Further assessment of the reliability and validity of a Nutritional Risk Index: analysis of a three-wave panel study of elderly adults.

This paper reports on the further assessment of the reliability and validity of a short (16-item), portable method for assessing nutritional risk which is easily administered in the typical social survey setting. Data were obtained from a three-wave panel study of 401 randomly selected, noninstitutionalized elderly persons (age 65 and over) in St. Louis. Reliability was assessed by both internal consistency and test-retest methods. Reliability coefficients (internal consistency) of .603, .544, and .515 were obtained at T-1, T-2, and T-3, respectively. Cross-panel intercorrelations (test-retest) ranged between .67 and .71. Validity was assessed using factor analysis and various outcome measure comparisons for those at risk versus those not at risk. A five-factor orthogonally rotated solution explained 47.9 percent of the variance in the 16 items. Individuals with higher risk scores had significantly poorer health as measured by other standard indexes, and used significantly more health services than those with lower risk scores. These results underscore the potential of the Nutritional Risk Index (NRI) as a screening device for use among the elderly.