• inappropriate medication;
  • elderly;
  • Medicare Part D;
  • Beers criteria

OBJECTIVES: To empirically estimate changes of potentially inappropriate medication (PIM) use attributable to the Medicare Part D prescription drug benefit.

DESIGN: Difference-in-difference strategy in the quasi-experimental design with a control group.

SETTING: U.S. nationally representative community-dwelling sample of older adults.

PARTICIPANTS: One thousand seven hundred seventy-four adults aged 65 and older in the 2005 and 2006 Medical Expenditure Panel Surveys were followed up for 2 years with five rounds of interviews.

MEASUREMENTS: PIM use was identified based on the 2002 Beers criteria. Analyses were conducted for likelihood of PIM use and number of PIM prescriptions using logit models and negative binomial models, respectively.

RESULTS: There was a trend of less likelihood of PIM use for all older adults from 2005 to 2006 (odds ratio=0.67, 95% confidence interval (CI)=0.52–0.86). After accounting for this secular trend and potential confounders, no significant difference of the likelihood of PIM use was found between Part D enrollees and nonenrollees, although enrollees were found to use significantly more PIM prescriptions in round 5 (in 2006) than nonenrollees (incidence rate ratio=1.56, 95% CI=1.08–2.25).

CONCLUSION: This initial evidence suggests that Medicare Part D could result in more PIM use in older enrollees than in nonenrollees, although the overall likelihood of PIM use has decreased in all older community-dwelling adults. Future research is needed to examine the effect over the longer term and focusing on particular categories of PIMs.