Frequency and Predictors of Polypharmacy in US Medicare Patients: A Cross-Sectional Analysis at the Patient and Physician Levels

Background: Polypharmacy in older patients increases the risk of medication-related adverse events and can be a marker of unnecessary care. Objectives: The aim of this study was to describe the frequency of polypharmacy among patients 65 years of age or older and identify factors associated with the occurrence of patient-level and physician-level polypharmacy. Methods: We performed a cross-sectional analysis of 100% Medicare claims data from January 1, 2016 to December 31, 2016. All patients with continuous Medicare coverage (Parts A, B, and D) throughout 2016 who were 65 years of age or older and who were prescribed at least one medication for at least 30 days were included in the analysis. Each patient was attributed to the primary care physician who prescribed them the most medications. Physicians treating fewer than ten patients were excluded. We defined polypharmacy based on the highest number of concurrent medications at any point during the year. We used hierarchical linear regression to study patient- and physician-level characteristics associated with high prescribing rates. Results: We identified 25,747,560 patients attributed to 147,879 primary care physicians. The patient-level mean [standard deviation (SD)] concurrent medication rate was 5.6 (3.3), and the physician-level mean (SD) was 5.6 (1.1). A total of 6108 physicians (4.1% of sample) had a mean concurrent number of medications greater than two SDs above the physician-level mean. At the patient level in the adjusted model, a history of HIV/AIDS, diabetes mellitus, solid organ transplant, and systolic heart failure were the comorbidities most strongly associated with polypharmacy. The relative difference in number of medications associated with these comorbidities were 1.89, 1.39, 1.32, and 1.06, respectively. At the physician level, increased time since medical school graduation and smaller practice size were associated with lower rates of polypharmacy. Conclusions: Patterns of high prescribing to older patients is common and measurable at the physician level. Addressing high outlier prescribers may represent an opportunity to reduce avoidable harm and excessive costs.