Nutritional profiling of frail and obese, community dwelling older subjects: Results from a national survey

Introduction: Less attention has been given to the detection and nutritional status and needs of obese frail/sarcopenic older subjects. The aim of this study was to characterize the nutritional composition in older (≥65 years), frail-prone, obese subjects (defined by either waist circumference [WC] or body mass index [BMI]). Methods: A cross-sectional study with analysis of the national survey “Mabat Zahav”. Random sample of 1751 community dwelling Israeli older adults (≥65 years). Eleven nutritional factors formerly linked to frailty were a-priori selected based on the current literature. Data was extracted from a 24-hour dietary recall. Adherence for each nutritional factor was defined using the Dietary Reference Intakes (DRI), and aggregated into a sum score of the overall adherence (ranging from “0” to “11”, where “fair” adherence was defined as ≥6; inadequate adherence otherwise). Frailty likelihood was estimated using a validated non-direct model, and associations of nutritional factors with frailty-likelihood in obese vs non-obese individuals were examined. Additionally, a decision tree procedure based on machine learning was applied in order to capture nutritional factors related to frailty, stratified by gender, as well as by WC and/or BMI. Results: Overall, the prevalence rates of frailty and pre-frailty were 7.1 and 57.6%, respectively. A “fair nutritional adherence” was less common among frail-prone compared to robust subjects (23.1% vs. 32.1%; p < 0.0001). The intake of most frailty-related nutritional factors did not co-segregate according to the presence of abdominal or BMI-defined obesity. Still, compared to robust normal/overweight subjects, frail-prone obese (by BMI) individuals had a higher rate of inadequate nutritional adherence (odds-ratio 1.842; p < 0.05). Of all 11 nutritional factors, folate in obese women and vitamin A (as retinol) and calcium in non-obese and obese men, respectively, were recognized as the most prominent predictors of frail-prone prevalence by the machine learning process. Although BMI was more closely associated with impaired intake of the 11 selected nutritional components than WC, this association was eliminated when frailty status, low income and education were considered. Conclusions: Frail-prone subjects differed from robust subjects in their nutritional intake. Nutritional inadequacies related to frailty-likelihood were mostly seen among obese women and non-obese men. In the prediction of inadequate adherence to the DRI of 11 nutritional components, obesity is a weaker predictor than frailty, lower education and low income in older Israeli adults.