Reliability of a Frailty Index Based on the Clinical Assessment of Health Deficits in Male C57BL/6J Mice

We investigated the reliability of a newly developed clinical frailty index (FI) that measures frailty based on deficit accumulation in aging mice. FI scores were measured by two different raters independently in a large cohort (n = 233) of 343-430 day-old male C57BL/6J mice. Inter-rater reliability was evaluated with correlation coefficients, the kappa statistic, and intra-class correlation coefficients (ICC) in three separate groups of mice (n = 45, 50, and 138 mice/group) sequentially over 3 months. After each group was evaluated, descriptions of techniques used to identify health deficits were amended. Mice had comparable overall FI scores regardless of rater (0.213 ± 0.002 vs 0.212 ± 0.002; p = .802), although discordant measures declined as techniques were refined. Correlation coefficients (r2 values) between raters improved throughout the study and mean kappa values increased (mean ± SEM; 0.621 ± 0.018, 0.764 ± 0.017, and 0.836 ± 0.009 for groups 1, 2, and 3; p < .05). Values for intra-class correlation coefficient also improved from .51 (95% confidence interval = 0.11-0.73) to .74 (0.54-0.85) and .77 (0.67-0.83). FI scores increased over 3 months (p < .05), but did not differ between raters. These results show a high overall inter-rater reliability when the clinical FI tool is used to assess frailty in a large cohort of mice.