Potentially inappropriate medication use in hospitalized elders

Abstract

Prescribing of potentially harmful medications has not been well documented in hospitals. The objective of the study was to determine the rate of and factors associated with potentially inappropriate medication (PIM) prescribing in a large inpatient sample. The study was a retrospective cohort of the period between September 1, 2002, and June 30, 2005. We used multivariable logistic regression to identify patient, physician, and hospital characteristics associated with PIM prescribing. The study collected data from 384 US hospitals. The sample was composed of patients aged >or=65 years admitted with 1 or more of 7 common medical diagnoses. The percentage of patients prescribed PIMs as defined using a modified Beers list was measured. Multivariable-adjusted odds ratios for PIM use were computed. Of the 493,971 patients, 49% received at least 1 PIM, and 6% received 3 or more, most commonly promethazine, diphenhydramine, and propoxyphene. Patient, physician, and hospital characteristics were all associated with PIM use. Patients with myocardial infarction or heart failure were most likely (61% and 52% vs. 46% for pneumonia), men (47% vs. 49% for women) and those in managed care plans (44% vs. 49% for other plans) were less likely, and patients >or=85 years were least likely (42% vs. 53% for patients aged 65-74 years) to receive PIMs (P < .0001 for all comparisons). For high-severity PIMs, internists and hospitalists had similar prescribing rates (33%), cardiologists had a higher rate (48%), and geriatricians had the lowest rate (24%). The proportion of elders receiving PIMs ranged from 34% in the Northeast to 55% in the South, and variation at the individual hospital level was extreme. At 7 hospitals, PIMs were never prescribed. Wide variation in the use of PIMs is associated with hospital and physician characteristics. Care may be improved by minimizing this non-patient-centered variation.