Predictors of Antipsychotic Medication Use in Nursing Home Residents

Abstract

The prevalence of antipsychotic medication use in nursing home patients continues to be substantial and may be increasing despite restrictions on their use (Briesacher et al., 2005). These medications are indicated for psychotic symptoms such as delusions and hallucinations, but are not well-supported for use in controlling problematic behaviors associated with cognitive impairment. A 2004 study found that although diagnoses related to psychotic symptoms were strongly predictive of antipsychotic use, there were other personal and clinical predictors, such as a diagnosis of dementia and behavioral symptoms, that might contribute to inappropriate antipsychotic use (Kamble, Chen, Sherer, & Aparasu, 2008). To our knowledge, no studies have used validated scales measuring the severity of behavioral symptoms and cognitive impairment to examine the degree to which these factors might be associated with antipsychotic use. Two-hundred eighty nursing home patient medical records were randomly selected from nineteen nursing homes, and data were extracted from the Minimum Data Set 2.0 (MDS 2). We used logistic regression analyses to predict use of antipsychotics in the previous seven days. We controlled for age because prior research suggests that younger patients are more likely to receive antipsychotics than older patients. The severity of cognitive impairment was measured with the validated Cognitive Performance Scale (CPS) (Morris et al., 1994), and the severity of behavioral symptoms was measured using the validated Aggressive Behavior Scale (ABS) (Perlman & Hirdes, 2008). We also included a facility-level variable, the number of licensed nursing staff hours per resident per day as recorded by the Centers for Medicare and Medicaid Services' website. We found severity of cognitive impairment (OR = 1.35, 95% CI, 1.05-1.74), severity of behavioral symptoms (OR = 1.38, 95% CI, 1.01-1.88), and age (OR = .97, 95% CI, .95-.99) to be predictive of antipsychotic use. We found the number of staffing hours (OR = .98, 95% CI, .97-1.00) to be nearly significant in the prediction of antipsychotic use. Descriptive statistics, however, indicate that two of nineteen facilities contributed forty percent of people on an antipsychotic. The results support the notion that cognition and behaviors are important drivers of antipsychotic use and that facility characteristics and staffing should be further explored as a possible issue related to this outcome.