Background: Preoperative screening for malnutrition has become mandatory in The Netherlands. A sensitive method to diagnose malnutrition would save time and improve effectiveness.
Methods: A prospective cross-sectional study of 488 adult elective preoperative outpatients was performed. The accuracy of self-reported height and weight was compared with measured data and 3 commonly used malnutrition screening tools. Interobserver agreement was calculated by the intraclass correlation coefficient, studied in Bland and Altman plots, and analyzed by using Cohen's κ statistic. Accuracy was expressed in sensitivity, specificity, and false-negative rates.
Results: Differences between self-reported and measured data were significant, but clinically irrelevant, because only 1 patient was falsely identified as well nourished. Intraclass correlation coefficient for height, weight, and body mass index was high (.97-.99). Bland-Altman plots showed that the mean ± standard deviation differences and 95% limits of agreement between both methods were as follows: height, .0096 m (±.0262, -.0417 to +.0609 m); weight, -1.28 kg (±2.29, -5.76 to +3.20 kg); body mass index, -.72 kg/m(2) (±1.11, -2.92 to +1.46 kg/m(2)). The κ coefficient was .84 (95% confidence interval, .75-.94). Sensitivity was .97 and specificity was .98. Sensitivity and false-negative rates of self-reported data were better overall compared with the screening tools.
Conclusions: Self-reported data provide highly sensitive information to diagnose malnutrition in preoperative outpatients.
Copyright © 2012 Elsevier Inc. All rights reserved.