Concurrent and predictive validity of nutritional screening tools in hospitalized pediatric patients: Protocol for a single-center, prospective, observational study

Abstract

Background and aim

Early identification of nutritional risk is essential in the care of hospitalized children. Although several screening tools exist, evidence on their performance in identifying at-risk patients and predicting nutrition-related outcomes remains limited, especially in non-Caucasian populations. This protocol outlines the rationale and methods to assess the concurrent and predictive validity of nutritional screening tools in hospitalized pediatric patients.

Methods

This prospective diagnostic accuracy study will enroll at least 204 patients aged 2–17 years, consecutively admitted to the emergency department of a tertiary pediatric hospital in Colombia and expected to remain hospitalized for ≥3 days. Within 48 hours of admission, nutritional risk will be assessed using STRONGkids, PYMS, and WHO criteria, alongside the Subjective Global Nutritional Assessment (SGNA) as the reference standard. The primary outcome is the concurrent validity of each screening tool to detect disease-related malnutrition, assessed via sensitivity, specificity, predictive values, likelihood ratios, and Kappa coefficient. Secondary outcomes include time to complete each tool and associations between baseline nutritional status, clinical outcomes, and sociodemographic variables. Predictive validity will be analyzed through ROC curves and area under the curve (AUC) values for each tool using a composite adverse outcome: ICU admission, hospital stay ≥7 days, in-hospital mortality within 30 days, and infectious complications.

Discussion

This study will generate prospective evidence on the diagnostic accuracy of widely used pediatric screening tools, using an appropriate reference method. Findings may help identify the most effective tool for early nutritional risk detection in underrepresented pediatric populations.