Presentation Title

Report
Risk Pooling to Achieve Universal Coverage:
Ghana’s National Health
Insurance Scheme
Slavea Chankova
Abt Associates Inc.
In collaboration with:
I Aga Khan Foundation I BearingPoint
I Bitrán y Asociados I BRAC University
I Broad Branch Associates
I Forum One Communications I RTI International
I Training Resources Group
I Tulane University’s School of Public Health
I. BACKGROUND
The National Health Insurance Scheme (NHIS)
 Established by legislation in 2003
 Goal: equitable and universal access to health care
 Coverage reached 66% in 2010
 Evaluation of NHIS
 Designed in anticipation of NHIS implementation
 Collaboration between Health Systems 20/20 project and
Health Research Unit - Ghana Health Service
Key Features of the NHIS
 Managed by district-level mutual health insurance
schemes
 Providers: all public health facilities and accredited
private providers
 Benefits: 95% of disease conditions, essential drugs
 Enrollment
 Open to all with sliding-scale premium contributions
starting at about $5 per adult
 Premium exemptions for children (under 18), elderly
(70+), indigent, and pregnant women (as of 2008)
II. EVALUATION DESIGN
Evaluation Questions
 Who has enrolled in the NHIS?
 Do enrollment rates differ among different socio-economic groups?
 Is there evidence of adverse selection in NHIS enrollment?
 How well-targeted have premium exemptions been?
 What is the impact of the NHIS on the utilization of
health services?
 What is the impact of the NHIS on out-of-pocket
expenditures for health care?
Evaluation Design
 Pre-post study design
 Baseline in September 2004
 Endline in September 2007
 Implementation of NHIS in study sites started in 2005
 Cross-sectional household surveys in 2 districts
 Nkoranza (had CBHI at baseline)
 Offinso
Study Sample
Baseline
2004
Endline
2007
Number of households
1,805
2,520
Number of individuals
9,554
11,770
Individuals reporting illness/injury in
past 2 weeks
413
411
Individuals reporting hospitalization in
past 12 months
203
208
Women reporting delivery in past 12
months
298
312
Analytic Methods
 Pre-post comparison of means for key indicators
 Regression models
 Control for differences in socio-economic characteristics
between baseline and endline samples
 Probit and logistic regression models
 Results were robust to analytic methods
III. RESULTS
Sample Characteristics
 Poor rural population
 General improvements in socio-economic characteristics between
2004 and 2007
 Health insurance coverage:
Baseline 2004
(Nkoranza CBHI)
Endline 2007
(NHIS)
Nkoranza
35%
45%
Offinso
0%
25%
23%
35%
Total sample
Who Enrolls in NHIS?
 Enrollment increases with wealth quintile
 Poorest are 3 times less likely to enroll compared to the richest
60%
% of wealth quintile enrolled in NHIS
52%
50%
40%
39%
Middle
Middle-rich
40%
30%
30%
20%
18%
10%
0%
Poorest
Middle-poor
Richest
Who Enrolls in NHIS?
Factors associated with higher likelihood of NHIS enrollment*
 Richer wealth quintile
 Education of household head
 Female headed household
 Female gender
 Age: children and the elderly more likely to enroll, compared to
18-49 yr old
 Self-reported chronic illness
 At least one household member is part of a community
solidarity scheme
* Results from multivariate regression (variables with statistically significant coefficients)
Targeting of NHIS:
Premium Exemptions for Children & Elderly
 Age-based exemptions have worked as intended
 But nearly all enrolled (97%) had paid a registration fee
% of NHIS members exempt from premium
100%
100%
98%
99%
90%
80%
70%
60%
50%
40%
30%
20%
10%
6%
4%
18-34
35-49
11%
0%
0-4
5-17
50-69
70+
Targeting of NHIS:
Premium Exemptions for the Poor
 Exemptions have not benefited primarily those in the lowest
wealth quintile
% of NHIS members exempt from premium
100%
90%
80%
70%
64%
65%
Poorest
Middle-poor
60%
59%
62%
60%
62%
Middle
Middle-rich
Richest
Total
50%
40%
30%
20%
10%
0%
Adverse Selection in Enrollment
 Strong evidence of adverse selection based on health
status
 NHIS-insured almost 3 times as likely to report illness in past 2 weeks,
compared to uninsured
 55% of those with chronic illness insured, compared to 34% of those
without
 No evidence of self-selection in enrollment related to
pregnancy
 36% of women with delivery in the past 12 months were insured at
time of delivery, compared to 33% of women who did not have a
delivery
Utilization of Care for Recent Illness or Injury
100%
2004
Percent of individuals sick in past 2 weeks
90%
2007
76%
80%
70%
70%
60%
50%
50%
40%
44%
36%
37%
30%
20%
10%
0%
Used medication at home (p=0.009)
Sought care from informal
providers (p<0.0001)
Sought care at a modern health
care provider (p<0.0001)
Utilization of Maternal Health Care
 No significant changes between 2004 and 2007 in proportion of
pregnant women receiving key maternal health services
% of women with delivery in past 12 months
100%
2004
90%
80%
70%
2007
73%
68%
60%
54%
55%
50%
40%
30%
20%
6%
10%
6%
0%
4 or more ANC visits (p>0.10)
Delivery in health facility (p>0.10)
Delivery by c-section (p>0.10)
Likelihood of OOP Expenditures for Care
 Significant decrease in probability of incurring OOP
expenditures for recent curative care, hospitalization,
antenatal care (ANC), and delivery
100%
% with positive OOP expenditures for care
90%
87%
2004
88%
87%
80%
2007
74%
70%
60%
57%
55%
47%
50%
43%
40%
30%
20%
10%
0%
Recent curative care (p<0.01)
Hospitalization (p<0.01)
ANC (p<0.01)
Delivery (p<0.01)
Changes in OOP Expenditures for Care
 Average expenditures for treatment declined significantly for
most services:
 41% decrease for curative care (from $2 at baseline)
 44% decrease for hospitalization (from $25 at baseline)
 No significant decrease for ANC (remained at about $3)
 30% decrease for delivery (from $8 at baseline)
 No significant changes in average amount paid by those who
had positive OOP expenditures
Limitations
 Results from 2 districts (out of 138) so cannot be generalized
to whole country
 Changes between 2004 and 2007 likely reflect impact of
NHIS, but may also be influenced by other factors (e.g. other
socioeconomic or policy changes occurring simultaneously)
 Small samples for some indicators (e.g. hospitalization) limit
the ability of the study to detect significant changes
IV. CONCLUSIONS & POLICY
IMPLICATIONS
NHIS Enrollment
 Age-based exemptions from NHIS premiums for children and
the elderly have worked as intended
 But this may have potential implications for NHIS sustainability
 Strong wealth effects observed for NHIS enrollment
 Exemptions for the poorest groups need to be strengthened to ensure
equitable enrollment in NHIS
 Evidence of adverse selection: those with poorer health
status are more likely to enroll and more likely to utilize care
 Implications for DMHIS sustainability
Utilization and OOP Expenditures
 Substantial increase in use of formal medical services for
illness; decrease in self-treatment and informal care-seeking
 However, no improvement in maternal care-seeking
 Need to explore non-financial barriers for seeking care
 Insurance has been very effective at reducing out-of-pocket
expenditures for curative care and hospitalization, as well as
for maternal care
Acknowledgements
 Abt Associates -- Health Systems 20/20:
 Laurel Hatt, Sara Sulzbach, Hong Wang, Ha Nguyen
 Ghana Health Service/Health Research Unit:
 Dr. John Gyapong, Bertha Garshong
 USAID:
 Yogesh Rajkotia, Karen Cavenaugh, Mary Ellen Stanton
Reports related to this presentation
are available at: www.HS2020.org
Presentation will be posted at:
http://www.abtassociates.com/HSRsymposium
Abt Associates Inc.
In collaboration with:
I Aga Khan Foundation I BearingPoint
I Bitrán y Asociados I BRAC University
I Broad Branch Associates
I Forum One Communications I RTI International
I Training Resources Group
I Tulane University’s School of Public Health

similar documents