Accuracy of non-paralytic anthropometric data for nutritional screening in older patients with stroke and hemiplegia.

Abstract

Although malnutrition commonly affects stroke patients, there are no validated screening tools. We verified the usefulness of non-paralytic anthropometric measurements for the nutritional screening of stroke. A cross-sectional study was conducted in consecutive stroke patients with hemiplegia aged ⩾65 years, with Mini Nutritional Assessment Short Form score ⩽11. Diagnostic malnutrition criteria from the European Society for Clinical Nutrition and Metabolism were the reference standards: body mass index (BMI) <18.5 kg/m2 or weight loss >10%+BMI <22 or 20 kg/m2. Non-paralytic anthropometric measurements (calf circumference (CC), arm circumference (AC), triceps skinfold (TSF) and arm muscle circumference (AMC)) and serum albumin concentration (Alb) at admission were the index tests. Cutoffs were determined by receiver operation curve and Youden index, and accuracy by area under the curve (AUC) and kappa value. Functional independence measures at discharge and discharge destination were collected. We included 488 patients (224 men and 264 women) with a mean age of 78.8 years and mean BMI of 22.0 and 21.1 kg/m2, respectively. Eighty-one men and 124 women had malnutrition. The AUC for CC, AC, TSF, AMC and Alb was 0.859, 0.825, 0.764, 0.745 and 0.670 for men, and 0.881, 0.843, 0.796, 0.742 and 0.658 for women, respectively. In both sexes, CC had the highest kappa (0.533, 0.608; both P<0.001) with cutoff values ⩽31 and ⩽30 cm. Patients with lower CC showed significantly worse functional outcomes and lower proportion of return to home (P<0.001). Non-paralytic CC indicated malnutrition with sufficient accuracy and good correlation with functional capacity; it may be a useful nutritional screening tool for stroke.