Characterizing Potential Medication Related Harm in the 30-Days Following Hospitalization Using Linked Provincial Administrative Health Data

Abstract

Introduction Patients transitioning between hospital and home are at risk of experiencing adverse drug events. Hospital-based care teams require a better understanding of how therapeutic decisions and patient risk factors impact adverse health outcomes following discharge. Objectives and Approach The objective was to describe medications prescribed at discharge and quantify the association between these medication classes and risk of emergency department (ED) visits within 30 days. A retrospective, population-based cohort study was conducted using linked health administrative databases of adults discharged from hospital to the community between April 2016-March 2017, in Ontario, Canada. To distinguish medications prescribed at discharge from those by community providers, we examined medications dispensed in the 30-days following hospitalization but prior to another healthcare encounter (physician visit, ED visit, or rehospitalization). Adjusted logistic regression was used to determine the association between medication classes at discharge and ED visits in 30 days post discharge. Results Our study cohort included 214, 011 patients. Median age was 76 (IQR:66-82), 52% were females, 92% were discharged home and 8% were discharged to long term care. Patients were dispensed a median of 3 medications (IQR: 2-6). The most common medications included hydromorphone (16% of all patients, 71% of which represented new use), pantoprazole (15% of patients, 40% new use), furosemide (13% of patients, 34% new use), amlodipine (10% of patients, 35% new use) and oxycodone (9% of patients, 70% new use). To account for confounding by indication, we adjusted for age, sex, discharge service, discharge disposition, in hospital diagnoses and length of stay. Medications which were most highly associated with an increased risk of ED visits included antipsychotics, antiepileptics, anti-infectives, systemic corticosteroids, antihypertensives and cardiac stimulants. Conclusion / Implications Characterizing potential medication related harm in the 30-days following hospitalization is an important first step in understanding which patients may benefit most from hospital-based medication related interventions.