Abstract
BACKGROUND: Appetite loss is a major concern in various clinical contexts but the self-assessment of appetite might be compromised in some populations such as in individuals with dementia. The use of surrogate responders might be useful but no study so far has investigated whether hetero-administrated questionnaires might be valid tools to assess appetite. METHODS: Forty men and women aged 66-97 years, without dementia, were recruited from the Medical care and Rehabilitation Unit of the Toulouse University Hospital, France. Within the same day, patients' appetite was assessed using the SNAQ, a 4-answers questionnaire that was administrated separately to the patients as well as to two certified nursing assistant (CNA) surrogate responders answering on behalf of the patients. Intraclass correlation coefficients (ICC) as well as Kappa's and Gwet's AC1 coefficients were computed to test the agreement between the patients and each surrogate assessor and between the two surrogate assessors. RESULTS: When comparing patients versus surrogate responders' SNAQ scores, ICC (95% Confidence Interval (CI)) showed weak to moderate correlations, ranging from 0.087 (-0.224, 0.384) to 0.519 (0.254, 0.713). The rate of correct categorization was 65.0-67.5 % and Kappa's and Gwet's coefficients ranged from 0.300 (0.006, 0.594) to 0.360 (0.082, 0.638), indicating poor agreements between patient and surrogates. When comparing the surrogate responders with each other, the percentage of similar diagnosis was 82.5% with corresponding Kappa's and Gwet's coefficients of 0.650 (0.417, 0.883) and 0.652 (0.406, 0.898), indicating good agreements among surrogate raters. CONCLUSION: We found weak-to-moderate coefficients for the SNAQ used as a hetero-administrated questionnaire. Further investigation across diverse clinical settings and using larger samples are needed to provide further insights to our preliminary observations.