Diapositive 1 - International AIDS Society

Pr Jean-Paul MOATTI
University of Aix-Marseille 2 (France)
INSERM/IRD Research Unit 912 SE4S
Economic & Social Sciences, Health Systems &
Societies
HIV and health
expenditures surveys : the
need for data at individuals’
level
THE CONTEXT
Health economic literature = strong inefficiencies
in health systems (in general and exacerbated in
developing countries)
Heterogeneity in unit costs for delivering HIV
services (beyond economies of scale & scope)
International and domestic funding bodies asking
for information about « value for money »
Scale and Average Unit Cost of VCT programs in 5 countries
US$ Average Unit Costs
1,000
100
10
1
1
10
Mexico
100
1,000
Annual clients receiving VCT
Uganda
Russia
India
Source: Preliminary analysis of PANCEA data. Unpublished data. 2006
10,000
100,000
South Africa
THE CONTEXT
Health economic literature = strong inefficiencies
in health systems (in general and exacerbated in
developing countries)
Heterogeneity in unit costs for delivering HIV
services (beyond economies of scale & scope)
International and domestic funding bodies asking
for information about « value for money »
THE METHODOLOGICAL TRADE-OFF
Inter-countries and Intra-country comparisons =>
standardized easy to collect indicators…..but =
High risk of misinterpretations (econometric
results sensitive to heterogeneity of biases in data
collection and models specification)
VS
Population surveys and patients’cohorts =>
detailed data at individuals’ or households’ level =
more amenable for proper analysis….but =
increased implementation difficulties on a routine
basis
THREE EXAMPLES OF THE USE OF
INDIVIDUAL DATA
ARV source prices (Brazil) (Meiners et al., Vienna
Conference, THAE0205- Thursday 22nd)
EVAL-ANRS survey among HIV+ patients in
health care centers (Cameroon) (Boyer et al., AIDS
2010, 24: S5-S15)
Micro-simulation of DHS data for macroeconomic impact estimation of alternative
scenarios (Cameroon, Swaziland, Tanzania)
(UNAIDS, 2010)
Data and methods (Brazil)
Object: adult ARV transactions (2009 USD)
Analysed period: 1996 - 2009
Data: transaction,
characteristics
drug
and
market
Source: Dept of STD, AIDS and VH, MoH
5/13
Results
Coverage and mean ARV expenditures (Brazil)
200,000
6000
5000
160,000
140,000
4000
120,000
100,000
3000
80,000
2000
60,000
40,000
USD (2009)
Number of adult patients
180,000
1000
20,000
0
0
N=607
Treatment coverage
Mean Cost/Patient/Year
6/13
Results
PYD according to supply type (Brazil)
6000
USD (2009)
5000
4000
3000
2000
1000
0
Total
Branded ARVs
Non-Branded ARVs
7/13
CAMEROON: the EVAL – ANRS 12 116 research project
Cross-sectional and multicentre study (Sept., 2006 to Apr.,
2007)
- 3151 HIV infected adults interviewed in 27 HIV services at the 3
levels of the healthcare delivery (Central; Provincial; District)
- Random sample
- Face-to-face interview and anonymous questionnaires
- Medical information card and blood sample (CD4 cell counts)
- Response rate: 91%
- 97 physicians in the same 27 HIV services
- All-inclusive
- Anonymously self-reported questionnaires
- Data on characteristics of healthcare services
- Data on decentralization policy
CAMEROON
Direct out-of-pocket costs related to HIV infection (by
month – median) (n=2412 ART-treated pts)
Health expenditures (exp.)* (in FCFA X 103)
- TOTAL (exp. >0 : 98%)
- ART (exp.>0 : 88%)
- Transportation (exp. >0 : 85%)
- Consultations (exp. > 0 : 33%)
- Other drugs (exp. >0 : 20%)
Proportion of expenditures in household
income by quintiles
- 1st quintile
- 5th quintile
Yde &
Dla
Prov.
District
8,6
3,0
1,0
2,0
5,0
6,6
3,0
1,2
1,0
5,0
5,8
3,0
2,0
1,0
1,5
27,2%
3,5%
9,0%
2,3%
16,0%
2,0%
- Catastrophic Health Expenditures (≥ 20% households’ resources) : 44%
CAMEROON EVAL ANRS Survey
Factors associated with the risk of catastrophic health exp.* (n=2412)
Coeff
p
Monthly income by equivalent adult (1st quintile = ref.)
- 2nd quintile
- 3rd quintile
- 4th quintile
- 5th quintile
-1.1
-1.8
-2.4
-3.0
***
***
***
***
Wealth index
-1.3
***
Free access to ART (interaction term) : - Constant
- Free access: provincial level
- Free access: district level
-1.3
-0.1
-0.9
***
NS
**
Transportation length to the hospital < 1 hour
-1.1
***
Consultation with a private doctor outside the reference hospital
0.4
**
Consultation with a traditional healer: constant
- variance of random effect
0.7
0.7
**
**
-0,6
- 0,6
***
*
Decentralization: - central level (ref.)
- Provincial level
- District level
* Adjustment variables: gender, age, matrimonial status, area of residence, CD4 at initiation, time
since HIV diagnosis, nb of perceived symptoms / technologic level of the medical centre
Estimates of alternative scenarios of
scaling-up of ART treatment in an
agent-based microsimulation model
Ventelou(1,2), Arrighi(1,2), Afridi(1,2), Greener(3), Lamontagne(3), Moatti(2)
Contact Author:
(1) CNRS GREQAM / INSERM Unit 912 and PACA Regional Center for Disease Control
(2) INSERM Unit 912 and PACA Regional Center for Disease Control
(3) UNAIDS
CAMEROUN
–
–
–
–
2004 Cameroon Demographic and Health Survey (EDSC III)
Large dataset : 35,000 individuals sampled
Numerous data on socio-economics and perception of AIDS
10,900 Adults aged15-49 are retained
– Linked with a HIV Blood Test Record– 9,551 tests were performed
–
–
–
–
Results in a Sample of 8,186 HIV+ and HIV- individuals
46.2% Men ; 53.8% Women (weighted)
HIV Prevalence = 7.5% (weighted)
Every Individual represents xxx
Agents in the database are:
HIVnegative / HIVpositive / HIVpositive+needing ART
The proportion of PLWHIV needing ART has been obtained from WHO data
(not given in the dataset) - differentiated across age classes and
genders, for taking into account a probable longer date of infection
among the oldest. We randomly assign agents to the HIV+TN status
4 Status:
Future
states =
obtained
by artificial
“ageing”
(Markov)
CBA:
Aid Freeze vs. Universal Access
- Universal Access dominates Aid Freeze only
on the long-run
CBA - Self Financing Ratio
300%
- Gains are lower
(GDP per capita...)
250%
Aid Freeze
SFR
200%
150%
Universal
Access
100%
50%
0%
2009
2014
2019
2024
2029
2034
Ad-Hoc Technical Advisory Group
on Costing HIV/AIDS Interventions
(WHO, GFATM, PEPFAR)- June 2010
“WHO and partners should go forward
with two levels of program-level ART costing:
a routine data collection across a few basic
cost categories at national level; and secondly
a more detailed exercise to guide countries in
producing reliable cost figures for
comparative analysis”
Recommendations
Need of Multi-country/multi-sites surveys
with individual data on HIV expenditures
 Need of an Operational Research pooled
mechanism