The Elderly Nutritional Index for Geriatric Malnutrition Assessment (ENIGMA): concurrent, construct and predictive validity in an external evaluation cohort of community-dwelling older persons

Abstract

Abstract We previously developed a malnutrition risk index, the Elderly Nutritional Index for Geriatric Malnutrition Assessment (ENIGMA) with good predictive accuracy for mortality risk in an original population cohort (SLAS1). Herein, we further evaluate the concurrent and predictive validity of the ENIGMA construct in an external validation cohort (SLAS-2) of 2824 community-dwelling older adults aged 55+ years. They were assessed on the ENIGMA index, Mini Nutritional Assessment-Short Form (MNA-SF) and the Geriatric Nutritional Risk Index (GNRI), known correlates of malnutrition, and baseline and follow-up functional dependency and 10-year mortality risk. Higher ENIGMA risk categories were significantly associated (P < 0·001) with lower education, living alone, smoking, low physical activity, BMI < 18·5 kg/m2, poorer muscle strength and functional mobility, exhaustion, physical frailty, homocysteine, glomerular filtration rate, Hb, red and white blood cell counts, platelets, systemic inflammation indexes, metabolic syndrome, CVD, cognitive impairment and depressive symptoms (Geriatric Depression Scale ≥ 5). ENIGMA scores showed statistically significant (P < 0·001) correlations but low-to-moderate concordance with MNA-SF (r = 0·148, agreement = 45·9 %, kappa = 0·085) and GNRI scores (r = 0·156, agreement = 45·8 %, kappa = 0·096). Controlling for known correlates of malnutrition, only high-risk ENIGMA among the indexes significantly predicted baseline functional dependency (OR = 1·64, 95 % CI 1·01, 2·65) and mortality (hazard ratio = 1·65 (95 % CI 1·04, 2·62). ENIGMA marginally out-performed MNA-SF and GNRI in predicting baseline functional dependency (AUC: 0·625 v. 0·584 v. 0·526), follow-up functional dependency (AUC: 0·594 v. 0·525 v. 0·479) and 10-year mortality risk (AUC: 0·641 v. 0·596 v. 0·595). The concurrent and predictive validity of the ENIGMA construct is replicated in an external evaluation study of community-dwelling older persons.