Transitions between diabetes and frailty and their joint association with mortality in older Chinese adults: a five-state multi-state Markov analysis of the CHARLS cohort
Abstract Background. Diabetes and frailty frequently co-occur in older adults, but their joint contribution to mortality, and the transitions between the two conditions, remain poorly characterised in Chinese populations. Methods. We analysed five waves (2011–2020) of the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative cohort of community-dwelling adults aged ≥ 45 years (n = 16,658). A five-state Markov multi-state model (S0 neither condition; S1 diabetes only; S2 frailty only; S3 both; S4 death) was fitted to panel data with multiple imputation (m = 50). Death intensities were compared across states, additive interaction was quantified with the relative excess risk due to interaction (RERI), and transitions between the two conditions were summarised as transition intensity ratios (TIRs). Fifteen sensitivity analyses were pre-specified; two further analyses restricted the cohort to participants free of both conditions at baseline and reversed the assumed order of within-interval changes. Results. Over nine years 2,417 participants (14.5%) died. Death intensity was highest with frailty alone (≈ 0.16 per person-year) versus neither condition (≈ 0.07); diabetes alone added little, and additive interaction was unclear (RERI − 0.15, 95% CI − 1.70 to 1.39). Both transition ratios were below 1 (TIR diabetes→frailty 0.22, 95% CI 0.18–0.29; TIR frailty→diabetes 0.65, 95% CI 0.49–0.85). Restricting to the 14,061 participants free of both conditions at baseline left the first ratio essentially unchanged (0.248, 95% CI 0.191–0.305) but reduced the second to 0.099 (95% CI 0.056–0.141), indicating that this direction is shaped by the prevalent-case design. Conclusions. Frailty carries the dominant mortality burden in this cohort and diabetes adds little beyond it, with no clear evidence of additive interaction. Transition ratios between the two conditions are not interpretable as causal acceleration; they describe associations conditional on the baseline state. Early frailty detection and shared risk-factor reduction should be prioritised.
Authors
- Tianrui Cun (ORCID: https://orcid.org/0009-0007-0043-2590)
- Aidi Xu
Institutions
- Chuxiong Normal University (CN)
- The People's Hospital of Chuxiong Yi Autonomous Prefecture (CN)
Publication Details
- Journal
- Research Square
- Published
- 2026-11-12
- DOI
- https://doi.org/10.21203/rs.3.rs-11113516/v1
- Primary Topic
- Frailty in Older Adults
- Type
- preprint