External validation of the A2SD2 and ISAN scales for predicting infectious respiratory complications of ischaemic stroke

Abstract

Background and purpose: Pneumonia as a complication of stroke is associated with poor outcomes. The A2DS2 and ISAN scales were developed by German and English researchers, respectively, to predict in-hospital stroke-associated pneumonia. We conducted an external validation study of these scales in a series of consecutive patients admitted to our hospital due to ischaemic stroke.

Method: These predictive models were applied to a sample of 340 consecutive patients admitted to hospital in 2015 due to stroke. Discrimination was assessed by calculating the area under the ROC curve for diagnostic efficacy. Calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test and graphing the corresponding curve. Logistic regression analysis was performed to determine the independent predictors of respiratory infection secondary to stroke.

Results: We included 285 patients, of whom 45 (15.8%) had respiratory infection after stroke according to the study criteria. Mean age was 71.01±12.62 years; men accounted for 177 of the patients (62.1%). Seventy-two patients (25.3%) had signs or symptoms of dysphagia, 42 (14.7%) had atrial fibrillation, and 14 (4.9%) were functionally dependent before stroke; the median NIHSS score was 4 points. Mean scores on A2DS2 and ISAN were 3.25±2.54 and 6.49±3.64, respectively. Our analysis showed that higher A2DS2 scores were associated with an increased risk of infection (OR=1.576; 95% CI: 1.363-1.821); the same was true for ISAN scores (OR=1.350; 95% CI: 1.214-1.501).

Conclusion: High scores on A2DS2 and ISAN were found to be a strong predictor of respiratory infection associated with acute stroke in a cohort of consecutive patients with stroke. These easy-to-use scales are promising tools for predicting this complication in routine clinical practice.

Keywords: Ictus isquémico; Infección respiratoria; Ischaemic stroke; Modelos predictivos; Neumonía; Outcome; Pneumonia; Predictive models; Pronóstico; Respiratory infection.