Association of preventable adverse drug events with inpatients' length of stay-A propensity-matched cohort study.

PURPOSE: Using clinical administrative data (CAD) of inpatients, we aimed to identify ICD-10 codes coding for potentially preventable inhospital adverse drug events (ADE) that affect the length of hospital stay (LOS) and thus patient well-being and cost.

METHODS: We retrospectively assessed CAD of all inpatient stays in 2012 of a German university hospital. Predefined ICD-10 codes indicating ADE (ADE codes) were further specified based on expert ratings of the ADE mechanism and ADE preventability in clinical routine to particularly identify preventable inhospital ADE. In a propensity-matched cohort design, we compared patients with one or more ADE codes to control patients with regard to differences in LOS for three situations: all cases with an ADE code, cases with an inhospital ADE code, and cases with a preventable inhospital ADE code.

RESULTS: Out of 54 032 cases analysed, in 8.3% (N=4 462) at least one ADE code was present. Nine of 128 evaluated ADE codes were rated as preventable in clinical routine, relating to 220 inpatients (4.9% of all identified inpatients with at least one ADE code and 0.4% of the entire cohort, respectively). Out of 48 072 evaluable inpatients for propensity score matching, 7 938 controls without ADE code and 4 006 cases with ADE code were selected. In all three settings, cases showed prolonged LOS vs controls (delta 1.13 d; 0.88 d and 1.88 d, respectively), significantly exceeding the maximum LOS as defined for each Diagnosis-Related Group.

CONCLUSION: Inpatients with ADE codes referring to inhospital, potentially preventable ADE exceeded the maximum hospital stay fully reimbursed by insurance companies, indicating unnecessary long and costly inpatient stays.