4CPS-179 Concordance analysis between the CONUT index and individual markers: evaluating nutritional status through unweighted and weighted kappa measures

<h3>Background and Importance</h3> Nutritional assessment using the CONUT index can be useful for detecting hospital malnutrition. Measuring the concordance between CONUT and its individual markers (albumin, cholesterol, lymphocytes) is essential to validate its accuracy, identify potential biases, and ensure the index provides a more comprehensive and balanced evaluation of nutritional status compared to isolated parameters. <h3>Aim and Objectives</h3> Assess the concordance between the CONUT index and individual markers, considering the impact of unweighted and weighted Kappa measures. <h3>Material and Methods</h3> We conducted a cross-sectional, observational study in a mid-level complexity hospital. We collected all lab results that used the CONUT index, along with individual parameters (albumin, total cholesterol, and lymphocyte count), excluding those with missing data. We used the following cut-off points to classify nutritional status: Albumin (g/dL): ≥ 3.5 (Normal), 3.0–3.49 (Mild), 2.5–2.99 (Moderate), &lt; 2.5 (Severe malnutrition); Total Cholesterol (mg/dL): ≥ 180 (Normal), 140–179 (Mild), 100–139 (Moderate), &lt; 100 (Severe malnutrition); Lymphocyte count (/mm³): ≥ 1600 (Normal), 1200–1599 (Mild), 800–1199 (Moderate), &lt; 800 (Severe malnutrition). Kappa, weighted Kappa and Altman’s criteria were applied to assess the concordance between the CONUT index and individual parameters. <h3>Results</h3> We analysed a total of 3,426 lab results from 2,882 patients. The agreement between the CONUT index and the individual markers was: CONUT vs. Albumin: Kappa values: 0.141 (unweighted – poor), 0.319 (linear – low), 0.502 (quadratic – moderate). CONUT vs. Cholesterol: Kappa values: 0.218 (unweighted – low), 0.401 (linear – low), 0.589 (quadratic – moderate). CONUT vs. Lymphocytes: Kappa values: 0.248 (unweighted – low), 0.418 (linear – low), 0.580 (quadratic – moderate). <h3>Conclusion and Relevance</h3> According to Altman’s criteria, the agreement between the CONUT index and individual markers (albumin, cholesterol, and lymphocytes) varies significantly depending on the type of weighting applied. This suggests that while exact category matches between CONUT and the markers may be rare, there is a stronger correlation when the proximity of categories is considered. This reflects that the overall trends in nutritional status classification are more similar when small differences in categorisation are accounted for. Further refinements in the weighting of individual markers within the CONUT system could improve its precision and reduce potential biases in specific clinical contexts. <h3>References and/or Acknowledgements</h3> <h3>Conflict of Interest</h3> No conflict of interest