High risk medication regimens and medication related predictors of hospital readmission in elderly home care patients.

Abstract

Adverse drug events are a primary cause of hospitalization in the elderly. Nearly 70% of the $177.4 billion dollars spent on drug related morbidity and mortality in the U.S. is due to hospitalizations. Polypharmacy, inappropriate medications or medication regimen complexity have all been implicated as precursors to adverse drug events and as indicators of high risk medication regimens. Understanding the relationship between medication regimens and readmission is important when evaluating potential errors in administration, risk-benefit ratios, and readmission risk. However, due to definitional and measurement issues, the high risk medication regimen remains an elusive concept. This study characterizes medication regimens, defines high risk medication regimens, and determines if high risk medication regimens predict re-hospitalization in home healthcare clients over age 65. An exploratory, secondary analysis of OASIS data and medication records from 15 home care agencies was used to characterize medication use in 911 older adults discharged from the hospital to their first episode of home care in 2004. Conceptual and operational definitions of polypharmacy, potentially inappropriate medications, medication regimen complexity, and high risk medication regimens were developed. Logistic regression and structural equation modeling were used to examine the relationship between comorbidity, a variety of risk factors supported by the literature, high risk medication regimens (defined by polypharmacy, potentially inappropriate medication regimens, and medication regimen complexity) and re-hospitalization to determine if high risk medication regimens predicted rehospitalization in these subjects. Factor analysis revealed that high risk medication regimens are composed of polypharmacy, potentially inappropriate medication regimens, and medication regimen complexity, and that a model using this concept rather than individual medication variables proved to be the most predictive and parsimonious model. The model accounted for 10% of variance in re-hospitalization in this sample. Additionally, high risk medication regimens appear to have as much influence as comorbidity on hospital readmission. Future research should include high risk medication regimens as a predictor of readmission and previously completed studies may need to be re-evaluated in light of these findings. Both the findings and the methodology will be useful in examining predictive potential of high risk medications regimens in other settings.