作者
Gotaro Kojima,Yu Taniguchi,Akihiko Kitamura,Shoji Shinkai
摘要
To explore comparability of Kihon Checklist (KCL) and Kaigo-Yobo Checklist (KYCL) to Frailty Index (FI) in predicting risks of long-term care insurance (LTCI) certification and/or mortality over 3 years.Prospective cohort study.1023 Japanese community-dwelling older adults from the Kusatsu Longitudinal Study of Aging and Health.Frailty status was quantified at baseline using KCL, KYCL, and 32-deficit and 68-deficit FI. Relationships of the measures were examined using Spearman rank correlation coefficients. Cox regression models examined the risk of new certification of LTCI or mortality according to KCL, KYCL, and FI. Predictive abilities of KCL and KYCL were compared with FI using area under the receiver operating characteristic curve (AUC), C statistics, net reclassification improvement (NRI), and integrated discrimination improvement (IDI).Mean age was 74.7 years and 57.6% were women. KCL and KYCL were significantly correlated to 32-FI (r = 0.60 and 0.36, respectively) and to 68-FI (r = 0.88 and 0.61, respectively). During the follow-up period, 92 participants (9%) were newly certified for LTCI or died. Fully adjusted Cox models showed that higher KCL, KYCL, 32-FI, and 68-FI were all significantly associated with elevated risks [hazard ratio (HR) = 1.03, 95% CI = 1.01-1.04, P < .001; HR = 1.04, 95% CI = 1.02-1.05, P < .001; HR = 1.03, 95% CI = 1.01-1.05, P = .001; HR = 1.04, 95% CI = 1.02-1.06, P < .001, respectively, per 1/100 increase of max score]. AUC and C-statistics of KCL and KYCL were not different statistically from those of 32-FI and 68-FI. Predictive abilities of KCL were superior to 32-FI in NRI and IDI but inferior to 68-FI in category-free NRI, and those of KYCL were superior to 32-FI in IDI but inferior to 68-FI in NRI.Although KCL and KYCL include smaller numbers of items than standard FI, both tools were shown to be highly correlated with FI, significant predictors of LTCI certification and/or mortality, and compatible to FI in the risk prediction.