The Use of Physician Financial Incentives and Feedback to Improve Pediatric Preventive Care in Medicaid Managed Care

METHODS

This randomized, controlled trial evaluated the impact of feedback, with or without financial incentives, on physician compliance with pediatric preventive care guidelines. The study population included primary care physician practices serving pediatric members of a managed care plan in Philadelphia. We conducted the study between 1993 and 1995 with Healthcare Management Alternatives, Inc (HMA), a Medicaid plan structured like an Independent Practitioner Association, with primary care physicians paid by capitation.

At the start of the study, we identified all primary care physician sites in HMA's network having at least 25 pediatric members up to the seventh birthday. We excluded smaller pediatric sites to increase the likelihood that an adequate number of charts would be available for audit, and to focus the study on those sites with larger pediatric panels, which would be more likely to respond to feedback and/or financial incentives related to compliance with pediatric care guidelines. In total, 53 sites were randomly assigned to one of three study groups: feedback and incentive (FB+I), feedback only (FBO), or control. We stratified the randomization by practice type (solo/group) to ensure adequate representation of each. Four sites—2 in the FBO group and 2 in the control group—discontinued participation in the health plan during the study. Thus, the final sample included 49 sites (19 FB+I, 15 FBO, 15 control).

The health plan's pediatric preventive care guidelines, chart audit tools to assess physician compliance with these guidelines, and audit result feedback reports were used for the study. The study focused on compliance with guidelines for children between the first and seventh birthdays (ages 1 through 6). HMA's preventive care guidelines (Table 1) were adapted from recommendations by the US Preventive Services Task Force and the Philadelphia Department of Health, and were reviewed and approved by the health plan's peer review committees, to increase physician acceptance and buy-in. The guidelines were distributed in brochure form to providers in all three study groups shortly before the study began.

Pediatric preventive care audit tools were developed by HMA to assess compliance with the guidelines. Chart audits were performed for practice sites in all three groups at 6-month intervals: baseline and three follow-up points, for a total of four audits during 18 months. For each audit, trained nurse-abstractors used by the health plan contacted the site, scheduled the on-site record review, and requested a specific random sample of charts for members ages 1 through 6 years, who had been enrolled with the provider for at least 6 months. The chart request lists were generated by the health plan's information systems department. At each audit, the HMA nurse reviewed 15 records, using a standardized tool. The nurse-reviewers were blinded to which sites were in each of the three study groups. Intrarater reliability tests indicated a high degree of reproducibility (r ² = 0.89).

Audit data were entered into a scoring program developed by HMA, to produce summary reports and compliance scores. For each site, compliance scores were calculated for each of the indicators presented in Table 1, as the number of reviewed charts for which the indicator was met, divided by the number of charts for which the indicator was applicable. For example, if 12 of the 15 reviewed charts were for children >age 2, and 6 of these had blood pressure recorded at least once per year after age 2, the compliance score for blood pressure would be 50% (6/12). Compliance for immunizations was defined as receipt of vaccines on or before the maximum recommended age, with a 2-month grace period. A total compliance score also was calculated as the total number of indicators met across all charts, divided by the total number of applicable indicators across those same charts.

Sites that were randomized to the control group were audited at 6-month intervals, but did not receive feedback reports or financial incentives. Sites in the FBO group received a feedback report after each audit, showing their performance (compliance scores on each indicator and total compliance score) in comparison with the performance of all audited sites. The FBO group letter for the first cycle of feedback also included an explanation that the health plan would now be conducting audits and providing feedback reports on a semiannual basis.

The FB+I sites received semiannual feedback, plus a cover letter with the feedback report notifying them if they had qualified for a bonus based on their performance. At the start of the study, a letter was mailed to all physicians at the FB+I sites notifying them that HMA was starting a new quality-based financial incentive system linked to pediatric preventive care audit scores. The details of the bonus and the rules for qualifying were explained. Eligibility for bonuses was based on the total compliance score. Bonus eligibility required a minimum compliance score of 20% for each indicator. The 3 sites with the highest total compliance scores received a full bonus (20% of the site's total 6-month capitation for pediatric members up to the seventh birthday); the 3 next best scoring sites received a partial bonus (10% of capitation); and the 3 sites showing the most improvement from the last audit, also received this partial bonus, provided that their total compliance score increased by at least 10%. These percentages were based on a previous survey of managers that suggested bonus levels that would garner physicians' attention.¹³ A minimum bonus amount was established ($500) in consideration of smaller sites.

Although randomization, auditing, and bonus distribution were site-specific, we directed all correspondence to individual physicians in the FB+I and FBO groups to maximize awareness of the study. To determine levels of awareness, we surveyed all intervention group physicians after baseline audits and introductory letters. The survey included a brief description of the program and questions about physician awareness. A second mailing targeted only nonresponders and responders who reported being unaware of the program.

An analytic file was created containing information on characteristics of the sites in each study group (number of providers, specialty, HMA panel size, and so forth) and audit results at baseline and each of the three follow-up points. Eight sites had a missing value for one audit. We used mean values from the respective study group to impute values for these sites. Other estimation techniques, such as regression imputation, did not produce significantly different results. Imputation did not affect our conclusions, but did improve the power of the study.

Statistical testing included χ² and Student'st tests for descriptive comparisons, and repeated measures ANOVA with the site as the unit of analysis for between-group effect, time effect, and group-by-time interaction.