Cognitive and Epigenetic Predictors of Healthcare Utilization in the Canadian Longitudinal Study on Aging: Factor- and Indicator-level Examinations

Abstract

Background: Healthcare utilization among older adults is a valuable indicator of health status and provides a mechanism linking illness to healthcare costs. Cognitive function and epigenetic age—as indicators of nervous system and cellular integrity at a biological level—are both correlated with age and may be important predictors of healthcare utilization. It is unclear how strongly the two predictor categories are associated, however, and it is not known whether they are best understood as correlated processes under a general “systemic resilience” (SR) construct, or as separate factors with independent influence on healthcare utilization. In the present research study, I examine indicator-level and factor-level predictors of healthcare utilization, with a focus on cognition and epigenetics. In doing so, the factor-level and indicator-level associations with healthcare utilization are evaluated, with an eye toward testing the validity of a superordinate SR construct. Objectives and hypotheses: This study examined the factor structure of SR as a higher-order construct, encompassing epigenetic age and cognitive function. It compared factor-level associations with healthcare utilization (emergency department [ED] visits and hospitalizations) to those observed at the indicator-level, while evaluating age and sex as potential moderators. A priori hypotheses were that a higher-order model would best fit the data, the associations would be stronger for women (vs. men) and older (vs. middle-aged) adults, and significant pathways would emerge at both the factor and indicator levels. Methods: Data were drawn from the Canadian Longitudinal Study on Aging (CLSA) Comprehensive Cohort (n=30,097; age range=45-85 at enrolment), focusing on the subsample who completed epigenetic assays (n=1,478). Structural equation modelling (SEM) was used to evaluate competing structural configurations and predict healthcare utilization. Age and sex moderation were examined using multi-group analysis. Logistic regressions were used to examine indicator-level associations. Results: A correlated two-factor model representing epigenetic age and cognitive function as distinct but related constructs, rather than as components of a higher-order SR model or a single unified factor was selected as the most appropriate model based on fit indices and parsimony. Cognitive function emerged as a predictor of hospitalizations at the factor-level (b -0.254; 95% CI; -0.461, -0.046). Supplemental analyses suggested no significant sex moderation, while evidence for age moderation was inconclusive. At the indicator-level, analyses suggested the mental alternation test (MAT) and intrinsic epigenetic age acceleration (IEAA) were reliable predictors of ED visits. Similarly, the animal fluency test (AFT) was a predictor of hospitalizations. Conclusions: Cognitive function and epigenetic age may be best considered as correlated, but fundamentally independent processes in older adults. Among factor-level predictors of healthcare utilization outcomes, cognitive function was reliable but not epigenetic age. At the indicator level, mental flexibility and cell-intrinsic aging were predictive of select facets of healthcare utilization. These findings suggest that both cognitive and epigenetic markers have some value in predicting future healthcare costs among older adults, but that systemic resilience may be less useful at the whole organism level.