Misclassification of Nutrition Status among Elderly Based on Combination of Stature Predictor

The combination of arm span and knee height as the stature predictors on elderly was arranged to obtained the most precise and accurate prediction of actual height. However, there was not much research that calculated prediction errors. This study aimed to measure the size of the misclassification of elderly nutritional status which calculated by predictive combination regression model. The study was an observational study with cross-sectional design. The study population was 60–69 years old men and women in Wonogiri District, Central Java. The sample size were 65 men and 71 women, which were chosen purposively in the community. Anthropometric measurements on knee height and arm span and the actual height were done by standardize technique. Data of BMI were analysed using Friedman pos hoc Wilcoxon test then clasified to nutrition status to calculated sensitivity and specificity. Among the men subjects, the overweight and underweight status using BMI were overestimate by 8.3% and 12.5%, respectively. Among the women subjects, overestimate also occurred on underweight by 12.5%. In all respondents, underestimation of underweight was 7.2% and overestimation in the normal nutritional status was 3.7%. It was concluded that some misclassification of BMI, by predicted height among elderly based with combination regression model was found.