Moderating effect of country-level health determinants on the association between hobby engagement and mental health: cross-cohort multi-level models, meta-analyses, and meta-regressions
Daisy Fancourt,Taiji Noguchi,Jessica K Bone,Jacques Wels,Qian Gao,Katsunori Kondo,Tami Saito,Hei Wan Mak
DOI: https://doi.org/10.1016/S0140-6736(23)02091-3
Abstract:Background: Theories from anthropology, evolutionary psychology, and sociology have focused on the potential adaptive benefits of hobby engagement for mental health in older adults. However, previous studies have used data from single countries, potentially biased by specific measurement and methodological approaches, cohort effects, or cultural specificities. Whether there are genuine benefits for mental health in older adults cross-culturally remains unknown. This study explored the consistency of this association across 16 different nations. Methods: For this epidemiological study, we used data from adults aged 65 years or older across 16 countries in the USA, Europe, and Asia, represented in five longitudinal studies (ELSA, JAGES, HRS, SHARE and CHARLS; N=93 263, 45-62% female, mean age 72-76 years, data collected 2008-20). We harmonised measures of self-reported engagement in hobbies and past-times, depressive symptoms (validated scales), and Likert scale responses for self-reported health, happiness, and life satisfaction. We conducted fixed-effects models and longitudinal regression models of hobbies and mental health for each country and then pooled in multinational meta-analyses. We accounted for all time-constant factors including those unobserved (eg, genetics, past leisure behaviour, medical history, psychological traits) and identified time-varying factors (eg, sociodemographic background, clinical conditions, daily functioning). We tested the potential moderating effects of country-level determinants of health in meta-regressions and multilevel models. Findings: Meta-analytic fixed-effects findings showed that having a hobby was associated with fewer depressive symptoms (pooled coefficient -0·10, 95% CI -0·13 to -0·07, I2=69·5%, H2=3·28), and higher levels of self-reported health (0·06, 0·03 to 0·08, I2=48·1%, H2=1·93), happiness (0·09, 0·06 to 0·13, I2=67·0%, H2=3·03), and life satisfaction (0·10, 0·08 to 0·12, I2=33·6%, H2=1·51). Results were consistent in meta-analyses of longitudinal regression models testing directionality of findings. Macro-level factors such as life expectancy, world happiness index, country wealth, and income inequality predicted prevalence of hobby engagement, but they showed only marginal moderating effects on the association between hobbies and mental health. Interpretation: Despite some heterogeneity in measurement between the cohorts, the apparent universality of the health benefits of hobbies internationally suggests that facilitating greater opportunities for engagement across demographic groups and between countries could be an important part of multidisciplinary care. Findings have implications for social prescribing schemes (currently in trial in many countries) and multidisciplinary work on origins and human behavioural patterns of hobby engagement. Funding: National Endowment for the Arts, Wellcome Trust, Belgian Nnational Scientific Fund (FNRS).