Examining the Long-term Spillover Effects of a Pay-forPerformance Program in a Healthcare System That Lacks Referral Arrangements

Background  Several studies have examined the intended effects of pay-for-performance (P۴P) programs, yet little is known about the unintended spillover effects of such programs on intermediate clinical outcomes. This study examines the long-term spillover effects of a P۴P program for diabetes care.Methods  This study uses a nationwide population-based natural experimental design with a ۳-year follow-up period under Taiwan’s universal coverage healthcare system. The intervention group consisted of ۷۶۸۸ patients who enrolled in the P۴P program for diabetes care in ۲۰۱۷ and continuously participated in the program for three years. The comparison group was selected by propensity score matching (PSM) from patients seen by the same group of physicians. Each patient had four records: one pertaining to one year before the index date of the P۴P program and the other three pertaining to follow-ups spanning over the next three years. Generalized estimating equations (GEEs) with difference-in-differences (DID) estimations were used to consider the correlation between repeated observations for the same patients and patients within the same matched pairs.Results  Patients enrolled in the P۴P program showed improvements in incentivized intermediate clinical outcomes that persisted over three years, including proper control of glycated hemoglobin (HbA۱c) and low-density lipoprotein cholesterol (LDL-C). We found a slight positive spillover effect of the P۴P program on the control of non-incentivized triglyceride [TG]). However, we found no such effects on the non-incentivized high-density lipoprotein cholesterol (HDL-C) control.Conclusion  The P۴P program has achieved its primary goal of improving the incentivized intermediate clinical outcomes. The commonality in production among a set of activities is crucial for generating the spillover effects of an incentive program.

Keywords:

Pay-For-Performance , Spillover Effect , Multitasking , Diabetes Mellitus , Intermediate Clinical Outcome