Answers are grounded in this paper and cite their sources.
Background and Objectives: Dysglycaemia and a modified partial physical-frailty phenotype are each associated with cognitive decline. We examined whether their joint association with episodic memory decline was additive across three ageing cohorts, and whether the same pattern held for mortality. Methods: We analysed the China Health and Retirement Longitudinal Study (CHARLS, n = 5842), the English Longitudinal Study of Ageing (ELSA, n = 2520, with a wave-6 landmark) and the Health and Retirement Study (HRS, n = 6224). Dysglycaemia was defined from cohort-specific HbA1c or self-reported diabetes; the frailty exposure required at least one of the available modified physical-frailty criteria. The outcome was immediate plus delayed recall at or below the participant’s own baseline score minus 1.5 cohort-specific standard deviations at one or more follow-up assessments. Modified Poisson models with robust variance estimated relative risks and the relative excess risk due to interaction (RERI) with delta-method confidence intervals; cohort estimates were pooled by inverse-variance weighting. Mortality in HRS and a set of sensitivity analyses were also examined. Results: The three cohorts contributed 14,586 participants and 2860 observed decline events. Each exposure was associated with higher risk of decline. The pooled RERI was −0.262 (95% CI −0.465 to −0.059; p = 0.011) with no evidence of heterogeneity (Q = 0.16, df = 2, p = 0.92; I2 = 0%). Cohort-specific RERIs were −0.211 (CHARLS), −0.298 (ELSA) and −0.290 (HRS), and the joint relative risk was below the additive expectation in all three cohorts. The model-based population attributable fraction for either exposure was 9.5% (95% CI 5.8 to 13.2), of which 82% was assigned to singly exposed participants; 2692 of the 2860 observed decline cases (94.1%) occurred outside the jointly exposed group. When decline and death were combined in the 6645 participants in whom that composite is observable, the RERI was −0.027 (95% CI −0.270 to 0.216). Standardised six-year mortality rose from 10.4% to 20.3%, and on a common six-year window the paired comparison of risk-scale interactions was inconclusive (difference −0.013, 95% CI −0.065 to 0.036). Conclusions: The joint association with memory decline was sub-additive, and model-based burden was concentrated in singly exposed participants. The pattern was confined to an endpoint defined relative to each participant’s own baseline recall, did not extend to a composite of decline or death, and does not establish screening or treatment effects.
No takes yet. Share an insight, caveat, or question.
Li et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: