Circadian syndrome and incident multisite pain among middle-aged and older Chinese adults: component-specific and longitudinal analyses from CHARLS

Circadian syndrome (CircS) combines short sleep duration, depressive symptoms, and metabolic abnormalities, but it is unclear whether the composite provides information beyond its individual components. We examined the prospective association of CircS with incident multisite pain and evaluated component-specific contributions, follow-up structure, and selection bias. We used 2011 baseline data from the China Health and Retirement Longitudinal Study and pain assessments in 2013 and 2015, providing up to four years of follow-up. CircS was defined as at least four of seven components. Participants with baseline multisite pain were excluded, although those with single-site pain were retained. Modified Poisson regression with robust standard errors was the primary analysis for risk ratios (RRs). Cox models were supplementary because onset was interval-censored. We additionally examined individual components, leave-one-component-out definitions, alternative thresholds, a flexible score model, a discrete-time model, inverse-probability-of-selection weighting, survey-weighted estimates, and extended confounder adjustment. Among 6,124 participants initially free of multisite pain, 1,514 (24.7%) developed multisite pain during follow-up. CircS was associated with higher risk in the fully adjusted primary model (RR = 1.19, 95% CI: 1.09–1.30). The association remained in the 2015-only analysis (RR = 1.16, 95% CI: 1.04–1.30) and after excluding all baseline pain (RR = 1.18, 95% CI: 1.07–1.31). Depressive symptoms (RR = 1.86) and short sleep (RR = 1.41) showed the strongest component associations. Removing depressive symptoms from the CircS definition attenuated the association to RR = 1.04 (95% CI: 0.95–1.13). Adding CircS to a conventional model changed the tenfold cross-validated AUC from 0.669 to 0.672. Baseline CircS was associated with incident multisite pain among adults initially free of multisite pain, but not necessarily pain-free. Component and leave-one-out analyses indicated that depressive symptoms and short sleep accounted for much of the association, and CircS added little discrimination beyond conventional factors. CircS should therefore be interpreted as a composite marker of co-occurring vulnerabilities rather than evidence of an independent integrated mechanism.