Abstract
Objective This study aims to assess the predictive accuracy and clinical utility of the Collaborative Respiratory Assessment Score (CoRAS) in identifying pneumonia risk among stroke patients in Kaifukuki Rehabilitation Wards (KRWs). CoRAS, developed by a multidisciplinary team, incorporates eight clinical parameters to quantify respiratory risk. The study retrospectively applies CoRAS to a cohort of stroke patients and evaluates its predictive performance using statistical methods, including logistic regression and receiver operating characteristic (ROC) curve analysis. Additionally, the study compares CoRAS with traditional risk factors such as age and pre-stroke care level to determine its unique contribution to pneumonia prediction. The findings aim to validate CoRAS as an effective tool for early risk stratification, supporting multidisciplinary collaboration and targeted interventions in KRWs. Methods We devised CoRAS as a scoring system based on eight clinical parameters: consciousness level, SpO₂, need for suctioning, history of respiratory diseases, FILS (Food Intake Level Scale), Hoffer criteria, nutritional status, and OHAT-J (Oral Health Assessment Tool - Japanese version). These parameters were weighted proportionally to sum up to a total score of 100. Assessments were independently conducted by seven multidisciplinary professionals, and the score was retrospectively applied to data from 629 stroke patients admitted to our hospital. Data on pneumonia occurrence were collected and analyzed. Results Pneumonia was observed in 48 (7.6%) of the 629 patients. The highest variance inflation factor (VIF) among the eight parameters was 2.11, validating the application of a linear combination model. ROC analysis showed an area under the curve of 0.860. Logistic regression revealed that CoRAS had an adjusted odds ratio of 1.06 per point for predicting pneumonia. Conclusion CoRAS demonstrated satisfactory predictive validity for pneumonia onset among stroke patients in the KRW, suggesting its utility as an assessment tool for multidisciplinary teams.