Drinking at least weekly, loneliness and social isolation over time: Bidirectional within‐person associations in an Australian prospective cohort study

Abstract Background and aims Past research has shown a link between alcohol consumption and loneliness/social isolation; however, the nature of the relationship remains unclear. We aimed (1) to estimate bidirectional longitudinal associations between drinking at least weekly, loneliness and social isolation; and (2) to distinguish within‐person change from stable between‐person differences. Design Prospective cohort study using ten annual outcome waves (2014–2023), with wave 13 (2013) providing descriptive context and time‐constant baseline socio‐demographic covariates. Random‐intercept cross‐lagged panel models (RI‐CLPMs) were primary, dynamic panel models (DPMs) assessed robustness and conventional cross‐lagged panel models were sensitivity analyses. Models were unadjusted, partially adjusted for baseline socio‐demographic covariates and fully adjusted for additional time‐varying covariates. RI‐CLPMs separated deviations around participants' expected levels from relatively time‐invariant differences in their average levels. Setting Australia; the Household, Income and Labour Dynamics in Australia (HILDA) Survey is a nationally representative household panel, although analyses were unweighted. Participants Unbalanced panel of 33 835 Australians aged 15+. Former‐drinker person‐wave observations were excluded. Across waves 13–23, 87.8%–93.3% of individual‐interview respondents completed the self‐completion questionnaire. Measurements Time‐constant baseline covariates were measured at wave 13, and outcomes were measured at waves 14–23. Drinking at least weekly (≥1 day/week) was derived from an annually administered drinking‐frequency item. Loneliness was derived from three 7‐point items and social isolation was operationalised as low perceived social support using four 7‐point items; responses ranged from 1 (strongly disagree) to 7 (strongly agree). Median scores ≥4 were classified as lonely and scores <4 as socially isolated, following prior HILDA operationalisations. Results Among observed wave 13 responses, 6719/14 019 (47.9%) reported drinking at least weekly, 3884/14 076 (27.6%) were classified as lonely and 812/14 082 (5.8%) as socially isolated; participants were 50.9% female and had a mean age of 36.0 years (standard deviation = 22.8). In fully adjusted RI‐CLPMs, loneliness predicted subsequent social isolation [pooled unstandardised coefficient (b) = 0.020, 95% confidence interval (CI) = 0.002–0.037 and social isolation predicted subsequent loneliness (b = 0.028, 95% CI = 0.007–0.049). Drinking at least weekly was autoregressive (b = 0.151, 95% CI = 0.135–0.168) but had no clear prospective association with either social measure. DPM conclusions were consistent with RI‐CLPM conclusions. Summary Loneliness and social isolation appeared to reinforce each other over time within individuals in this Australian cohort. Drinking at least weekly was not consistently associated with within‐person changes in social connection at annual intervals.