Background: Pay-for-performance (P4P) has been widely adopted, and increasingly recognized in intervening management of chronic diseases. However, due to the heterogeneity of P4P settings as well as demographic differences, results are generally inconsistent and controversial. Previous reviews either failed to focus on P4P, studying the whole concept of financial incentive in the diabetes field or insufficiently synthesized results.
Objectives: To explore whether P4P positively influences quality indicators of diabetes mellitus and the size of the effects and to evaluate the quality of the body of evidence for each relevant indicator using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system.
Methods: Databases including Ovid MEDLINE, EMBase, PubMed, the Cochrane Library were comprehensively searched for the effects of P4P programs in terms of patient outcomes and physician behaviors. Studies covering detailed data were included and synthesized. The quality of the body of evidence for each quality indicator was determined using the GRADE system.
Results: From 492 identified articles, 16 interrupted time series studies, four controlled before-after studies and one quasi-experiment study were included. (Figure 1) Twelve studies were also included for quantitative analysis. Results of meta-analysis showed that P4P produced a generally positive effect in most indicators (e.g. patients with record of total cholesterol or blood pressure). However, these results were inconsistent. The percentage of patients with HbA1c ≤7% or 53 mmol/mol showed a pooled odds ratio of 0.98 in patients, but a pooled mean difference of 19.71% in the physicians. The odds ratios of receiving tests/reaching an outcome level were also diverse in patients (OR ranged from 0.98 to 3.32). Besides, process indicators had higher rates of improvement than outcome indicators. (Tables 1-6)
Conclusions: P4P has variable impacts on patient outcomes of diabetes as well as physician behaviors, with various effects from negligible to strongly beneficial. Considering the low to medium quality of included studies, the conclusion should be cautiously interpreted.