Pay-for-Performance Quality Measures: Why Do New Physicians Score Lower?

Background: Public reporting of quality metrics for organizational benchmarking is well-established. Most metrics are well-specified, eg, the proportion of patients with diabetes whose hemoglobin A1c is at target or the proportion of patients in a specific age range having colorectal cancer screening. Patients become eligible for inclusion by developing a condition (diabetes), achieving a specific age (screening), or joining the organization — with an undocumented history of prior care. Newly hired physicians may have relatively more of the latter than established physicians with stable panels. We examine whether adjusting for how patients “enter” a panel impacts physician-level quality scores. Methods: We used administrative and electronic health record data of a large ambulatory group practice. Our study included 389 primary care physicians (PCPs) in 2011–2014 (1,261 PCP-years). We examined adult primary care practice quality metrics for diabetes management and screening services (colorectal/breast/cervical cancer screening and chlamydia). A quality score (number of patients who met the target divided by number eligible) was constructed for each metric. Patients were classified into three groups: 1) new to the organization, 2) newly eligible for screening or newly diagnosed with diabetes, and 3) continuing eligible patients. We compared each set of quality scores for the three patient groups. Adjusted panel-based scores (ratio of observed score to expected score based on patient mix) were compared between newly hired and established PCPs. Results: Quality scores varied across the patient groups (eg, 57% vs76 % vs 63% for hemoglobin A1c control, 37% vs 44% vs 73% for colorectal cancer screening, 60% vs 44% vs 50% for chlamydia screening, at 6th-month post-enrollment/diagnosis for new and 6th calendar-month for continuing patients). The “quality gaps” between new and continuing patients were wider shortly after enrollment or diagnosis but narrowed over time. In general, new PCPs had lower scores than established PCPs, but the differences were reduced after this adjustment for patient mix. Conclusion: Time since a patient is eligible for a quality metric impacts the likelihood of that patient being at goal. When assessing quality metrics across physicians within an organization, the composition of the physicians’ panels should be considered