Risk Factors and Predictive Model of Diarrhea Among Patients with Severe Stroke

OBJECTIVE: To investigate the risk factors and predictive model of diarrhea among patients with severe stroke. METHODS: The study analyzed the retrospective clinical data of patients with new-onset stroke who had been admitted to the intensive care unit at the Department of Neurology of X Hospital, between September 2017 and April 2018. All data were analyzed with a binary logistic regression, and a logistic regression equation was used to build a predictive model of diarrhea among patients with severe stroke. RESULTS: A total of 153 patients with severe stroke were included in this study, including 45 patients (29.41%) with diarrhea. The binary logistic multivariate analysis showed that the National Institutes of Health Stroke Scale score at admission (odds ratio [OR], 1.123; 95% confidence interval [CI], 1.016-1.242), the Glasgow Coma Scale score at admission (OR, 1.563; 95% CI, 1.048-2.330), antibiotic use (OR, 2.168; 95% CI, 1.041-4.514), gavage feeding time (OR, 1.260; 95% CI, 1.098- 1.445), and hospital stay before the occurrence of diarrhea (OR, 0.652; 95% CI, 0.552-0.770). The receiver operating characteristic curve was 0.862 (95% CI, 0.799-0.925), the specificity was 0.778, and the sensitivity was 0.843. CONCLUSIONS: The National Institutes of Health Stroke Scale score at admission, the Glasgow Coma Scale score at admission, antibiotic use, gavage feeding time, and hospital stay before the occurrence of diarrhea independently predict diarrhea among patients with severe stroke. This model can be used to predict the risk of diarrhea among patients with severe stroke.