OBJECTIVE
The "Kihon Checklist" (a frailty checklist), consisting of 25 items, is used for screening frail elderly, based on the Japanese long-term care insurance system. However, few reports have investigated predictive ability of incident long-term care insurance certification in the Kihon Checklist. The purpose of this study was to investigate inter-relationships and accuracy as a screening test of individual items and criteria in the Kihon Checklist for incident long-term care insurance certification.
METHODS
In December 2006, we distributed a questionnaire including the Kihon Checklist to individuals older than 65 years living in Ohsaki City, Japan. Among the valid respondents, we followed those who gave informed consent to follow-up, had more than 1 item of response on the Kihon Checklist, and were not qualified for long-term care insurance certification at the baseline. We further excluded individuals who died or moved away in the one year follow-up, analyzing 14,636 elderly. The age- and sex-adjusted odds ratio (OR) and 95% confidence interval (95%CI) for newly incident long-term care insurance certification were estimated by logistic regression analysis. Independent variables were each of the items and criteria in the Kihon Checklist used for screening of "frail elderly". In addition, we estimated the sensitivity and specificity, and conducted receiver operating characteristic (ROC) analysis for each criteria domain.
RESULTS
5,560 (38.0%) matched the criteria of "frail elderly". During the one year of follow-up, 483 (3.3%) required newly incident long-term care insurance certification. All of the items in the Kihon Checklist were significantly associated with incident long-term care insurance certification (range of ORs: 1.45-4.67). In addition, all of the criteria also significantly predicted the risk of incident long-term care insurance certification (range of OR: 1.93-6.54). The OR (95%CI) for "frail elderly" was 3.80 (3.02-4.78). Among the various domains, "20 items other than five related to prevention and support for depression" had the largest area under the ROC curve.
CONCLUSION
All items and criteria used for screening frail elderly in the Kihon Checklist are useful for predicting the risk of incident long-term care insurance certification during a one-year period. However, the strength of the relation and accuracy for screening test were variable among items or domains, and criteria values could be improved.