Background: Nutrition risk screening for community-living seniors is of great interest in the health arena. However, to be useful, nutrition risk indices need to be valid and reliable. The following three studies describe construct validation, test-retest and inter-rater reliability of SCREEN II.
Methods: Study (1) seniors were recruited from the general community and from a geriatrician's clinic to complete a nutritional assessment and SCREEN II. 193 older adults provided medical and nutritional history, 3 days of dietary recall and anthropometric measurements. A dietitian reviewed all information collected and ranked seniors on risk: 1 (low) to 10 (high risk). Receiver operating characteristic curves were completed. An abbreviated SCREEN II was developed through statistical analysis and expert ranking of the 17 items. Studies (2) and (3) seniors were recruited from the community to self-administer (n = 149) or be interviewed (n = 97) using SCREEN II twice within 2 weeks. For self-administration one index was completed via mail. Interviewer administration was completed via telephone with two interviewers. Intra-class correlations were calculated.
Results: (1) Total and abbreviated SCREEN II have increased sensitivity and specificity as compared to SCREEN I in identifying seniors at nutritional risk. (2) Test-retest reliability was adequate (intra-class correlation (ICC) = 0.83). (3) Inter-rater reliability was adequate (ICC = 0.83).
Conclusions: SCREEN II appears to be a valid and reliable tool for the identification of risk for impaired nutritional states in community-living older adults, and is an improvement over SCREEN I.