Distributional effects of payment for performance in the health sector. Examining effects on structural quality, performance outcomes and service utilisation in Tanzania

Introduction: Payment for performance (P4P) involves the allocation of financial incentives to health workers and/or facilities for reaching pre-defined performance targets or measures. P4P has been used in high-income countries (HICs) to improve healthcare quality, and recently has been applied in low-and middle-income countries (LMICs) to improve the coverage and quality of health services and strengthen health systems. The available evidence on the effect of P4P is mixed, but with some promising results of improvements in the incentivised indicators. However, most evaluations of P4P have focused on average programme effects on the incentivised services, paying little attention to distributional effects of P4P. Specifically, little is known about the effects of P4P on structural quality of care (e.g. availability of medical commodities), and similarly on the understanding of the heterogeneity of the P4P effects among subgroups of providers and populations. This PhD work aims to fill that knowledge gap. It estimates the effect of P4P on the availability and stock-out of medical commodities, and examine the differential effects of P4P across subgroups of health facilities and populations in Tanzania. Data sources: The study collected data in intervention and control areas through facility and household surveys, and facility payments data from administrative records. Baseline data were collected in January 2012 with a follow-up 13 months later. Facility survey across 150 facilities (75 facilities from each study arm) included data on the availability and stock-out of medical commodities (drugs, supplies and equipment), and facility characteristics. Household survey across 3000 women who delivered within 12 months prior to the survey (20 women per facility catchment area), and a similar sample size in the follow-up survey, captured information about individual and household characteristics and maternal and child health service utilisation. Analyses: A difference-in-differences (DID) regression model was used to estimate the average effects of P4P on the availability and stock-out of medical commodities (Paper I). The DID model was further extended by including a three-way interaction term (i.e. average effect and subgroup indicator) to capture the differential effects of P4P across facilities’ subgroups (Paper I and II), and across populations subgroups (Paper III). Assessment of differential effects were based on outcomes which improved significantly due to P4P (i.e. availability of drugs and supplies, institutional delivery rates and uptake/ provision of antimalarial drugs during antenatal care (ANC)). Descriptive measures of inequality were also used to assess the distribution of facility payouts across facilities’ subgroups (Paper II). Results: Paper I reports that P4P improved the availability of drugs and supplies and reduced their stock-out rates, but had no effect on the availability of medical equipment. The improved effects were generally similar across facilities, but relatively higher among facilities serving a poor population and located in rural areas. Paper II finds that facility payments were initially higher among higher level facilities (hospitals and health centres than dispensaries), the better resourced than worse resourced facilities, and facilities serving wealthier than poorer populations, but these inequalities in payouts declined over time. The effect of P4P on institutional delivery rates was greater among facilities with low baseline performance, serving middle wealth populations, located in rural areas, than among their counterparts; whereas the effect on provision of antimalarial drugs was similar across facilities subgroups. Paper III finds that the effect of P4P on institutional deliveries was greater among women in the poorest households, who lived in rural areas and who did not have health insurance than among their counterparts. P4P effect on the uptake of antimalarial drugs was equally distributed across population subgroups. Conclusion: The study findings suggest that the P4P programme can improve structural quality of care in terms of the availability of medical commodities. It can further enhance more equitable performance among facilities as the worse-off providers improved most in this study. Similarly, P4P can enhance equitable service utilisation since the service use increased mostly among the worse-off populations