cardiovascular disease and diabetes: how can oecd … · cardiovascular disease and diabetes: how...
TRANSCRIPT
CARDIOVASCULAR DISEASE AND DIABETES:
HOW CAN OECD HEALTH SYSTEMS
DELIVER BETTER OUTCOMES?
Progress report 7th November 2013
• A: undertake a descriptive analysis of CVD and diabetes indicators and identify potential reasons for cross-country differences and trends
• B: Analyse health system performance in relation to CVD and diabetes outcomes
• C: Explain variation in health system performance
Project objectives
• Use relevant CVD and diabetes indicators incorporating lifestyles, utilisation, quality and outcomes
• To identify potential reasons for cross-country variation and changes over time, use: – existing peer-reviewed and grey literature to
comment on trends
– Results of questionnaires
• Describe organisational aspects of CVD/diabetes care
A: Descriptive analysis
• How do countries vary in their ability convert health inputs to outcomes?
• Use of multi-level modelling – Country-specific time-trends
– Country-specific fixed-effects
• Series of models: – Dependent variables
– Independent variables
• Which countries have performed better/worse than OECD average?
B: Health system performance analysis
• What characteristics and policy settings help explain potential health system variation? 1. Extension of multilevel model
2. Panel data techniques
3. Heart failure sub-project (with European Society of Cardiology)
• Characteristics and policy settings derived from HSC survey and our own survey: – CVD and diabetes specific
– Broader indicators that may affect CVD and diabetes
C: Explaining variation in health system
performance
Task Timeframe
1st CVD/diabetes survey design/distribution
Jan 2013 √
Return 1st questionnaire May 2013 √
Design and test 2nd CVD/diabetes survey
May-Jun 2013 √
2nd CVD/Diabetes survey in the field Jun to Sep 2013 √ (extended)
Draft report (A) End 2013 extended to Q1 2014
Finalise analysis (B and C) Jan-Apr 2014
Draft report (A,B,C) Nov-May 2014
Final report (A,B,C) End 2014
Project time frame and expert inputs
A: CVD mortality compared
0
100
200
300
400
500
600
700
800
900
196
0
196
2
196
4
196
6
196
8
197
0
197
2
197
4
197
6
197
8
198
0
198
2
19
84
198
6
19
88
199
0
199
2
199
4
199
6
199
8
200
0
200
2
200
4
200
6
200
8
201
0
Mortality from circulatory diseases Mortality from all other causes
Age-standardised rates per 100 000 population
A: Percentage fall in CVD mortality 1985-2011
-80
-70
-60
-50
-40
-30
-20
-10
0
Slo
vak
Re
pu
blic
Me
xico
Gre
ece
Hu
nga
ry
Ice
lan
d
Po
lan
d
Cze
ch R
ep
ub
lic
Ge
rman
y
Swe
de
n
Ch
ile
Be
lgiu
m
Swit
zerl
and
Un
ite
d S
tate
s
OEC
D
Fran
ce
Ital
y
Fin
lan
d
Ne
w Z
eal
and
Au
stri
a
Ne
the
rlan
ds
Slo
ven
ia
Can
ada
No
rway
Jap
an
Un
ite
d K
ingd
om
Ko
rea
Spai
n
Luxe
mb
ou
rg
De
nm
ark
Ire
lan
d
Po
rtu
gal
Au
stra
lia
Isra
el
Esto
nia
A: Mortality and premature mortality
AUS
AUT
BEL
CAN
CHL
CZE
DNK
EST
FIN
FRA
DEU
GRC
HUN
ISL IRL
ISR
ITA
JAPKOR
LUX
MEX
NLD
NOR
NZL
OECD
POL
PRT
SVK
SVN
ESP
SWE
CHEGBR
USA
10
020
030
040
050
060
0
Mort
alit
y r
ate
- p
er
100 0
00 p
opu
latio
n
0 250 500 750 1000 1250 1500Potential years of life lost - per 100 000 population
A: Univariate relationships example:
diabetes admissions and AMI case-fatality
AUS
AUTBEL
CANCHE
CHL
CZEDEU
DNK
ESPFINGBR
HUN
IRLISL
ISRITA
KOR
MEX
NOR NZLPOL
PRT
SWEUSA
R² = 0.3443
0.0
5.0
10.0
15.0
20.0
25.0
0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0 450.0
AM
I ca
se-f
ata
lity
ra
te -
pe
r 1
00
ad
mis
sio
ns
Diabetes related admissions per 100 000 population
A: Univariate relationships example:
diabetes prevalence and pharmaceutical consumption
AUS
BEL
CAN
DNK
EST
FIN
FRA
DEU
HUN
ICE
ITA
KORLUX
NLD
NOR
PRT
SVKSVN
ESPSWE
GBR
01
02
03
04
05
06
07
08
09
0
DD
D for
dru
gs
use
d fo
r d
iabte
s
0 2 4 6 8 10 12Diabetes prevalence (% of population)
• Descriptive data:
– Main health outcomes trends - by major CVD category, gender, age
– Pharmaceutical consumption
– Continue examining of univariate relationships between :
• Mortality, PYLL, case-fatality, ESRF
• Utilisation, risk factors, quality indicators
A. Further descriptive analysis:
• Basic model:
B: performance analysis
empirical framework
• y is outcome of interest in country i at time t.
• β0 and β1 are the intercept and slope coefficients that vary by country.
• I is the explanatory variable of interest.
• refers to fixed slope coefficients for p control variables x.
• е is the residual error term
(1)
• Series of models : -Dependent variables:
• Mortality rate by major CVD groups
• Potential Life-Years Lost (PLYL) by major CVD groups.
• CVD case-fatality rates (AMI, hemorrhagic stroke, ischemic stroke)
• Diabetes admissions and complications (dialysis, transplants, lower extremity amputations)
– Independent variables: • Physical inputs (e.g. number of doctors)
• Financial inputs by health sector
– Control variables (income, public coverage, lifestyle, time)
B: variables
AMI case-fatality (ln) Coef. Std. Err. P>z
Hospital expenditure (ln) -0.265 0.151 0.079
% public funding -0.094 0.277 0.733
Mean income (ln) 0.039 0.122 0.746
% smoke 0.010 0.006 0.118
Alcohol (annual l. per cap) -0.010 0.012 0.412
Year -0.054 0.006 0.000
Constant 112.672 11.458 0.000
B. Preliminary results (model 8: AMI case-
fatality)
Likelihood ratio test to determine whether random slope model is justified: LR chi2(2) = 26.93 (Prob > chi2 = 0.0000)
B. Preliminary results (model 8)
Random-effects Parameters Estimate Std. Err.
country_id: Unstructured
var(hospital expenditure) 0.284 0.151
var(Constant) 13.505 7.347
var(Residual) 0.005 0.000
cov(hosp.exp,_cons) -1.955 1.052
Observations 193
Countries 26
Ave. obs. Per country 7.4
B. Preliminary results (model 8) – random slopes
DNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNKDNK
LUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXLUXNORNORNORNORNORNORNORNORNORNORNORNORNORNORNORNORNORNORNORNORNORNORNORCZECZECZECZECZECZECZECZECZECZECZECZECZECZECZECZECZECZECZECZECZECZECZEAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUS
CANCANCANCANCANCANCANCANCANCANCANCANCANCANCANCANCANCANCANCANCANCANCAN
ICEICEICEICEICEICEICEICEICEICEICEICEICEICEICEICEICEICEICEICEICEICEICE
SWESWESWESWESWESWESWESWESWESWESWESWESWESWESWESWESWESWESWESWESWESWESWE
AUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUTAUT
NLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLDNLD
FRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAFRAJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAPJAP
PRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRTPRT
DEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUDEUCHECHECHECHECHECHECHECHECHECHECHECHECHECHECHECHECHECHECHECHECHECHECHE
SVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKSVKUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAUSAPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOLPOL
ISRISRISRISRISRISRISRISRISRISRISRISRISRISRISRISRISRISRISRISRISRISRISR
HUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNHUNNZNZNZNZNZNZNZNZNZNZNZNZNZNZNZNZNZNZNZNZNZNZNZ
BELBELBELBELBELBELBELBELBELBELBELBELBELBELBELBELBELBELBELBELBELBELBEL
FINFINFINFINFINFINFINFINFINFINFINFINFINFINFINFINFINFINFINFINFINFINFINKORKORKORKORKORKORKORKORKORKORKORKORKORKORKORKORKORKORKORKORKORKORKOR
ESPESPESPESPESPESPESPESPESPESPESPESPESPESPESPESPESPESPESPESPESPESPESPCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHSCHS
-2-1
01
2
vari
atio
n fr
om
ave
rage
0 5 10 15 20 25Rank
• Extend to other dependent variables
– Stroke case-fatalities, CVD mortality, diabetes outcomes.
• Extend to other independent variables
– Disease specific hospital expenditure
– Number of doctors
• Further work on convergence
– Alternative model specifications
B. Further performance analysis:
• Most of the work ahead of us, but:
– ESC/OECD subproject (Prof Aldo Maggioni)
– Letter of agreement signed/draft protocol
• Combines data from:
– ESC’s Heart Failure Long-Term Registry
• 12 000 patients in 21 countries, including 14 OECD countries
• Chronic and acute HF
– OECD HSC and CVD/Diabetes survey
• To what extent to HSC/policy/organisational aspects influence heart failure treatment guidelines and outcomes?
• Outputs: OECD report, paper, Congress of the European Society of Cardiology
C: Explaining variation
Next steps summary
• 2nd CVD/diabetes questionnaire:
– Please contact the Secretariat (if not already done so)
– Continue collating and analysing survey questionnaire
– Interview experts November, December
– Finalise Q4 2013
• Descriptive analysis:
– Write up results, integrating survey and interview results
– Finalise Q1 2014
• Health system performance analysis
– Draft report by May 2014
• Explaining CVD and diabetes performance (incl. ESC/OECD sub-project)
– Draft report by May 2014
• Maggioni, Aldo P., et al. (2013) "EURObservational Research Programme: regional differences and 1-year follow-up results of the Heart Failure Pilot Survey (ESC-HF Pilot)." European Journal of Heart Failure.
• Maggioni, A. P., Anker, S. D., Dahlström, U., Filippatos, G., Ponikowski, P., Zannad, F., ... & Pravdic, D. (2013). Are hospitalized or ambulatory patients with heart failure treated in accordance with European Society of Cardiology guidelines? Evidence from 12 440 patients of the ESC Heart Failure Long-Term Registry. European journal of heart failure, 15(10), 1173-1184.
• Or, Z., Wang, J., & Jamison, D. (2005). International differences in the impact of doctors on health: a multilevel analysis of OECD countries. Journal of Health Economics, 24(3), 531-560.
• Paris, V., M. Devaux and L. Wei (2010), “Health Systems Institutional Characteristics: A Survey of 29 OECD Countries”, OECD Health Working Papers, No. 50.
References