Validity and reliability of the new Canadian Nutrition Screening Tool in the 'real-world' hospital setting

BACKGROUND/OBJECTIVES: Nutrition screening should be initiated on hospital admission by non-dietitians. This research aimed to validate and assess the reliability of the Canadian Nutrition Screening Tool (CNST) in the 'real-world' hospital setting. SUBJECTS/METHODS: Adult patients were admitted to surgical and medical wards only (no palliative patients). Study 1-Nutrition Care in Canadian Hospitals (n = 1014): development of the CNST (3 items: weight loss, decrease food intake, body mass index (BMI)) and exploratory assessment of its criterion and predictive validity. Study 2-Inter-rater reliability and criterion validity assessment of the tool completed by untrained nursing personnel or diet technician (DT) (n = 150). Subjective Global Assessment performed by site coordinators was used as a gold standard for comparison. RESULTS: Study 1: The CNST completed by site coordinators showed good sensitivity (91.7%) and specificity (74.8%). Study 2: In the subsample of untrained personnel (160 nurses; one DT), tool's reliability was excellent (Kappa = 0.88), sensitivity was good (>90%) but specificity was low (47.8%). However, using a two-item ('yes' on both weight change and food intake) version of the tool improved the specificity (85.9%). BMI was thus removed to promote feasibility. The final two-item tool (study 1 sample) has a good predictive validity: length of stay (P<0.001), 30-day readmission (P = 0.02; X-2 5.92) and mortality (P<0.001). CONCLUSIONS: The simple and reliable CNST shows good sensitivity and specificity and significantly predicts adverse outcomes. Completion by several untrained nursing personnel confirms its utility in the nursing admission assessment.