The association between physician knowledge and inappropriate medications for older populations.

Abstract

Older patients are often prescribed potentially inappropriate medications (PIMs) given their age. We measured the association between a physician's general knowledge and their PIM prescribing. Using a 2013-2017 cross-sectional design, we related a general internist's knowledge (n=8196) to their prescribing of PIMs to fee-for-service Medicare beneficiaries, age ≥ 66 years with part D coverage, which they saw in the outpatient setting the year after their exam (n=875,132). Physician knowledge was based on the American Board of Internal Medicine's (ABIM) Internal Medicine Maintenance of Certification (IM-MOC) exam scores. Medications included 72 PIMs from the American Geriatric Society's Beers Criteria and appropriate alternatives to these medications. Logistic regressions controlled for physicians practice/training characteristics and patient-risk factors. Annually, 11.0% of patients received a PIM and 57.2% received an appropriate alternative medication. Patients seen by physicians scoring in the top versus bottom quartile were 8.6% less likely (95% confidence interval [CI]: -12.7 to -4.5, p < 0.001) to be prescribed a PIM and 4.7% more likely (95% CI: 1.7 to 7.6, p=0.001) to be prescribed an appropriate alternative medication. The difference in PIM prescribing grew to 12.1% fewer (95% CI: -15.1 to -9.1) patients when limiting the sample to the 58.9% of patients being prescribed a PIM or appropriate alternative medication. Among patients receiving any medication, this was similar to the percent difference in PIM prescribing between solo and large practices (≥50 physicians, -10.2%, 95% CI: 13.6-6.5, p < 0.001) or between group and academic practices (-11.7%, 95% CI: -15.3 to -7.9, p < 0.001). PIM prescribing was more positively associated with patient characteristics including age, gender, and total number of medications prescribed. Better physician general knowledge, as measured by an ABIM exam, was associated with fewer PIM prescriptions. Future research should examine whether general educational interventions, such as MOC, effect PIM prescribing.