Bi-directional associations between alcohol consumption and pain among non-manual workers: a random-intercept cross-lagged panel analysis in the British Whitehall II cohort study

Aims The direction and temporality of the association between alcohol use and pain remain understudied among non-manual workers. This study investigated bidirectional associations between alcohol consumption and pain among current and retired non-manual workers, with exploratory subgroup analyses by retirement transition status, retirement age, and socioeconomic position (SEP). Methods Survey data from Phases 7, 9, and 12 (2002–15) of the Whitehall II cohort study were analyzed (n = 5928, baseline mean age 60.7 years, 71% men). Alcohol consumption was assessed through self-reported intake converted to pure alcohol units. Pain severity was derived from musculoskeletal pain-site number and RAND-36 bodily pain measures, categorized as none, mild, or moderate/severe. Random-intercept cross-lagged panel models without equality constraints estimated within-person associations, with subgroup analyses by retirement transition and SEP. Results At baseline, 23.1% reported above-moderate pain severity and 30.0% exceeded recommended limits (>14 units/week). During 14-year follow-up, 47.3% remained retired, 10.1% were employed, and 40.1% transitioned from employment to retirement. Elevated alcohol consumption was associated with increased pain severity among all participants (βP7 → P9 = 0.07, 95% confidence interval [CI]: 0.02–0.12; βP9 → P12 = 0.04, 95% CI: 0.00–0.08), with stronger effect among midlife retirees (βP9→P12 = 0.15, 95% CI: 0.04–0.25) and low-SEP participants (βP9→P12 = 0.22, 95% CI: 0.04–0.51). Pain was associated with alcohol consumption at earlier intervals, but associations attenuated subsequently. Conclusions Elevated alcohol consumption might be associated with increased pain severity among non-manual workers, particularly midlife retirees and low-SEP individuals. Pain-to-alcohol associations were observed but were inconsistent across intervals.