[Validity of the "Kaigo-Yobo Check-List" as a frailty index]

A frailty index for Japanese older people is not yet available. This study examined the validity of "Kaigo-Yobo Check-List" (CL) as a frailty index.

The study site was Kusatsu town, Gunma prefecture. Out of 612 older persons aged 65 years and over who undertook a comprehensive geriatric assessment in 2007, results from 526 who had no missing data were used to examine the cross-sectional relationship between frailty as defined by Fried's criteria (= external criteria) and CL scores in order to evaluate concurrent and construct validity. Further, 916 older individuals aged 70 years and over who responded to the baseline survey in 2001 were followed for the subsequent 4 years and 4 months regarding the onset of ADL disability, service use under the Long Term Care Insurance program, and mortality. We examined the predictive validity of the CL for such adverse outcomes after adjustment for gender, age, and comorbidity.

The CL (cut-off point = 3/4) discriminated frailty from non-frailty at the sensitivity of 70.0% and specificity of 89.3%. The higher the CL score, the higher the prevalence of frailty; the trend was highly significant (P < 0.001). The Multitrail Multimethod Model showed that there were significant associations among three components of CL (homeboundness, falling, and lower nutrition) and four out of five components of Fried's frailty criteria (shrinking, exhaustion, low activity, and slowness), whereas those components of the CL did not have an association with the weakness component of Fried's frailty criteria. As compared with older persons who had CL scores of 3 points or below, those who had CL scores of 4 or more points had a significantly higher risk for developing adverse outcomes. Multivariate-adjusted odds ratios for ADL disability at 2 and 4 years after baseline were 5.25 (95% confidence interval, 2.79-9.89) and 3.42 (1.79-6.54), respectively. Likewise, multivariate-adjusted hazard ratios for the onset of service use under the Long Term Care Insurance program and mortality during the follow-up period of 4 years and 4 months were 3.50 (2.41-5.07) and 2.43 (1.70-3.47), respectively. Although the construct validity remained inconclusive, the "Kaigo-Yobo Check-List" showed

good concurrent and predictive validity as a frailty index. Since it comprises 15 easy-to-answer questions, it could be widely used for research on frailty and its preventive intervention.