Selection of potential anthropometric and undernutrition-related items was based on consensus literature. The criteria were developed using 15-year mortality in community-dwelling older persons聽鈮ヂ?5 years (Longitudinal Aging Study Amsterdam, n聽=聽1687) and validated in an independent sample (InCHIANTI, n聽=聽1142).
Groups distinguished were: (1) undernutrition (mid-upper arm circumference <25聽cm or involuntary weight loss聽鈮?聽kg/6 months); (2) risk of undernutrition (poor appetite and difficulties climbing staircase); and (3) no undernutrition (others). Respective hazard ratio鈥檚 for 15-year mortality were: (1) 2.22 (95%CI 1.83-2.69); and (2) 1.57 (1.22-2.01) ((3)聽=聽reference). The area under the curve (AUC) was 0.55. Comparable results were found stratified by sex, excluding cancer/obstructive lung disease/(past) smoking, using 6-year mortality, and applying results to the InCHIANTI study (hazard ratio鈥檚 2.12 and 2.46, AUC 0.59).
The developed set of criteria (SNAQ65+) for determining (risk of) undernutrition in community-dwelling older persons shows good face validity and moderate predictive validity based on the consistent association with mortality in a second independent study sample.