Applying Propensity Score Matching to Assess the Impact of Community Health Insurance on the Demand for Health Care in Rural Burkina FASO

Posted: 22 Jun 2007

See all articles by Devendra Gnawali

Devendra Gnawali

Heidelberg University

Subhash Pokhrel

Brunel University London - School of Health Sciences and Social Care

Mamadou Sanon

CRSN, Nouna

Manuela De Allegri

Heidelberg University

Rainer Sauerborn

University of Heidelberg - Medical School - Tropical Hygiene and Public Health

Date Written: June 2007

Abstract

Rationale: During recent years community health insurance (CHI) has emerged as an alternative health care financing mechanism in low-income countries. It is believed that it has the potential to lower the financial barriers at the point of service use and hence it is likely to increase the demand for health care. However, the empirical evidence remains weak and there is a real paucity of community trials to assess the impact of CBI.

Objectives: The objective of this paper is to detect any differences in the demand for health care between individuals who have enrolled in the community health insurance and those who have not.

Methodology: We used household survey panel data collected from the demographic surveillance system (DSS) area of the Nouna Health District, Burkina Faso in May/June 2006. The whole DSS area was divided into 33 clusters. Following a step-wedge cluster randomized community trial design, CHI was offered to households in a stepwise fashion (11 clusters per year). The year 2006 marked the final point of the design when CHI was offered to everyone in the trial area. Due to low enrollment, all enrolled households were selected for interview. In order to control for potential biases due to self-selection into the enrolment group, we applied propensity score matching to estimate health care demand. First, we used a logit specification to model enrolment decision (participation) and used it to derive propensity scores. We then matched each participant to one or more non-participants on propensity score using nearest neighbor and caliper matching techniques. Finally, we ran a multivariate analysis of health care demand based on the new sample.

Results: A total of 2580 (4.3%) were enrolled in CHI from the target population (60000). Of the 10804 individuals interviewed in the survey, a total of 526 individuals (4.9%) reported at least one illness episode in the one-month recall period, out of which 364 (69.2%) sought health care outside home. The propensity score matching showed that there were notable differences in terms of health care visits for acute care between insured and non-insured population.

Conclusion: Our study indicates that community health insurance influences health-care demand and in order to detect true differences in health care demand between insured and non-insured population, techniques like propensity-score matching may be useful. We discuss the implications of our results.

Keywords: Community health insurance, propensity score matching, health-care demand

Suggested Citation

Gnawali, Devendra Prasad and Pokhrel, Subhash and Sanon, Mamadou and De Allegri, Manuela and Sauerborn, Rainer, Applying Propensity Score Matching to Assess the Impact of Community Health Insurance on the Demand for Health Care in Rural Burkina FASO (June 2007). iHEA 2007 6th World Congress: Explorations in Health Economics Paper, Available at SSRN: https://ssrn.com/abstract=994644

Devendra Prasad Gnawali (Contact Author)

Heidelberg University ( email )

Department of Tropical Hygiene and Public Health
INF 324, 69120 Heidelberg
Heidelberg
Germany

Subhash Pokhrel

Brunel University London - School of Health Sciences and Social Care ( email )

United Kingdom
0044 18952 68745 (Phone)

Mamadou Sanon

CRSN, Nouna ( email )

Centre de Recherche en Santé de Nouna
Nouna, Kossi
Burkina Faso

Manuela De Allegri

Heidelberg University ( email )

Grabengasse 1
Heidelberg, 69117
Germany

Rainer Sauerborn

University of Heidelberg - Medical School - Tropical Hygiene and Public Health ( email )

Heidelberg
Germany

Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
1,495
PlumX Metrics