Development, validity, and utility of a model-based intrinsic capacity composite score in community-dwelling older persons

BACKGROUND To address the lack of a composite intrinsic capacity (IC) score based on the World Health Organization (WHO) Integrated Care for Older People (ICOPE) Screening tool, we propose a model-based approach to computing the composite score by developing and validating a model that uses the ICOPE screening items to predict the presence of impaired IC in community-dwelling older adults. METHODS In this cross-sectional study, a sample of 1,235 participants (mean[SD], 68[7]years) completed a multi-domain geriatric and fitness assessment, from which we (i) operationalized the ICOPE Steps 1-2 screening/assessment and (ii) derived a 5-point count-based Step 1 composite IC score and an 8-point Step 2 composite IC score. Proportional-odds regression analysis, leveraging on ICOPE screening items, was used to predict, for a given person, the probability that the Step 2 score was ≤6points (impaired IC) and the corresponding mean score (model-based composite IC score). RESULTS The model c-statistic for impaired IC was 0.81 (95%CI, 0.79 to 0.86) and calibration was excellent. The model-based IC scores showed stronger discriminative validity than did the count-based scores for the prefrailty/frailty, sarcopenia, and restricted life-space-mobility outcomes (AUC differences, 0.05-0.12; P-values < 0.001). At a decision threshold of 20%, the model-based approach yielded greater net benefit (0.016), and it allowed a greater proportion of participants to potentially delay Step 2 assessment (n = 633 vs 317) without compromising positive and negative predictive values. CONCLUSIONS If externally validated, composite IC scores derived from the proposed model-based approach have the potential to facilitate more granular risk stratification and IC monitoring.