Abstract TP321: Is Weekend Admission Associated With Increased 30-day Stroke Mortality? The Effect of Adjustment Using a Claims-Based Stroke Severity Index

Abstract

Introduction: Whether weekend admission is associated with increased stroke mortality in Taiwan remains uncertain, partly because of an inadequate case-mix adjustment in other studies using an insurance claims databases. Hypothesis: Adding the 7-item claims-based stroke severity index (SSI) to a multivariate logistic regression model might alter the analysis of the effect of weekend admission on 30-day stroke mortality. Methods: We identified, in the Taiwan Longitudinal Health Insurance Database, which is linked with the National Death Registry, patients hospitalized for acute ischemic stroke between 2001 and 2013. The primary outcome was mortality 30 days post-admission. In base logistic regression models with and without the SSI, we tested the odds ratio (OR) of 30-day mortality in patient admitted on weekends using the covariates of age, sex, year of admission, Charlson’s comorbidity index, brain surgery, physician specialty and surgical volume, hospital ownership, accreditation, and patient volume. Results: We analyzed 46,007 consecutive hospitalized stroke patients (mean age: 68.8 ± 12.0 years; male: 59%), with an SSI of 7.5 ± 5.3 (range: 4.1-27.1), 23.0% were admitted on the weekend, and 4.2% died within 30 days. Patients who died within 30 days were more likely to have been admitted on a weekend (4.9% vs. 4.0%, p < 0.001). Nevertheless, patients admitted on a weekend had a higher SSI than those admitted on a weekday (7.8 vs. 7.4, p < 0.001). In multivariate logistic regression models, weekend admission was associated with 30-day mortality (OR: 1.22, 95% CI: 1.10-1.35) in the base model but not in the base model plus SSI (OR: 1.07, 95% CI: 0.95-1.20). Conclusions: We confirmed that, after stroke severity had been adjust by adding the SSI, weekend admission did not increase the 30-day mortality of stroke patients in Taiwan. A case-mix adjustment in comparative outcome studies of stroke patients is important when using an insurance claims database.