Objectives: To investigate the effectiveness of a pay-for-performance program (P4P) to increase the receipt of quality care and to decrease hospitalization rates among patients with diabetes mellitus.
Study design: Longitudinal study of patients with diabetes enrolled in a preferred provider organization (PPO) between January 1, 1999, and December 31, 2006.
Methods: We used multivariate analyses to assess the effect of seeing P4P-participating physicians on the receipt of quality care (ie, glycosylated hemoglobin and low-density lipoprotein cholesterol testing) and on hospitalization rates, controlling for patient characteristics.
Results: Patients with diabetes who saw P4P-participating physicians were more likely to receive quality care than those who did not (odds ratio, 1.16; 95% confidence interval, 1.11-1.22; P <.001). Patients with diabetes who received quality care were less likely to be hospitalized than those who did not (incident rate ratio, 0.80; 95% confidence interval, 0.80-0.85; P <.001). During 1 year, there was no difference in hospitalization rates between patients with diabetes who saw P4P-participating physicians versus those who did not. However, patients with diabetes who saw P4P-participating physicians in 3 consecutive years were less likely to be hospitalized than those who did not (incident rate ratio, 0.75; 95% confidence interval, 0.61-0.93; P <.01).
Conclusions: A P4P can significantly increase the receipt of quality care and decrease hospitalization rates among patients with diabetes in a PPO setting. Although it is possible that the differences observed between P4P-participating physicians and non-P4P-participating physicians were due to selection bias, we found no significant difference in the receipt of quality care between patients with diabetes who saw new P4P-participating physicians versus non-P4P-participating physicians during the baseline year. Further research should focus on defining the effect of P4Ps on intermediate outcomes such as glycosylated hemoglobin and low-density lipoprotein cholesterol levels.