Comparison of early warning scoring systems for predicting stroke occurrence among hospitalized patients: A study using smart clinical data warehouse.

BACKGROUND

This study aimed to evaluate the predictive ability of two widely used early warning scoring systems, the Modified Early Warning Score (MEWS) and the National Early Warning Score (NEWS), for predicting stroke occurrence in hospitalized patients.

METHODS

The study enrolled 5,474 patients admitted to the intensive care unit from the general ward using data from the Smart Clinical Data Warehouse (CDW). MEWS and NEWS were calculated based on vital signs and clinical parameters within four hours of stroke onset. Stroke occurrence was categorized as ischemic or hemorrhagic. Logistic regression and receiver operating characteristic curve analyses were performed to assess the predictive abilities of the scoring systems.

RESULTS

Of the enrolled patients, 33.9% (n = 1853) experienced stroke, comprising 783 cases of ischemic stroke and 1,070 cases of hemorrhagic stroke. Both the MEWS and the NEWS were found to significantly predict overall stroke occurrence with a cutoff value of 4 (MEWS>4; OR [95% CI]: 13.90 [11.51-16.79], p<0.001; NEWS>4; OR [95% CI]: 6.71 [5.75-7.83], p<0.001). Parameters, such as prior malignancy, atrial fibrillation, AVPU response, heart rate, respiratory rate, and oxygen saturation, are also associated with stroke occurrence. The predictive ability of MEWS and NEWS was good for overall stroke occurrence. (AUC of MEWS: 0.92, 95% CI [0.91-0.93], p<0.001; AUC of NEWS: 0.85, 95% CI [0.84-0.86], p<0.001). The predictive ability was considered fair for ischemic stroke but good for hemorrhagic stroke.

CONCLUSION

MEWS and NEWS demonstrated significant predictive abilities for overall stroke occurrence among hospitalized patients, with MEWS slightly outperforming NEWS.