Inappropriate use of nonpsychotropic medications in nursing homes.

OBJECTIVES To determine the prevalence and patient-specific predictors of the use of 10 presumptively inappropriate medications used to treat medical conditions among nursing home residents, and to use this information to examine alternative screening strategies using computerized assessment data to identify residents who are at high risk of receiving inappropriate medications. DESIGN Retrospective, cross-sectional study. PATIENTS All persons residing in all 252 nursing homes in two states during the last 6 months of 1991 (N = 21,884). MEASUREMENTS Data were from Minimum Data Set Plus (MDS+) assessments, gathered as part of the Health Care Financing Administration (HCFA) Multistate Nursing Home Casemix and Quality Demonstration Project. The MDS+ is an expanded version of the federally mandated Minimum Data Set (MDS) that includes additional information on medications and their doses and schedules (frequency, standing vs prn). The reliability of the MDS has been demonstrated previously. Medications were defined as inappropriate using explicit criteria from published literature. Outcome measures were the standing use of each or any of 10 presumptively inappropriate medications used to treat medical (rather than psychiatric or behavioral) conditions. Potential predictors of inappropriate medication use included patient demographic characteristics, payer, a proxy measure for length of stay and admission source, functional status, number of standing medications, and state. MAIN RESULTS A total of 12% of residents were prescribed one or more of 10 presumptively inappropriate medications on a standing basis, a figure that differed substantially between states (14.0% vs 7.4% (P < .001)). The most prevalent inappropriate medications were dipyridamole (5.4% of residents), amitriptyline (3.3%), and methyldopa (1.8%). Among patients receiving 0 to 3, 4 to 6, and 7+ medications, 5%, 12%, and 19%, respectively, were receiving at least one inappropriate medication. In multivariate logistic regression analyses, the strongest predictors of inappropriate medication use were state and the total number of standing medications prescribed. Including other statistically significant predictors of inappropriate medication use (age > 65 years, never having been married, severe functional limitations, being a long-stay patient, and medical diagnosis) did not substantially improve the overall predictive ability of the model. CONCLUSIONS A substantial proportion of nursing home residents receives presumptively inappropriate medications to treat medical conditions. Selecting persons prescribed large numbers of medications for further review may be the most efficient method for nursing home or pharmacy personnel to identify residents at high risk of receiving inappropriate medications. Extensive additional information on residents' characteristics, although widely available through the Minimum Data Set, does not significantly improve the ability to identify residents receiving inappropriate medications for medical conditions. State-specific policies or provider practices also influence the likelihood of presumptively inappropriate medication use among nursing home residents and deserve further investigation.