Toward Standardization of Intrinsic Capacity Assessment: A Methodological Comparison of the Responsiveness of Composite Scoring Methods

Intrinsic capacity (IC) is increasingly used as an intervention outcome; however, IC composites lack a standardized scoring method, hindering comparisons across trials. To determine whether the scoring method influenced the detection of intervention effects, we applied multiple composite scoring approaches to the same data and compared their responsiveness to a 12-week exercise intervention in older adults at risk of mild cognitive impairment. METHODS: Pre- and post-intervention IC data were obtained from 59 community-dwelling older adults assigned to resistance training (n=34) or stretching and flexibility (n=25). The five-domain IC composite was computed repeatedly using alternative scoring approaches: z-score (concurrent and baseline-anchored variants), percentage of maximum possible (POMP), binary classification (0/1), and a three-category ordinal scale. Responsiveness was quantified using the standardized response mean (SRM). RESULTS: Responsiveness ranged from negligible to large for identical data, indicating that the intervention effect depended on the scoring method. The ordinal and POMP composites were the most responsive (SRM 0.74-0.88 and 0.50-0.76). Within the z-score composite, responsiveness depended entirely on the standardization anchor: the concurrent specification was insensitive (SRM ≈ 0), whereas baseline-anchoring showed moderate responsiveness (0.62-0.71). The binary scoring was the least sensitive, showing a ceiling effect. CONCLUSIONS: Ordinal three-category scoring was the most responsive, suggesting that approaches combining a fixed reference with a category-based structure are well suited for detecting change in IC. This finding requires confirmation against external criteria for IC, such as its predictive validity for adverse health outcomes, along with domain weighting and instrument selection.