PP415 - A NEW SCREENING TOOL TO PREDICT RISK OF MALNUTRITION IN PATIENTS WITH OVERWEIGHT AND OBESITY

PP415

A NEW SCREENING TOOL TO PREDICT RISK OF MALNUTRITION IN PATIENTS WITH OVERWEIGHT AND OBESITY

J. Borkent¹, B. VAN DER MEIJ^1,*, M. de van der Schueren¹

¹HAN University of applied sciences, Nijmegen, Netherlands

Rationale: Current malnutrition screening tools (MSTs) are not sensitive to the GLIM criteria and do not specifically address risk of malnutrition in patients with overweight or obesity. We aimed to develop a screening tool to fill this gap.

Methods: Included were 766 hospital patients with overweight or obesity (BMI ≥25 kg/m²)  (SCOOP project). More than 50 malnutrition risk factors and anthropometric measures were included as possible determinants. The reference method,  GLIM, was based on low muscle mass (measured by Bioelectrical Impedance Analysis) or unintentional weight loss (>5% in one month, >10% indefinite of time), combined with reduced intake or inflammation.

Data were split into a training set (70%) and a test set (30%). Separate LASSO logistic regression models were developed for the phenotypic outcomes ‘low muscle mass’ and ‘unintentional weight loss’, with bootstrap resampling used to select top predictors. Predicted probabilities were combined in a meta-model. Model performance against GLIM was evaluated using ROC/AUC, and a web-based calculator was created for individual-level prediction.

Results: The prediction model includes low appetite, problems with doing physical activities, low food intake, >3 doctor visits in the past year, no daily dairy/soy consumption, calf circumference (in cm), nausea, swallowing difficulties, and unintentional weight loss. It shows good performance with an AUC of 0.80, sensitivity 0.79, specificity 0.70, positive predictive value 0.50, negative predictive value 0.90. The accuracy of the model is 0.72.

Image:

Conclusion: This screening tool is the first to accurately identify risk of malnutrition in patients with overweight or obesity, addressing gaps in existing MSTs. The web-based calculator offers a practical, easy-to-use tool for early detection and timely interventions. The predictive validity of this tool needs to be established in future studies.

Disclosure of Interest: None declared