output-based healthcare

Evaluation of the output-based aid project in Uganda began with a thoughtful dialogue with key partners. Experts from the Mbarara University of Science and Technology, University of California at Berkeley, KFW Development Bank, Venture Strategies for Health and Development, Marie Stopes International, and the PSP-One project provided input. Given the constraints to a randomized design and longitudinal follow-up of general population respondents, the evaluation team developed a pre-/post-intervention with a control arm for each survey.

The 2006 baseline study compared intervention and control areas on household assets (long-term socio-economic status), general healthcare and STI-specific healthcare utilization, food security (short-term socio-economic status), general health and STI symptoms, contraceptive prevalence, community level trust, and prevalence of gonorrhea and syphilis in three study arms: greater parishes containing one or more OBA clinics, parishes without OBA clinics, and parishes that have private clinics similar to OBA clinics. There are 15 parishes with OBA contracted clinics. There are also 15 parishes randomly selected by probability proportional to size (PPS) without replacement from the remaining 225 parishes in greater Mbarara district, and 11 parishes purposively selected from Bushenyi district which have one or more private clinics that treat STIs similar to the OBA providers. Although it is common practice to refer to the intervention area as greater Mbarara district, in late 2005 the Mbarara district was divided into four new districts: Ibanda, Isingiro, Kiruhura, and a smaller Mbarara. For the time being, we refer to all four districts as “greater Mbarara”.

Within each parish, two villages were selected PPS without replacement and the local political leaders and health workers informed of the upcoming study’s objectives. The principal investigators from MUST sought verbal permission from local leadership explaining that the study would randomly select 36 respondents from each village between the ages of 15 and 49. The investigators also detailed that each respondent would be asked for their consent after the interviewer explained the purpose of the survey in the local language Lunyankole. If consent was given (assent for respondents under 18 with guardian consent) a 45-minute confidential interview and free syphilis and gonorrhea screening was conducted. Free confidential same-day treatment was provided for positive results when possible and free anonymous follow-up treatment provided by local government services when same-day treatment was not possible or sufficient. Referrals were given to gonorrhea and syphilis cases that do not clear after follow-up treatment. All community leaders accepted the study’s aims and gave their permission for the study teams to work in the communities. They also agreed to provide complete household lists for sampling purposes as well as act as guides to help survey teams locate selected households.

There is strong interest in the global donor community and recipient governments to measure the impact of development assistance. Venture Strategies technical advisor and University of California, Berkeley doctoral student, Ben Bellows has worked with Dr. Fred Bagenda and Dr. Edgar Mulogo of Mbarara University of Science and Technology in the evaluation of healthcare utilization and epidemiologic impact of the OBA network.

The impact evaluation uses two household surveys at pre- and post-intervention in communities with and without the voucher. The unit of comparison is a parish, an administrative unit with 1000 to 3000 residents. For consistency, the OBA evaluation refers to “district > county > subcounty > parish > village” when in common usage other terms may be applied by government agencies with their specific regulatory or political mission. The research domains are broadly defined as: quality of service, utilization, voucher targeting, and program performance. Quality will be assessed from the patient perspective as well as objective assessments of provider skill. Utilization data will show which OBA facilities are performing well, provide insights on facility characteristics that may be strengthening or challenging service delivery, and indicate whether the program is reaching a desired number of patients. For targeting, donors and managers are keen to know whether there is “leakage” of the subsidy - in the OBA scheme leakage occurs when a large fraction of the voucher patients could have otherwise afforded the full cost of the service. Program performance indicators include incidence of discovered fraud and degree of competition in local healthcare markets. See “Report on Initial Findings from 2006 Baseline Survey” for results of the household survey.