43
Ensuring Quality of Health Care Data: A Canadian Perspective Data Quality Asia Pacific Congress 2011 Heather Richards Consultant Canadian Institute for Health Information (CIHI) Tel:+1 250 220 2206 Email: [email protected] 1

Heather Richards

Embed Size (px)

Citation preview

Page 1: Heather Richards

Ensuring Quality

of Health Care Data:

A Canadian Perspective

Data Quality Asia Pacific Congress 2011

Heather RichardsConsultant

Canadian Institute for Health Information (CIHI)

Tel:+1 250 220 2206

Email: [email protected]

1

Page 2: Heather Richards

> The Canadian Institute for Health Information

> Data Quality Challenges in Canada

> Strategies to Ensure Data Quality:

– CIHI’s Data Quality Framework

– Data Quality Reporting Tools and Studies

– Techniques for Communicating Data

Quality

Agenda

2

Page 3: Heather Richards

The Canadian Institute for

Health Information

3

Page 4: Heather Richards

Canadian Institute for Health Information

> National, independent, not-for-profit agency,

established in 1994

> One of Canada’s leading sources of high-quality, reliable and timely health information

> 27 health databases

and registries

> 7 offices

4

Page 5: Heather Richards

CIHI’s Mandate

> Coordinate, develop, maintain and disseminate

health information on Canada’s health system

and the health of Canadians

5

Page 6: Heather Richards

CIHI's Mandate Con't

> Provide accurate and

timely information

required for:

– Sound health policy

– Effective management

of the health system

– Public awareness about

factors affecting

good health

6

Page 7: Heather Richards

Data Quality Challenges

in Canada

7

Page 8: Heather Richards

Data Challenge: Variety of Partners

> Accommodating different coding standards at

provincial/territorial level versus national level;

> Recognizing different uses of the data and different

focus on data quality;

> Adjusting for differing data collection methods

8

Page 9: Heather Richards

CIHI PartnersRegional

health authorities

Health

Canada

Professional

associations

Private sector

organizations

Researchers

Non-governmental

organizations

Ministries

of health

Statistics

Canada

Health

facilities

CIHI

9

Page 10: Heather Richards

Data Challenge: Secondary Data Collector

> CIHI does not collect data directly

> Our data comes from:

– provincial governments;

– hospitals; and

– professional associations

10

… this means that

we cannot affect first hand

how that data is captured and collected.

Page 11: Heather Richards

> CIHI relies on data providers (some are voluntary

data providers) to report accurate information

> Poor quality data often result from difficulties in

collection standards, coding standards and

chart documentation – and lack of training

11

Data Challenge: Secondary Data Collector

Page 12: Heather Richards

Data Challenges: Other

> Variety of databases and usability

> Data flow and timeliness

> Coding and comparability

> Hospital practices and data completeness

12

Page 13: Heather Richards

End-stage renal failure

13www.vancouversun.com

Page 14: Heather Richards

Question: Are Risk Factors Completely

Captured at all Facilities?

14

Prevalence of Pulmonary Edema

0%

10%

20%

30%

40%

50%

60%

70%

Inter-Quartile Range: 13-27%

Page 15: Heather Richards

Questionnaire Reveals a Correlation of

Data Completeness to Hospital Practices

15

Reviews select documentation

Reviews all documentation

Prevalence of Pulmonary Edema

0%

10%

20%

30%

40%

50%

60%

70%

IQR: 16-29%IQR: 8-21%

Page 16: Heather Richards

Chart Review Confirms Under-Reporting

16

Prevalence (%)

Data Captured

by Dialysis

Clinic Staff

Data Captured

by CIHI coder

during Chart

Review

Pulmonary edema 22 27

Sensitivity=62%

Specificity=93%

PPV=77%

NPV=87%

Epidemiologists and clinical

researchers prefer seeing

these statistics…

Page 17: Heather Richards

Strategies to Ensure Data Quality

17

• CIHI’s Data Quality Framework

• Data quality reports and studies

• Techniques for communicating DQ

Page 18: Heather Richards

Strategies to Ensure Data Quality

18

CIHI’s Data Quality Framework

Page 19: Heather Richards

CIHI’s Data Quality Framework

> Objective approach to

assessing data quality

and producing standard

documentation

> Three parts

1. Work Cycle

2. Assessment Tool

3. Documentation

19

Page 20: Heather Richards

1. Data Quality Work Cycle

20

Plan

ImplementAssess

Page 21: Heather Richards

2. Data Quality Assessment Tool

> Provides a consistent

approach for defining

data quality

> Five dimensions

– Accuracy

– Comparability

– Timeliness

– Usability

– Relevance

5

19

61

Dimensions

Characteristics

Criteria

21

Page 22: Heather Richards

2. Data Quality Assessment Tool

22

AccuracyComparability

Timeliness

Usability

Relevance

CoverageCapture and collection

Unit non-response

Item (partial) non-response

Measurement error

Edit and imputation

Processing and estimation

Population of reference explicitly stated

Coverage issues are documented

Frame validated

Under or over-coverage rate

Page 23: Heather Richards

Assessment Tool: Educational Component

23

Page 24: Heather Richards

3. Data Quality Documentation

> Details the data quality

documentation required

for each data holding

24

Page 25: Heather Richards

Metadata Documentation

Retain

knowledge

about the

management

of a

database

with the

database.

25

Page 26: Heather Richards

Strategies to Ensure Data Quality

26

Data Quality Reporting Tools

and Studies

Page 27: Heather Richards

Deputy Minister Data Quality Reports

> Bird’s eye view

> Broad DQ scope:

assess accuracy,

timeliness, comparability

and usability

> Specific audience:

Deputy Ministers of

Health

27

Page 28: Heather Richards

Features of the Deputy Minister

Data Quality Reports

> Each indicator is important to the success

of a database and has a defined action to

improve performance

– Snapshot of results across all jurisdictions

– Trending over time

> 11 databases

– 8 from CIHI

– 3 from Statistics Canada

28

Page 29: Heather Richards

Components of the Deputy Minister

Data Quality Reports

29

Database-

specific reports

Technical

documents

P/T indicator

tables

Trending

results

Flags table

Each DM package

Page 30: Heather Richards

Trending: Discharge Abstract Database

2003-04

2004-05 2005-06

2006-07

2007-08

2008-09

2009-10

0.0

0.5

1.0

1.5

2.0

2.5

Optimal Value = 0

Indicator 1: Total Outstanding Hard Error Rate, per 1,000 Abstracts

30

Page 31: Heather Richards

Response to Reports

Positive:

> Highlights to DM the value of a

database; increases coverage of

data holdings

> Reveals systemic problems

causing DQ issues; helps Deputy

Ministers prioritize and reallocate

resources

> Congratulates on past DQ

improvements; facilitates creation

of DQ improvement action plans

31

Page 32: Heather Richards

Reabstraction Studies

> Detailed review

> Narrow DQ scope: assess

coding consistency,

correctness, completeness

> Wide audience

32

Page 33: Heather Richards

Reabstractorrecodes from chart

Application reveals original data

Application compares data

Reabstractorassigns reasons for differences

Study Methods

> A chart review to

recapture the data and

compare

33

Page 34: Heather Richards

Overview

Determine study

method

Develop data

collection tool

Train coders,

collect data

Process and

analyze data

Share results

34

Study Objectives

Page 35: Heather Richards

Reabstraction Study Example

DAD: Discharge Abstract Database

> Data on acute-care hospital activity

> Data supports:

– funding and system planning decisions at government level

– management decisions at the facility level

– clinical research at the academic level

35

Page 36: Heather Richards

Strategies to Ensure Data Quality

36

Techniques for Communicating

Data Quality

Page 37: Heather Richards

Communicating Data Quality Using

Different Lenses

37

Statistics for OECD

international

comparisons

Health

indicators

Categorizing

hospitalizations

for hospital

management purposes

Isolating determinants

of good health

Assessing

quality of care

Clinical

research

purposes

such as

survival analysis

Page 38: Heather Richards

Health

Indicators

> Assess population health and

health system performance

> Will look at one indicator:

ACSC hospitalizations

38

Page 39: Heather Richards

Health Indicator: ACSC Hospitalizations

39

2001-022002-03 2003-04 2004-05 2005-06

2006-072007-08

0

100

200

300

400

500

600

Age-Standardized Rate of ACSC Hospitalizations per 100,000 Population

459

326

Page 40: Heather Richards

2007-08 DAD Study: ACSC Hospitalizations

> Question: Is the decrease in ACSC

hospitalizations real or is it due to changes in

coding quality?

> Answer: The observed decrease is real

– National rates are indeed decreasing

– Reabstraction studies found that certain patient

populations had lower quality data

40

Page 41: Heather Richards

2007-08 DAD Study: ACSC Hospitalizations

Sensitivity

Grand mal status, epileptic convulsions 81%

Chronic obstructive pulmonary diseases 91%

Asthma 90%

Diabetes 95%

Heart failure and pulmonary edema 84%

Hypertension 100%

Angina 94%

Any ACSC hospitalization 90%

41

Page 42: Heather Richards

Data Quality Challenges that Lie Ahead

> The health sector is a changing landscape

– Electronic health record

– Health care funding

– New technologies

– New modes of delivering

health care

> New data will bring new quality challenges

42

Page 43: Heather Richards

43

“Taking health information further”