Temporal Validation of an Assessment Tool that Predicts a Possibility of Home Discharge for Patients with Acute Stroke

Abstract

Objectives: Several prediction models have been developed to assess discharge destinations for patients with acute stroke; however, few studies have performed external validation. We aimed to perform a temporal external validation of a prediction tool to identify stroke patients with a high possibility of discharge to home.

Materials and methods: From December 2017 to July 2019, consecutive patients with acute stroke were included. Clinical nurses and physical therapists applied the prediction model to assess the patients' possibility of home discharge. Whether or not the patient was discharged their own home was the outcome measured. We calculated the sensitivity and specificity of the model and evaluated the discrimination and calibration based on the area under the curve (AUC) and the calibration plot.

Results: Of the 1214 patients assessed, 618 (51%) were discharged home. Using the same cutoff values recommended in the study that first described the tool, we determined the sensitivity and specificity of 91% and 59%, respectively. The AUC to assess the model discrimination was 0.80 (95% confidence interval, 0.77-0.82) and the calibration plot showed acceptable agreement between the predicted and observed outcomes.

Conclusions: The tool showed a high sensitivity, as expected, in the present study, which examined external validity during the different study periods.

Keywords: Clinical prediction rule; Patient discharge; Stroke; Stroke rehabilitation; Validation studies.