Abstract 19489: Pharmacological Factors Associated With Congestive Heart Failure Hospital Readmission: A Case-Study Using 15,768 Heart Failure Patients From Two Health Systems
Introduction: Reduction of congestive heart failure patient hospital readmissions remains challenging despite much effort. Pharmacotherapy is known to be associated with readmission, but past work has not established if related factors also contribute or the degree that findings persist across health systems. We use data from two health systems to examine whether side effects and drug interactions are associated with CHF readmission.
Methods: A 14,348 patient cohort from Columbia University Medical Center (discovery) using ICD-9 codes and LVEF to identify patients and a separate 1,420 patient cohort with known CHF diagnosis from Mount Sinai Hospital (validation) were assembled. Drug adverse events were compiled from FDA side effects (Sider), predicted side effects (Offsides), and drug-drug interactions (Twosides). Results are controlled for age, sex, and race and FDR-adjusted for multiple testing.
Results: In the discovery cohort 78 drugs, 1169 FDA side effects, and 4165 predicted side effects were associated with 30-day readmission. In the validation cohort, four medications (ondansetron, OR=4.09; diphenhydramine, OR=4.84, insulin, OR=3.18; dextrose, OR=3.77) were associated. Two FDA side effects (increased creatinine, OR=0.49; hepatitis, OR=0.45) and 36 predicted side effects with were associated. Significant drug-drug interactions were not identified. Readmitted patients received more prescriptions than non-readmitted patients (16.9 vs. 11.5, p<10-6). Administration route, route changes, timing, or drug classes were not associated. Discovery cohort FDA side effects and drug exposures did not intersect with our validation cohort. 14/36 (38.9%) validation cohort predicted side effects were replicated, including findings like decreased CD8+ count (validation OR=4.41), heart-lung transplant rejection (4.23), and subendocardial ischemia (3.41).
Conclusions: Aspects of medications are associated with 30-day hospital readmission. Readmission odds associated with drugs may reflect underlying medical conditions instead of direct drug effects. Co-occurence of side effects across two health systems despite the lack of concurrence between drug exposures may suggest that common side effects contribute to readmission.
Author Disclosures: K.W. Johnson: None. K. Shameer: None. A. Yahi: None. R. Miotto: None. D. Ricks: None. J. Jebakaran: None. P. Kovatch: None. P. Sengupta: None. D.L. Reich: None. A. Kasarskis: None. S. Pinney: None. N. Tatonetti: None. J.T. Dudley: Ownership Interest; Significant; NuMedii Inc, Ayasdi, Inc, Ontomics, Inc.. Consultant/Advisory Board; Significant; Janssen Pharmaceuticals, GlaxoSmithKline, AstraZeneca, Hoffman-La Roche.
- © 2016 by American Heart Association, Inc.