[Research on screening for frailty: development of "the Kaigo-Yobo Checklist"]

Early detection of older persons at high-risk with an effective screening tool is a prerequisite for strategies to optimize care among the community-dwelling elderly. We have focused on risks regarding the likelihood of developing homeboundedness, falls, and poor nutrition, and proposed an original 18 item-scale called "the Kaigo-Yobo Checklist" as a questionnaire-based screening tool. This study examined the reliability and validity of this scale using a longitudinal cohort of community-dwelling elderly.

Among 1039 older persons aged 70 years and over living in Kusatsu Town, Gunma Prefecture in 2001, 916 persons (88.2%) responded to the baseline interview survey including assessment with the 18-items of "the Kaigo-Yobo Checklist". Using these data, we performed the following analyses: (1) elimination of improper items according to pass and non-response rates; (2) internal reliability analysis based on Cronbach's alpha and Good-Poor approaches; (3) partial correlation analysis with IADL scores (Instrumental Self-Maintenance subscale of TMIG-Index of Competence) treated as external criteria. Further, we collected individual records under the long-term care insurance system in Kusatsu Town from April 2000 to November 2005, and determined the predictive value of the checklist for onset of certification of long-term care insurance over 4 years (2001-2005) with the trend test and logistic regression models.

(1) Three among 18 items did not show pass rates of 75% to 95% with a non-response rate under 1%. Thus we excluded these three items from the original checklist, yielding a 15 item-scale with 15 points for full marks, i.e., a modified version of "the Kaigo-Yobo Checklist". (2) The Cronbach's alpha was 0.79, and Good-Poor analysis determined that the high-score group (> or =2 points) had a higher mean score for all items of the checklist compared to the low-score group ( < or =1 point) (P < 0.001). (3) The partial correlation coefficient between the checklist score and the IADL score was -0.64 (P < 0.001). (4) The baseline checklist score was positively and linearly associated with the risk of developing a state in need of care during the 4-year follow-up; the adjusted odds ratio for the increment of 1 point was 1.21 (95% confidence interval, 1.10-1.33) [1.24 (1.11-1.38) when deceased or moving-out cases were excluded from the analysis].

The modified version of "the Kaigo-Yobo Checklist" had concurrent and predictive validity, and good reliability as a questionnaire-based scale for screening high-risk older persons.