Date of Award

Spring 2026

Document Type

Open Access Dissertation

Department

Epidemiology and Biostatistics

First Advisor

Anwar T. Merchant

Abstract

Introduction: Metabolic dysfunction, depressive symptoms, and physical activity are key and independent interrelated determinants of cognitive aging, yet the temporal ordering and causal pathways linking these factors remain incompletely understood. Specifically, metabolic dysfunction may act as both a risk factor for cognitive decline and as an exposure-induced mediator-outcome confounder when examining behavioral and psychological pathways. This dissertation used longitudinal data from the Health and Retirement Study (HRS) to examine the role of metabolic dysfunction, depressive symptoms, and physical activity in shaping later-life cognitive performance using advanced causal mediation methods.

Methods: We analyzed nationally representative longitudinal data from U.S. adults aged 50 years and older participating in the HRS between 2014 and 2020. Metabolic dysfunction was indexed by the triglyceride-glucose (TyG) index derived from fasting biomarkers. Cognitive performance was measured using the validated 27-point Langa–Weir cognitive score. Aim 1 examined whether TyG predicted subsequent cognitive performance over 2 and 4 year follow-ups using multivariable regression incorporating survey weights. Aims 2 and 3 applied the Recanting Twins (RT) causal mediation framework using the crumble package in R to estimate path-specific effects on the additive scale in the presence of exposure-induced mediator-outcome confounding. Aim 2 examined the pathways linking physical activity (2014) to cognition through depressive symptoms and metabolic dysfunction (2016). Aim 3 examined pathways linking depressive symptoms (2014) to cognition through physical activity and metabolic dysfunction (2016).

Results: In aim 1, higher TyG levels were associated with poorer subsequent cognitive performance, with stronger and more consistent associations emerging over the longer 4-year follow-up. Continuous modeling detected modest short-term associations, whereas quartile-based differences were more consistent at longer follow-up among participants in the highest TyG quartile (Q4). In aim 2, physically active individuals had higher cognitive performance at both follow-ups. The largest contribution to the total effect (ATE: 2018: 0.195 [95% CI: 0.19, 0.2]; 2020: 0.505 [95% CI: 0.476, 0.533] was the direct pathway from physical activity to cognitive performance independent of depressive symptoms and metabolic dysfunction (P1: A→Y, 2018: 0.122 [95% CI: 0.119, 0.126]; 2020: 0.366 [95% CI: 0.331, 0.400]),and the indirect pathways through depressive symptoms (P4: A→M→Y, 2018: 0.06 [95% CI: 0.029, 0.091]; 2020: 0.124 [95% CI: 0.09, 0.158]) were significant also at both follow-ups. The pathways involving metabolic dysfunction were only significant at the longer follow-up and were smaller in magnitude. The extent by which the exposure-induced mediator-outcome confounding component attributable to TyG index was not distinguishable from zero. In aim 3, elevated depressive symptoms predicted cognitive decline at both follow-ups (ATE: 2018: -0.413 [95% CI: -0.447, -0.379]; 2020: -0.733 [95% CI: (-0.756, -0.71]). The direct pathway independent of physical activity and metabolic dysfunction accounted for the largest portion of the total effect (P1: A→Y, 2018: -0.371 [95% CI: -0.380, -0.362]; 2020: -0.651 [95% CI: -0.673, -0.629]). Also, the behavioral pathway through physical activity was significant at both follow-ups (P4: A→M→Y, 2018: -0.041 [95% CI: -0.0753, -0.007]; 2020: -0.062 [95% CI: -0.069, -0.056]). The metabolic pathways emerged only at the longer follow-up and were modest in magnitude. The exposure-induced mediator-outcome confounding component attributed to metabolic dysfunction was small but statistically significant.

Conclusion: This dissertation demonstrates that metabolic dysfunction, depressive symptoms, and physical activity jointly shape cognitive aging through temporally evolving pathways. Metabolic dysfunction predicts later cognitive performance and plays a modest but increasingly detectable role in behavioral and psychological mediation pathways over longer follow-up and, in the depressive-physical activity-cognitive performance pathways, the TyG index representing the metabolic dysfunction acted as a an exposure-induced mediator-outcome confounding. Physical activity and depressive symptoms exerted strong direct behavioral and psychological effects on cognition, highlighting mental health and physical activity as key targets for intervention. These findings provide causal evidence that improving depressive symptoms, promoting physical activity, and addressing metabolic dysfunction may collectively contribute to preserving cognitive health in the aging population.

Rights

©2026, Ali A. Alshayhan

Download

Available for download on Wednesday, May 31, 2028

Included in

Epidemiology Commons

Share

COinS