Cutoff value and optimal age discrimination thresholds of impaired intrinsic capacity in Chinese older adults: Based on the CHARLS database

OBJECTIVE: To analyze the cutoff value and optimal age discrimination thresholds of impaired intrinsic capacity (IC) in Chinese older adults. METHODS: This study was based on the CHARLS 2015 wave cross-sectional data and included 6205 Chinese participants aged 60 years and above. The total IC score assessment system was constructed by synthesizing the performance in five major domains: psychological, sensory, cognitive, locomotion and vitality (0-10, higher scores indicating better capacity). A locally-weighted regression (LOESS) smoothing curve was used to fit the trend of IC changes in the sample population. With the help of receiver operating characteristic curve (ROC) analysis, the cutoff value of IC impairment and the optimal age discrimination thresholds were identified, and the predictive efficacy was comprehensively evaluated with the area under the curve (AUC), sensitivity, specificity, and Youden index. Meanwhile, to control for confounding factors, binary logistic regression was performed with adjustment for covariates, and model discrimination was assessed using the C-index to verify the predictive performance of the optimal age discrimination threshold for IC impairment. Split-sample validation was performed to avoid circularity. RESULTS: The cutoff value of impaired IC of the older adults in China was 7 points, and the prevalence of impaired IC was 41.35%. According to the results of ROC curve analysis, the AUC of age predicting impaired IC in the older adults was 0.654 (95% CI: 0.640-0.668), and the optimal age discrimination threshold of impaired IC in the older adults was 70.5 years old, and that for impaired IC in males was 70.5 years old, and that for impaired IC in females was 67.5 years old. The results of the logistic regression model showed that after controlling for the effects of confounders, the optimal age discrimination threshold had a predictive value for changes in IC impairment, with the risk of IC impairment in older adults>70.5 years being 3.093 times higher than that in older adults≤70.5 years (95% CI: 2.713-3.525), with the Hosmer-Lemeshow test of 0.551 and a C-index of 0.616 (95% CI: 0.602 to 0.631). Split-sample validation confirmed the optimal age discrimination threshold (OR = 2.82, Hosmer-Lemeshow test = 0.643, C-index = 0.788). CONCLUSIONS: Age demonstrates high predictive value for impaired IC in older adults, with an overall threshold at 70.5 years (70.5 years for men, 67.5 years for women). The impairment in IC exhibits distinct age thresholds that can guide targeted intervention strategies.