Validation of a novel nutrition risk screening tool in stroke patients.

Background and objectives: We aimed to apply a novel nutrition screening tool to stroke patients and assess its reliability and validity.

Methods and study design: Cross-sectional data among 214 imaging-confirmed stroke patients were collected between 2015 and 2017 in two public hospitals in Hebei, China. Delphi consultation was conducted to evaluate the items in the NRS-S scale. Anthropometric indices including body mass index (BMI), triceps skin fold thickness (TSF), upper arm circumference (AMC) and mid-arm muscle circumference (MAMC) were measured. Internal consistency reliability, test-retest reliability, construct validity and content validity were assessed. In order to estimate content validity, two rounds Delphi consultation of fifteen experts were conducted to evaluate the items in the Nutrition Risk Screening Scale for Stroke (NRS-S).

Results: High internal consistency was indicated by Cronbach's alpha of 0.632 and a split-half reliability of 0.629; test-retest reliability of NRS-S items ranged from 0.728 to 1.000 (p<0.0001), except for loss of appetite (0.436, p<0.001) and gastrointestinal symptoms (0.213, p=0.042). Content validity index of 0.89 indicated robust validity of the items. Regarding construct validity, the Kaiser-Meyer-Olkin value was 0.579, and the result of the Bartlett test of sphericity was 166.790 (p<0.001). Three factors were extracted by exploratory factor analysis, which contributed to 63.079% of the variance. Confirmatory factor analysis was performed on the questionnaire, finding the p-value of the model to be 0.321, indicating a high model fitting index.

Conclusions: A novel stroke-specific nutritional risk screening tool demonstrated a relatively high reliability and validity in its clinical application.