IMPACT OF MEDICATION CLUSTERS ON POST-DISCHARGE ADVERSE DRUG EVENTS AND HEALTHCARE UTILIZATION AMONG OLDER ADULTS
Author(s)
Haowen Hsu, PharmD, MPH1, Gabriel Gazetta, MS2, Chi-Hua Lu, PharmD, MS1, David Jacobs, PharmD, PhD1.
1Department of Pharmacy Practice, School of Pharmacy and Pharmaceutical Sciences, University at Buffalo, Buffalo, NY, USA, 2Department of Industrial and Systems Engineering, School of Engineering and Applied Sciences, University at Buffalo, Buffalo, NY, USA.
1Department of Pharmacy Practice, School of Pharmacy and Pharmaceutical Sciences, University at Buffalo, Buffalo, NY, USA, 2Department of Industrial and Systems Engineering, School of Engineering and Applied Sciences, University at Buffalo, Buffalo, NY, USA.
OBJECTIVES: Medications at hospital discharge may substantially influence post-discharge adverse drug events (ADEs) and healthcare utilization (HCU) among older adults. Evidence is limited on how discharge medication clusters, rather than single agents, jointly affect these outcomes. This study characterizes discharge medication clusters and assesses their associations with 30-day ADEs and HCU using a regional hospital electronic health record (EHR).
METHODS: We extracted EHR data from a major hospital system in Western New York between 1/1/2020 to 12/31/2024. Adults ≥65 years with ≥1 hospitalization were included. The index admission was the first hospitalization with ≥1 discharge medication and ≥183 days of prior EHR history. Individuals were followed for 30 days to capture ADEs defined by ICD-10 codes, emergency department (ED) visits, and readmissions. Discharge medications were mapped to Anatomical Therapeutic Chemical (ATC) codes at the pharmacological/therapeutic level and clustered using latent class analysis. The Bayesian information criterion and entropy guided cluster selection. Clusters were labeled based on ATC composition following the study team's review. Multivariable logistic regression, adjusted for demographics, previous HCU, and comorbidities, estimated odds ratios (ORs) with 95% confidence intervals, and analyses used SAS 9.4.
RESULTS: Among 29,123 older adults, six discharge medication clusters were identified: inflammatory/respiratory (30.6%), pain management (29.0%), infection (13.5%), cardiovascular (10.4%), antithrombotic (9.9%), and multimorbidity (6.6%). The antithrombotic cluster was associated with a lower odds for 30-day ED visit (OR= 0.77 [0.63-0.94], p=0.01). For 30-day readmission, the inflammatory/respiratory (OR=1.18 [1.07-1.30], p=0.01), multimorbidity (OR=1.48 [1.28-1.72], p<0.01), and cardiovascular (OR=1.26 [1.11-1.42], p<0.01) clusters showed higher odds. Inflammatory/respiratory (OR=1.16 [1.04-1.30], p<0.01) and multimorbidity (OR=1.39 [1.17-1.65], p<0.01) clusters were associated with increased 30-day ADEs.
CONCLUSIONS: Distinct discharge medication clusters were differentially associated with 30-day ADEs and HCU. This identifies opportunities for tailored interventions targeting specific medication profiles to reduce preventable adverse outcomes among older adults during care transitions.
METHODS: We extracted EHR data from a major hospital system in Western New York between 1/1/2020 to 12/31/2024. Adults ≥65 years with ≥1 hospitalization were included. The index admission was the first hospitalization with ≥1 discharge medication and ≥183 days of prior EHR history. Individuals were followed for 30 days to capture ADEs defined by ICD-10 codes, emergency department (ED) visits, and readmissions. Discharge medications were mapped to Anatomical Therapeutic Chemical (ATC) codes at the pharmacological/therapeutic level and clustered using latent class analysis. The Bayesian information criterion and entropy guided cluster selection. Clusters were labeled based on ATC composition following the study team's review. Multivariable logistic regression, adjusted for demographics, previous HCU, and comorbidities, estimated odds ratios (ORs) with 95% confidence intervals, and analyses used SAS 9.4.
RESULTS: Among 29,123 older adults, six discharge medication clusters were identified: inflammatory/respiratory (30.6%), pain management (29.0%), infection (13.5%), cardiovascular (10.4%), antithrombotic (9.9%), and multimorbidity (6.6%). The antithrombotic cluster was associated with a lower odds for 30-day ED visit (OR= 0.77 [0.63-0.94], p=0.01). For 30-day readmission, the inflammatory/respiratory (OR=1.18 [1.07-1.30], p=0.01), multimorbidity (OR=1.48 [1.28-1.72], p<0.01), and cardiovascular (OR=1.26 [1.11-1.42], p<0.01) clusters showed higher odds. Inflammatory/respiratory (OR=1.16 [1.04-1.30], p<0.01) and multimorbidity (OR=1.39 [1.17-1.65], p<0.01) clusters were associated with increased 30-day ADEs.
CONCLUSIONS: Distinct discharge medication clusters were differentially associated with 30-day ADEs and HCU. This identifies opportunities for tailored interventions targeting specific medication profiles to reduce preventable adverse outcomes among older adults during care transitions.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
EPH178
Topic
Epidemiology & Public Health
Topic Subcategory
Safety & Pharmacoepidemiology
Disease
No Additional Disease & Conditions/Specialized Treatment Areas