Data Users 2008 Ottawa 1 Reshaping Official Health Statistics: Evolution of Administrative Health...
-
Upload
gervais-bridges -
Category
Documents
-
view
221 -
download
1
Transcript of Data Users 2008 Ottawa 1 Reshaping Official Health Statistics: Evolution of Administrative Health...
1Data Users 2008 Ottawa
Reshaping Official Health Statistics: Evolution of Administrative Health
Data in Canada
Michael Wolfson
Statistics Canada
2Data Users 2008 Ottawa
Three Major Phases Canadian context: constitution gives jurisdiction for
health care to provinces up to mid-1990s – direct uses of routinely collected
administrative data recent past to present – growth of record linkage future – introduction of electronic health records;
debate over “secondary use” n.b. some provinces much more advanced
3Data Users 2008 Ottawa
Phase 1 – Direct Uses of Administrative Data in Health Statistics
birth and death registration – since 19th century» mortality rates, life expectancy, ecological analysis
hospital in-patient admissions – since 1960s» basic prevalences of biomedically-defined disease» small area surgical procedure rate variations
partial exception: cancer registry
4Data Users 2008 Ottawa
Phase 2 – Broadened and more Powerful Use of Administrative Data via Record
Linkage
move from each administrative encounter to each individual person as basic unit of analysis → “trajectory” of encounters
n.b. Manitoba Centre for Health Policy actually the pioneer; longitudinally linked hospital + physician + nursing home + other records dating from late 1970s
three examples: census ↔ mortality, hospitals ↔ survey, hospitals ↔ hospitals
5Data Users 2008 Ottawa
Health Inequalities – Urban Life Expectancy at Birth, By Income
Quintile, Canada
Source: Wilkins et al, Statistics Canada, mortality and census data
66
68
70
72
74
76
78
80
82
1971 1976 1981 1986 1991 1996
Q1 - RichestQ2Q3Q4Q5 - Poorest
3.3 years
6Data Users 2008 Ottawa
66
68
70
72
74
76
78
80
82
84
86
1 2 3 4 5 6 7 8 9 10
Males Females
Health Inequalities – Household Life Expectancy by Sex and Income Decile
(assuming survival to age 25; from 1991 Census + mortality follow-up to 2001)
7.6 years
4.8 years
7Data Users 2008 Ottawa
4.23.8
4.3
6.0
0
1
2
3
4
5
6
7
Underweight Acceptable Overweight Obese
Hospitalization Rate (%) by Body Mass Index(2001-2002; excluding pregnancy and childbirth; excluding Quebec)
Age-sex standardized to Canadian population
48% 33% 17%2%
8Data Users 2008 Ottawa
0
20
40
60
80
100
120
140
160
180
200
116 Health Regions
Crude (Unadjusted) Rate (2.3 fold)
Variation in Hospitalization Rates Across Health Regions with and without Adjustments
(visits per 1,000)
9Data Users 2008 Ottawa
0
20
40
60
80
100
120
140
160
180
200
116 Health Regions
Crude (Unadjusted) Rate (2.3 fold)
Adjusted for Age and Sex (2.2 fold)
Variation in Hospitalization Rates Across Health Regions with and without Adjustments
(visits per 1,000)
10Data Users 2008 Ottawa
0
20
40
60
80
100
120
140
160
180
200
116 Health Regions
Crude (Unadjusted) Rate (2.3 fold)
Adjusted for Age and Sex (2.2 fold)
Also Adjusted for Illness, Health Care Use, Risk Factors (2.0)
Variation in Hospitalization Rates Across Health Regions with and without Adjustments
(visits per 1,000)
11Data Users 2008 Ottawa
0
20
40
60
80
100
120
140
160
180
200
116 Health Regions
Crude (Unadjusted) Rate (2.3 fold)
Adjusted for Age and Sex (2.2 fold)
Also Adjusted for Illness, Health Care Use, Risk Factors (2.0)
Also Adjusted for SES Factors (1.7 fold)
Variation in Hospitalization Rates Across Health Regions with and without Adjustments
(visits per 1,000)
12Data Users 2008 Ottawa
Underlying Person-Oriented Information for Heart Attack / Revascularization Analysis
Heart Attack (AMI)Treatment (revascularization = bypass or angioplasty)Death
one year observation window
one year follow-up window(excluded)
time
13Data Users 2008 Ottawa
0
5
10
15
20
0 10 20 30 40 50 60 70Percent Revascularized within 30 Days
30
Da
y M
ort
ali
ty R
ate
1995/96
2003/04
Heart Attack Survival in Relation to Treatment by Health Region, Seven Provinces, 1995/96 to 2003/04
14Data Users 2008 Ottawa
Important Caveats for the AMI → Revascularization → Mortality Results
revascularization is also intended to relieve symptoms other clinical aspects of treatment not taken into account, e.g.
thrombolysis, post discharge Rx no risk factors – obesity, physical fitness, smoking,
hypertension, lipids – considered no socio-economic factors considered n.b. in related analysis, co-morbidity (Charlson Index) was
included, with one-year mortality follow-up – results essentially unchanged
15Data Users 2008 Ottawa
Key Messages re Phase 2
use of administrative data is much more powerful if combined with record linkage, both within admin data sets and across to health surveys» privacy and vested interests remain major challenges
especially last set of results suggest major potential in Canada’s health care sector to improve health outcomes without more resources – working smarter, not harder» “you can’t manage what you don’t measure”
national data essential to give both the needed sample sizes and to provide the breadth of “natural experiments”
16Data Users 2008 Ottawa
Phase 2.5 – LHAD: Longitudinal Health and Administrative Data Initiative
(simple) idea: build a more analytically powerful database of longitudinally linkable individual level data» bring together a wide range of administrative data on health care encounters –
client registry, hospitals, Rx» plus over 500,000 Statistics Canada health survey responses (where consent to
link with provincial health care records has been given) – NPHS, CCHS, CHMS» plus vital events (births, deaths) and cancer registry» using sophisticated record linkage methodology
extreme care to protect confidentiality mechanism – governed by MoUs between Statistics Canada and each provincial
health ministry
17Data Users 2008 Ottawa
Phase 3 – Influencing the Content of Future Administrative Data for Statistical Purposes
emerging electronic health record (EHR) so far, driven by patient care considerations growing discussion of “secondary” or “health system” uses
of EHR» significant privacy concerns» important counter-moves, e.g. research community and “health
information summit”» idea: articulation of a carefully designed set of “use cases” /
“killer examples”
18Data Users 2008 Ottawa
Infoway – Conceptual Architecture
19Data Users 2008 Ottawa
Infoway Use Cases – the Lamberts (1)
An overview of health issues and interventions of the members of a fictional extended family who are the subjects of care in all subsequent use-cases
This use of a persistent set of actors is intended to provide commonality for discussion of information requirements, and to effectively illustrate the need for relevant health information to be captured and reused:
» in many different care settings» across many different disciplines» over time
20Data Users 2008 Ottawa
Infoway Use Cases – the Lamberts (2)
narrative form describes:» the health services delivery context for each encounter,» who the principle actors are,» the specific expectation for information
capture and reuse across and between encounters – the major outcomes expected from the use of this information
21Data Users 2008 Ottawa
Infoway Use Cases – the Lamberts (3)
Encounter (Clinical Use Case) ≡ narrative of interactions patient has with a provider in a health care setting such as the Emergency Room, an Outpatient Clinic, a Physician Office etc.
Clinical Activities ≡ lowest level of detail that describes the workflow event step for each actor’s (provider and patient) interactions with the Point of Service (PoS) systems and information sent or retrieved from the EHRi System.
22Data Users 2008 Ottawa
Infoway and “Secondary Use”
so far, limited interaction (privacy chill, physician resistance) idea: extend “use cases” to “health system” / secondary uses
» e.g. cancer registry → many disease registries, e.g. AMI, diabetes
» small area variations as a function of most relevant covariates
» standardized and regular assessment of health outcomes
» “continuity of care” metrics, e.g. GP → specialist → hospital → Rx, rehab → GP → home care, long term care etc.
» Rx post-marketing surveillance
» health care costs and outcomes as function of procedure volumes
» etc., etc.
23Data Users 2008 Ottawa
Concluding Comments
major growth over the past decade in use of administrative data in health statistics
excellent initiatives underway» growing use record linkage in partnership with provincial
health care providers» growing efforts to influence future content of health care
encounter data with broader statistical and “health system” uses in mind
concerns with “privacy chill” remain