STRONGkids: Predictive Validation in Brazilian Children.

OBJECTIVES: The aim of the study is to assess STRONGkids as a tool for predicting weight loss and length of hospital stay in children and to determine whether the anthropometric diagnosis of nutritional status at the time of admission was associated with weight loss and length of hospital stay.

METHODS: A methodological study recruiting 245 children age between 1 and 10 years of age admitted to a tertiary hospital. The participants were weighed daily until discharge. Validation of the STRONGkids tool for the identification of patients sustaining weight loss at the end of hospitalization involved the calculation of sensitivity, specificity, and positive and negative predictive values, and anthropometric assessment.

RESULTS: A total of 129 (52.7%) children lost weight at the end of hospitalization. Of these, 73 (56.6%) lost over 2% of their weight on admission. The tool had a sensitivity of 55.8%, a specificity of 38.8% and a positive predictive value of 50.3% in identifying children who lost weight. The anthropometric assessment had a sensitivity of 26.5%, a specificity of 75.9%, and a positive predictive value of 49.1%.

CONCLUSIONS: The model used to develop the STRONGkids tool incorporated clinical evaluation to a greater extent than the assessment of nutritional status. The tool, however, had a low sensitivity and a high percentage of false positives. Therefore, it should be considered as a preliminary evaluation tool and its use should be complemented with clinical data.