlled trial (RCT)
• Partnership Between CLAHRC GM and Stroke Association
• Statistician - Sarah Rhodes:
–Design of cluster RCT
–Trial Management
–CLAHRC Workshop on design of
implementation research
CLAHRC Greater Manchester 26 CLAHRC Greater Manchester 26
Wound Audit
• Audit of prevalence of complex wounds and their
treatment
• Partnership between GM CLAHRC and 5 NHS Primary
Care Trusts
• Statistician - Sarah Rhodes:
–Analysis of large data set
–Dissemination to NHS Trusts
–Paper on ‘Better Value Wound Care’
–Using results to plan future work
CLAHRC Greater Manchester 27 CLAHRC Greater Manchester 27
End of Life: Carer Survey
• The contributions of family care-giving at end of life: A national post-
bereavement census survey of cancer carers
• Partnership between CLAHRC GM and Dimbleby Cancer Care
• Statistician – Eva Batistatou
- Analysis of the census survey of cancer carers
- Dissemination of results
- Plan of future work
CLAHRC Greater Manchester 28 CLAHRC Greater Manchester 28
Type 2 Diabetes Prevention
• Evaluation of NHS Diabetes Prevention Programme
– Salford pathfinder.
• Partnership between CLAHRC GM and Salford
health trusts, local authority and voluntary sector.
Funded by Public Health England.
• Mixed methods study: literature review, analysis of
routine data and qualitative interviews.
• Principal Investigator – Sarah Cotterill
CLAHRC Greater Manchester 29 CLAHRC Greater Manchester 29
Diabetic neuropathy screening
• Corneal confocal microscopy (CCM) for screening and early
assessment of diabetic neuropathy
• Partnership between CLAHRC GM and Heidelberg Engineering.
• Feasibility study in 4 opticians (400 patients)
• Statistician – Sarah Cotterill
– Research design
– Project management team
– Contribution to analysis
– Reports, papers
CLAHRC Greater Manchester 30 CLAHRC Greater Manchester 30
CLAHRC PhD: Nia Coupe
• SMART-C: commitments to promote behaviour change in people who
are overweight or obese
• Supervisors
– Sarah Peters
– Sarah Cotterill
CLAHRC Greater Manchester 31 CLAHRC Greater Manchester 31
• ‘TIDieR or Messier? Exploring the utility of TIDieR beyond trials’
• TIDieR template used in 6 CLAHRC GM projects
• Findings:
–We found the TIDieR tool to be a useful tool for applied research
outside the context of clinical trials and we have identified four
amendments to enhance its utility
• Authors: Sarah Cotterill, Sarah Knowles, Anne-Marie Martindale,
Rebecca Elvey, Nia Coupe, Susan Howard, Paul Wilson, Michael
Spence.
CLAHRC Greater Manchester 32 CLAHRC Greater Manchester 32
NIHR CLAHRC GM
Project reports and papers:
http://clahrc-gm.nihr.ac.uk/our-work/
Matthew Gittins
Lecturer in Biostatistics [email protected]
PhD – Environmental Epidemiology
Manchester Guardian, 8 December 1952 Wilkins ET. Air Pollution and the London Fog of December, 1952. Ama Archives of Industrial Hygiene and Occupational Medicine. 1954;9(3):247-248.
Temperature oC-16
Lag Period (Dys)
-10
Rel
ativ
e R
isk
to 1
0oC
-5 0 5 10 15 20 25
1
6
12
18
24
30
0.97
1.00
1.04
All Cause
Temperature oC-16
Lag Period (Dys)
-10
Rel
ativ
e R
isk
to 1
0oC
-5 0 5 10 15 20 25
1
6
12
18
24
30
0.94
1.00
1.06
Pneumonia
Temperature oC-16
Lag Period (Dys)
-10
Rel
ativ
e R
isk
to 1
0oC
-5 0 5 10 15 20 25
1
6
12
18
24
30
0.98
1.02
1.06
COPD
Temperature oC-16
Lag Period (Dys)
-10
Rel
ativ
e R
isk
to 1
0oC
-5 0 5 10 15 20 25
1
6
12
18
24
30
0.98
1.01
1.05
Ischaemic Heart Disease
Black Smoke0.00
Lag Period (Dys)
43.50
Re
lativ
e R
isk
to 8
.0
87.001
6
12
18
24
30
0.98
1.01
1.04
All Cause
Black Smoke0.00
Lag Period (Dys)
43.50
Re
lativ
e R
isk
to 8
.0
87.001
6
12
18
24
30
0.98
1.02
1.05
Pneumonia
Black Smoke0.00
Lag Period (Dys)
43.50
Re
lativ
e R
isk
to 8
.0
87.001
6
12
18
24
30
1.00
1.02
1.04
COPD
Black Smoke0.00
Lag Period (Dys)
43.50
Re
lativ
e R
isk
to 8
.0
87.001
6
12
18
24
30
0.99
1.01
1.03
Ischaemic Heart Disease
Recent/Current Work
0
.05
.1
.15
.2
Rela
tive F
req
0 5 10 15 20 25 No. Cases per monthly report
OPRA 1996-2012 - 'Core' reporters only
• STRATEGIC - HTA clinical trial aiming to improve the uptake of cervical screening particularly in non-responders at first call.
• SSNAPIEST – Observational study describing and investigating factors that impact on stroke therapy received and the effect on health outcomes.
• THOR - Surveillance scheme monitoring occupational ill-health in the UK particularly attempting to account for excess zeros due to reporter fatigue.
Comparing missing data techniques
for intention-to-treat analyses
Liz Howarth
Missing data is a problem
• RCTs best method to compare treatments
– Complicated by non-compliance & missing data
• Non-compliance not necessarily a problem
– Pragmatic ‘ITT’ analysis ignores compliance
• Missing data IS a problem
– biased estimates, faulty inference
• Can we exploit links between compliance & missingness?
Compliance & missingness
• Some missing data techniques use auxiliary models
– Multiple Imputation
– Inverse Probability Weighting
• Does use of compliance in auxiliary models improve resulting ITT analysis?
Incomplete data ITT analysis Auxiliary model
Simulation study
• Generate data
– outcome & missingness both depend on compliance
• Compare different methods
– With and without compliance in auxiliary model
– ITT analysis on full data used as benchmark
• Bias and variability used as performance measures
PhD Aim: Develop statistical
methodology for the analysis of
complex multistage treatment
data.
Jack Wilkinson. Centre for Biostatistics Showcase 2017
Funder: NIHR
Supervisors: Steve Roberts, Andy Vail
PhD overview
IVF success rates
• 33 different ways of reporting pregnancy rates.
• 9 different ways of reporting live birth rates. Only 51% of websites reported live birth at all, 42% per cycle started – one to HFEA standard.
• 72% presented results without sample sizes, 26% without patient ages, 23% without specifying period. 80% did not report selection criteria.
Chemical Pregnancy 19% of clinics: -per cycle started -per egg recovery -per frozen cycle -per insemination -per transfer -unspecified denominator
Clinical Pregnancy 81% of clinics: -per cycle started -per egg recovery -per frozen cycle -per insemination -per transfer -per cycle -per first cycle -per treatment -unspecified -unspecified (cumulative) -per course of inseminations -per egg recovery (cumulative) -per three cycles -per blastocyst stage achieved -per embryo transferred
Pregnancy (undefined) 21% of clinics: -per cycle started -per frozen cycle -per insemination -per transfer -per cycle -per patient (cumulative) -per three cycles -unspecified denominator -per blastocyst stage achieved
Singleton Pregnancy 2% of clinics: -unspecified denominator
Continuing Pregnancy 2% of clinics: -per cycle started -per frozen cycle
OVERVIEW • Primarily applied statistician. • Cautiously skeptical re: much methodology research. • Given that most research is bad, how can statisticians be
more effective? • Teaching: statistics as reasoning and arguing with data –
not as a maths problem.
Calvin Heal. Centre for Biostatistics Showcase 2017
The effects of arthritis gloves on people with
Rheumatoid Arthritis or Inflammatory Arthritis with
hand pain: a multi-centre randomised controlled trial.
Pressure &
Warmth Warmth
or
For reducing hand pain during day activity.
Calvin Heal. Centre for Biostatistics Showcase 2017
Non-pharmacological interventions for spatial neglect due to
stroke and other non-progressive acquired brain injury in adults
Spatial neglect
• Lack of awareness of objects or people to the left
• Serious symptoms
• Often a result of injury to right cerebral hemisphere
Treatment / Rehabilitation
• Prismatic adaptation
• Transcranial direct-current stimulation
• Video feedback training
• Most treatments relatively new and have
limited evidence
Calvin Heal. Centre for Biostatistics Showcase 2017
A single-blind randomised controlled trial with a four
month and 12 month follow up comparing
Group-Metacognitive Therapy plus usual cardiac
rehabilitation (intervention group) with
usual CR alone (control group).
Background
• Heart attack/failure -> Cardiac rehabilitation
• 69000 patients annually (37% significant anxiety/depressive symptoms)
• Current treatment options have limited benefit
Metacognitive therapy