Abstract TMP80: Temporal Trends in Ischemic Stroke Mortality Stratified by Presence of Cancer

Abstract

Background: About 5% of acute ischemic stroke (AIS) patients have active cancer. Previous work has shown that cancer patients are more likely to die after AIS than non-cancer patients. However, few data exist on how these disparities are changing over time. Methods: We identified all patients hospitalized with AIS in the National Inpatient Sample from 1996-2013 using validated ICD-9 diagnosis codes. Our primary predictor was systemic cancer, which was a composite of solid and hematologic tumors with or without metastases. Primary brain tumors were excluded. Our primary outcome was in-hospital death. Survey weights were used to estimate nationally-representative annual rates of in-hospital death among AIS patients with and without cancer. Multivariable logistic regression models adjusting for age, gender, and race were constructed to evaluate the relationship between time, cancer history, and mortality. Results: We identified 9.6 million patients with AIS, of whom 491,704 (5.1%) had cancer. The in-hospital death rate among stroke patients with cancer decreased from 21.7% (95% confidence interval [CI], 20.6-22.9%) in 1996 to 13.9% (95% CI, 13.0-14.7%) in 2013, while the death rate among stroke patients without cancer decreased from 9.9% (95% CI, 9.6-10.1%) in 1996 to 6.4% (95% CI, 6.2-6.6%) in 2013 (Figure). After adjustment for demographics, a cancer diagnosis was associated with more than a doubling in odds of death after AIS (odds ratio [OR] 2.34; 95% CI, 2.29-2.38). However, during the 18-year study period, the demographic-adjusted odds of in-hospital death after AIS decreased similarly over time in patients with cancer (OR per year 0.97; 95% CI, 0.96-0.97) as compared to those without (OR per year 0.97; 95% CI, 0.97-0.98). Conclusions: Despite improved mortality rates over time, cancer patients with AIS continue to face more than a two-fold increased risk of in-hospital death as compared to non-cancer AIS patients.