Available via license: CC BY 4.0
Content may be subject to copyright.
Page 1/22
Psychological benets of hobby engagement in
older age: a longitudinal cross-country analysis of
93,263 older adults in 16 countries
Daisy Fancourt ( d.fancourt@ucl.ac.uk )
University College London
Hei Wan Mak
Department of Behavioural Science and Health, Institute of Epidemiology & Health Care, University
College London https://orcid.org/0000-0002-2013-1644
Taiji Noguchi
Department of Social Science, Center for Gerontology and Social Science, Research Institute, National
Center for Geriatrics and Gerontology
Jessica Bone
Department of Behavioural Science and Health, Institute of Epidemiology & Health Care, University
College London
Jacques Wels
MRC Unit for Lifelong Health and Ageing, University College London
Qian Gao
Department of Behavioural Science and Health, Institute of Epidemiology & Health Care, University
College London
Katsunori Kondo
Department of Social Preventive Medical Sciences, Center for Preventive Medical Sciences, Chiba
University
Tami Saito
Department of Social Science, Center for Gerontology and Social Science, Research Institute, National
Center for Geriatrics and Gerontology
Article
Keywords:
Posted Date: December 9th, 2022
DOI: https://doi.org/10.21203/rs.3.rs-2349781/v1
Page 3/22
Abstract
The growing ageing population has begun to pose a threat to global health due to social and
psychological challenges experienced by older adults. To mitigate this, many countries promote hobby
engagement to support and improve mental health. Yet, it remains unclear whether there is consistency in
benets across different cultural settings. We harmonised measures of hobby engagement and multiple
aspects of mental health across 16 nations represented in ve cohort studies. Fixed effects models and
multi-national meta-analyses were applied to compare the longitudinal associations between hobbies
and mental health internationally. Independent of confounders, having a hobby was associated with
fewer depressive symptoms, and higher levels of self-reported health, happiness and life satisfaction. But
prevalence of engagement and benets correlated with macro-level factors such as life expectancy and
national happiness levels. Consequently, ensuring equality in hobby engagement within and between
countries should be the priority.
Introduction
Ageing populations are an increasing global concern given the social and psychological challenges older
adults can experience, including loneliness, social isolation and worsening mental health, all of which are
associated with increasing physical multimorbidity and mortality1,2. In 2020, there were one billion people
aged 60 years or over. The World Health Organization predicts that the number will continue to rise, with
one in six people aged 60 or above globally by 2030 and an ageing population of 2.1 billion by 2050. This
is placing untenable burdens on global health and social care services, providing nancial and workforce
planning predicaments. To help meet older adults’ needs and to support the sustainability of health and
social care systems globally, there is a need to explore cost-effective strategies to enhance older adults’
mental health.
There is increasingly global interest in how engagement in psychosocial activities could address these
challenges3,4.Hobbies (dened as activities that people engage in during their leisure time for pleasure,
such as listening to and making music, drawing, dancing, model-making, reading, playing games, and
participating in societies/clubs) involve imagination, novelty and creativity, sensory activation, self-
expression, relaxation, and cognitive stimulation, all of which are positively related to mental health via
psychological, biological, social and behavioural pathways5. Participation in hobby groups can
additionally provide social support and reduce loneliness and social isolation5. For this reason, many
countries including the UK6, Japan7, the United States (US)8 have been promoting hobbies and leisure
activities as part of their policies and recommendations to support and improve mental health, with a
particular focus on investment to support participation amongst older adults.
These policies are underpinned by a large body of research which has shown how hobbies can enhance
multi-dimensional aspects of mental health, including negative symptomatology and clinical diagnoses
of depression and psychiatric conditions, experiential wellbeing (emotional quality of momentary and
everyday experiences), evaluative wellbeing (e.g. life satisfaction) and eudemonic wellbeing (e.g. purpose
Page 4/22
in life) for older adults. Meta-analyses of both observational and interventional studies involving
engagement in nature-based activities and volunteering have shown protective associations with
depressive symptoms9–12,. These ndings are supported by individual studies in the US (n=16,043)13,
Japan (n=37,627)14, UK (n=8,780)15, and China (n=6,287)16 showing concurrent and longitudinal
relationships (three to 12 years follow-up) between other types of hobbies such as community groups,
arts, and social clubs and lower incidence and prevalence of depression in older adults. Similarly, meta-
analyses of various types of leisure activities such as dancing, nature-based activities and gardening
have reported benets for positive aspects of wellbeing10,17–20. Again, these ndings are supported by
individual studies from the Sweden (n=157 twins)21, UK (n=10,989)22, Japan (n=6,093)23, and US
(n=12,111)24 focusing on broader activities such as volunteering, arts, cultural engagement and indoor
gardening.
However, the literature to date is hampered by several limitations. First, studies have focused on single
countries at a time, so given differences in denitions, outcome measures, and methodological
approaches between studies, it is unclear whether there is consistency in results across different cultural
settings, and thus whether ndings from one country population could be applied to populations in other
countries. Second, many studies have focused on specic sub-categories of hobbies (e.g. volunteering vs
nature-based activities vs arts participation vs cultural engagement etc), often applying conicting
denitions. Yet all hobbies share common 'active ingredients’ and activate similar causal mechanisms of
action; it has been proposed that there is little to differentiate in their potential to affect population-level
mental health outcomes25,26. Individualmeta-analyses focusing on specic hobby denitions thus
present only a fraction of the literature available on the topic and provide an incomplete picture to policy
makers.
The present study was therefore designed to harmonise measures of hobby engagement and mental
health in older adults across 16 nations represented in ve cohort studies. We undertook novel analyses
of representative data from the English Longitudinal Study of Ageing (ELSA), Japan Gerontological
Evaluation Study (JAGES), US Health and Retirement Study (HRS), Survey of Health, Ageing and
Retirement in Europe (SHARE), and China Health and Retirement Longitudinal Study (CHARLS). We
followed participants for three consecutive waves (between four and eight years). Fixed effects models
tested the longitudinal associations between engagement in hobbies and mental health, simultaneously
accounting for all time-constant factors (regardless of whether they were observed, e.g. genetics, past
leisure behaviours, medical histories and psychological traits) and identied time-varying factors (e.g.
socio-demographic backgrounds, clinical conditions, and diculties with activities of daily living). We
then pooled our ndings into novel multi-national meta-analyses.
Results
Participants
Page 5/22
To allow for comparison across all datasets, we limited the samples to participants aged 65 years and
above. To explore how changes in hobby engagement were associated with changes in older adults’
mental health over time, a total of 93,263 respondents who provided data across all study measures were
analysed: Austria (N = 2,524), Belgium (N = 2,304), China (N = 1,611), Czech Republic (N = 2,664), Denmark
(N = 1,006), England (N = 4,267), Estonia (N = 3,584), France (N = 2,705), Germany (N = 966), Italy (N =
1,915), Japan (N = 57,051), Slovenia (N = 1,272), Spain (N = 2,099), Sweden (N = 1,315), Switzerland (N =
1,776), and the US (N = 6,204).
The average age of the respondents across the different countries was between 71.7 and 75.9 years.
Generally, there was a higher proportion of females participating in the surveys (except for China, Japan
and Germany). More than seven out of ten were retired, except for those living in China, Japan and Spain.
Over 60% of the participants experienced long-standing mental or physical health conditions (Table S1).
For hobby engagement, Denmark (96.0%), Sweden (95.8%) and Switzerland (94.4%) had the highest
engagement levels, followed by Germany (91.0%), Austria (90.0%) and Japan (90.0%). China (37.6%),
Spain (51.0%), and Italy (54.0%) had the lowest engagement levels (Fig.1; Table S1).
(Fig.1)
Depressive symptoms
Older adults with a hobby reported fewer depressive symptoms over four to eight years follow-up. The
associations were found in Austria (coef=-0.24, 95%CI=-0.35,-0.13), Estonia (coef=-0.14,
95%CI=-0.22,-0.06), France (coef=-0.23, 95%CI=-0.31,-0.15), Japan (coef=-0.13, 95%CI=-0.15,-0.11), Spain
(coef=-0.12, 95%CI=-0.19,-0.05), Sweden (coef=-0.20, 95%CI=-0.34,-0.06), and the US (coef=-0.09,
95%CI=-0.13,-0.05) (Fig.2). Meta-analysis of the ndings showed that hobby engagement was overall
negatively associated with depressive symptoms (pooled coef=-0.10, 95%CI=-0.13,-0.07, I2 = 69.5%, H2 =
3.28).
Self-reported health
Older adults living in Austria (coef = 0.15, 95%CI = 0.05,0.25), Japan (coef = 0.10, 95%CI = 0.07,0.13),
Spain (coef = 0.08, 95%CI = 0.03,0.13), Sweden (coef = 0.22, 95%CI = 0.07,0.37), and Switzerland (coef =
0.20, 95%CI = 0.08, 0.32), reported greater self-reported health over follow-up if they had a hobby,
compared to those without (Fig.3). Meta-analysis of the ndings showed that hobby engagement was
overall positively associated with self-reported health (pooled coef = 0.06, 95%CI = 0.03,0.08, I2 = 48.1%,
H2 = 1.93).
Happiness
For happiness, analyses showed that older adults living in Austria (coef = 0.16, 95%CI = 0.05,0.27), China
(coef = 0.21, 95%CI = 0.10,0.32), England (coef = 0.10, 95%CI = 0.04,0.16), Germany (coef = 0.18, 95%CI =
0.03, 0.33), Italy (coef = 0.21, 95%CI = 0.14,0.28), Japan (coef = 0.10, 95%C = 0.08,0.12), Sweden (coef =
Page 6/22
0.21, 95%CI = 0.04,0.38), and the US (coef = 0.11, 95%CI = 0.07,0.15) were happier if they had a hobby
(Fig.4). Meta-analysis of the ndings showed that hobby engagement was overall positively associated
with happiness (pooled coef = 0.09, 95%CI = 0.06,0.13, I2 = 67.0%, H2 = 3.03).
Life satisfaction
Having a hobby was also related to higher levels of life satisfaction amongst older adults, particularly
those living in Austria (coef = 0.18, 95%CI = 0.07,0.29), Czech Republic (coef = 0.17, 95%CI = 0.06,0.28),
Denmark (coef = 0.26, 95%CI = 0.07,0.45), England (coef = 0.07, 95%CI = 0.02,0.12), Estonia (coef = 0.14,
95%CI = 0.04,0.24), France (coef = 0.11, 95%CI = 0.03,0.19), Japan (coef = 0.09, 95%CI = 0.07,0.11), Spain
(coef = 0.17, 95%CI = 0.11,0.23), Switzerland (coef = 0.18, 95%CI = 0.07,0.29), and the US (coef = 0.10,
95%CI = 0.06,0.14) (Fig.5). Meta-analysis of the ndings showed that hobby engagement was overall
positively associated with life satisfaction (pooled coef = 0.10, 95%CI = 0.08,0.12, I2 = 33.6%, H2 = 1.51).
(Figs.2–5)
Country-level predictors
Hobby engagement was shown to positively correlated with the world happiness index score27 (r = 0.63)
and life expectancy28 (r = 0.39), and were negatively correlated with country wealth29 (r=-0.62) and the
Gini index30 (r=-0.63) (Figs.6a-6d).
(Figs.6a-6d)
Meta-regressions were then conducted to explore whether the between-study heterogeneity can be
explained by these country-level predictors. For the prevalence of hobby engagement (Figures S1a-S1d),
country wealth (Figures S2a-S2d) and Gini index (Figures S3a-S3d), no associations were found between
effect sizes and the predictors. For the world happiness index score, no associations were shown
between effect sizes and the predictor, except for life satisfaction (Figures S4a-S4d). Results showed that,
with a condence level of 90%, for every additional unit in the world happiness index score, the effect size
of a study is expected to rise by 0.05 (95%CI=-0.01,0.11; p = 0.079). Finally, for life expectancy, a positive
correlation was shown between the predictor and self-reported health effect sizes; for every year increase
in life expectancy across countries, the association between hobby engagement and self-reported health
was 0.01-points larger (coef = 0.01, 95%CI = 0.01,0.02). No associations were shown for other outcomes
(Figures S5a-S5d).
Sensitivity analysis
When using multiple imputation to account for missing data, results were largely replicated (Table S2).
When including respondents aged 55 and above (except for Japan where all participants were aged 65
and over), the evidence for longitudinal associations between hobby engagement and the outcomes
across countries became stronger, likely due to the increase in sample size (Table S3).
Page 7/22
When stratifying the samples by gender, we found some variations between countries. However, our
pooled effect sizes from meta-analysis showed that engagement in hobbies remained benecial for both
females (depressive symptoms: pooled coef=-0.10, 95%CI=-0.15,-0.06; self-reported health: pooled coef =
0.06, 95%CI = 0.03,0.08; happiness: pooled coef = 0.08, 95%CI = 0.03,0.12; life satisfaction: pooled coef =
0.10, 95%CI = 0.07,0.14) and males (depressive symptoms: pooled coef=-0.09, 95%CI=-0.12,-0.06; self-
reported health: pooled coef = 0.06, 95%CI = 0.03,0.09; happiness: pooled coef = 0.10, 95%CI = 0.07,0.12;
life satisfaction: pooled coef = 0.09, 95%CI = 0.06,0.11) (Figures S6a-S9b).
Finally, a subgroup meta-analysis was conducted to assess whether the type of hobbies measures (i.e.
binary measure or index created from a list of options) was responsible for differences in effect sizes
between studies. Pooled analyses showed no subgroup differences between the measures and hobbies
continued to associate with all outcomes (p > 0.05; Table S4).
Discussion
This study was the rst to compare and contrast longitudinal associations between hobby engagement
and multidimensional aspects of mental health across 16 countries. The prevalence of hobby
engagement varied substantially across countries, from just two in ve people to over nine in ten. Meta-
analysis of the ndings revealed that having a hobby was associated with fewer depressive symptoms,
better self-reported health, more happiness and higher life satisfaction, with life satisfaction most
consistently related to hobbies. Sensitivity analysis showed that these associations did not vary by
gender. Both males and females could experience the benets of hobbies for their mental health. On
average, more older adults had hobbies in countries with higher world happiness index score and life
expectancy. Further, the relationship between hobby engagement and life satisfaction was stronger in
countries with higher happiness levels and the relationship between hobby engagement and self-reported
health was stronger in countries with longer life expectancy.
Our ndings are in line with the rich cross-disciplinary, international literature indicating that having a
hobby may enhance mental health amongst older adults and that such enhancement can be found in
both female and male participants. They also extend previous literature in several keyways. First, our
pooled effect sizes from meta-analysis revealed that hobby engagement was associated with fewer
depressive symptoms and higher levels of self-reported health, happiness and life satisfaction. The
results provide evidence for the consistency in results across cultural settings and countries, highlighting
the relevance to global public health. In particular, our results show that, of the four outcomes, hobby
engagement has the most consistent association with life satisfaction. Life satisfaction is a subjective
evaluation of one’s social, emotional and physical wellbeing, which can be independent of ‘objective’
health status or functional ability that tends to decline with age31. Hobbies have many qualities that
could enable them to support older adults’ life satisfaction, including feeling in control of their minds and
bodies, nding a purpose in life, and feeling competent in tackling daily issues26. The result for life
satisfaction is particularly promising given the benets of hobbies were not only seen in healthier
Page 8/22
samples but also in samples such as the US where a very high proportion of the respondents were living
with long-standing mental or physical health conditions.
However, it is also relevant to consider why, even in the context of signicant meta-analytic ndings,
associations between hobby engagement and mental health showed some variation across countries
and cultures. Notably, results from meta-regressions show that there may be a positive correlation
between world happiness index score and effect sizes for life satisfaction, suggesting that the effect
sizes of studies increase according to how happy people are on a country-level. Similarly, the effect sizes
of studies for self-reported health are also likely to increase with an increase in life expectancy of the
country. One interpretation for this is that individuals living in countries with higher life expectancies or
countries with greater happiness index may be more likely to have a hobby as shown in Figs.6c & 6d (e.g.
Denmark, Sweden and Switzerland), which may have inated the size of the coecients. However, there
are some exceptions. For example, while people living in Spain had a lower engagement rate
comparatively, the strength of the association between hobby engagement and life satisfaction was
similar to countries with much higher engagement rates (including Austria, Czech Republic and
Switzerland). This suggests that the health benets of hobbies (at least for life satisfaction) are not
simply driven by high prevalence of engagement rates but can also be found in countries where hobbies
are less popular. Similarly, the longitudinal protective associations between hobbies and the outcomes
continued to be found in countries with lower world happiness index scores such as Japan and China as
well as countries with lower life expectancy like the US. Taken together, our results suggest that having a
hobby may have the potential to be associated with improvements in health amongst the older
population universally. However, the differential participation rate across countries implied that some
countries may experience greater barriers to engaging in hobbies.
Indeed, hobbies are often perceived as an ‘asset’ possessed by older people who are healthier, happier
and wealthier. Within countries, previous literature has highlighted a social gradient in hobby
engagement, where gender, social class, ethnicity and health conditions could inuence the likelihood of
engagement amongst older adults32. Although our analyses adjusted for individual socio-economic
factors, we still found greater hobby engagement rates in more auent countries. Differences in hobby
participation are thus concerning, as they could contribute to or exacerbate health inequalities both within
and across countries. As a result, it is important that in working to capitalise on the ndings presented
here, a systematic approach is taken, considering both how to address individual-level barriers to hobby
engagement that older adults may face, as well as considering how societal interventions could be
designed to build stronger relationships at a public health level between hobby engagement and mental
health outcomes. In recent years, public health strategies such as social prescribing schemes in the UK,
US, Japan and part of Europe have focused on building hobby engagement into healthcare services,
providing new referral pathways that can help to address existing individual and societal barriers to
engagement, positively inuencing motivations and propensity to engage amongst older populations,
and in turn providing opportunities to strengthen the associations between hobby and health outcomes.
Page 9/22
Overall, our ndings have important policy and health implications for older adults, especially those who
are retired (the majority in our sample). Contemporary life-course research has demonstrated that the
concept of ageing has shifted from seniority to an emphasis on lifestyle and consumption34. This aligns
with the idea of ‘the third age’ emphasised in previous research, which suggests that older adults who
enter the retirement age are now presented opportunities for self-development and are liberated from the
previous label of an ‘old age pensioner’ and from ‘the fourth age’ of decline and dependency34. As
suggested in our ndings, hobbies such as physical activity, arts and cultural engagement, and social and
community participation have the potential to lengthen ‘the third age’ period and make it one of
‘productive ageing’ through protecting against age-related declines in mental health and enhancing
wellbeing, which have profound consequences for morbidity and mortality.
There are many strengths in this study including the use of ve nationally representative datasets to
compare and contrast the longitudinal associations between hobby engagement and health across 16
nations. It is also the rst study using public surveys to compare hobby engagement rates internationally,
as well as assessing the strengths of the associations against population statistics relating to country
wealth, Gini index, world happiness index score, and life expectancy. We were also able to stratify the
samples by gender to test the gender differences in the associations. However, the study is not without
limitations. Whilst xed effects analysis was used in which all time-invariant variables and important
time-varying variables were modelled, causality cannot be established due to the use of observational
data. Further, whilst there was overall relative homogeneity in the way that questions about hobbies were
asked and we collapsed questions into a binary indicator, some countries chose to list examples whereas
others did not, which may have led to differences in interpretations of the question by participants.
However, no differences in associations between hobbies and outcomes were found with different
measures. Future research is needed to also consider the frequency and the length of hobby engagement
in different countries. On a related note, more research is encouraged to explore whether modulation of
specic active ingredients in hobby engagement (such as the presence or absence of physical activity or
social interaction) could affect outcomes.
Conclusions
Our study provides three main additions to existing literature. First, similar to inequalities in life
expectancy, world happiness index score and country wealth, there are inequalities in hobby engagement
amongst older adults across countries. Over 90% of the sample in countries including Denmark, Sweden,
Switzerland and Germany reported having a hobby, whereas roughly a half of the sample living in
countries such as Italy, Spain and China (less than a half) had a hobby. Second, our xed effects analysis
revealed that having a hobby was longitudinally associated with fewer depressive symptoms, and higher
levels of self-reported health, happiness and life satisfaction. Third, our pooled effect sizes from meta-
analysis conrmed the associations between hobbies and the outcomes, with life satisfaction having the
strongest and most consistent association. No gender differences were found in meta-analysis. These
results provide evidence that the associations between hobby and the health outcomes are applicable in
Page 10/22
different countries and settings. However, we found that there may be differences in the associations
according to world happiness index score and life expectancy. In particular, the associations between
hobby engagement and self-reported health and life satisfaction may be stronger in countries with higher
world happiness index score and life expectancy.
This study has numerous policy and practical implications in public health. The universal health benets
of hobbies suggest that removing differences in hobby engagement across demographic groups and
between countries should be the priority; not only could it help relieve the burden of the healthcare system
that many countries have been experiencing, but it also has the potential to extend ‘the third age’ period
where older adults can enjoy hobbies for enjoyment and self-development. In this instance, further work is
required to identify, compare and contrast the enablers and barriers of hobby engagement across
different countries. Further, results from this study can also be used as evidence when formulating and
developing schemes to integrate psychosocial interventions into health services or public health
strategies (e.g. through social prescribing schemes) to reduce morbidity, mortality and healthcare burden
and enhance ageing experiences amongst older adults.
Online Methods
Data
ELSA.
ELSA started in 2002/03 and follows over 11,000 participants aged 50 years and above every two
years35. In this study, to be in line with the other datasets, we extracted a sample of adults aged 65 years
of above who responded in Waves 7 (2014/15; response rate=78.3%-81.4%), 8 (2016/17; response
rate=82.4%), and 9 (2018/19; response rate=79.5%) where hobby engagement and outcome variables
were measured. We only considered respondents providing data across all measures, which left an
analytical sample of 10,876 observations from 4,267 participants (2.5 per person, ranging from 2 to 3).
The study received ethical approval from the National Research Ethics Service Committee and all
participants provided informed consent.
JAGES.
JAGES is a large-scale population-based longitudinal cohort study about ageing established in
2010, mainly collected through self-administered mail surveys, targeting older people aged 65 and older
who do not receive long-term care insurance benets36. JAGES has conducted joint survey with
municipalities that are the public insurers of long-term care insurance every 3-4 years: Wave 1 (2010/11)
to Wave 4 (2019/2020). This study used data from Wave 2, 3 and 4 (30 to 64 municipalities [response
rate: 52.4% to 71,1%]). Of them, the sample with complete data on hobby engagement and health
outcomes in at least two waves were recruited. Therefore, the sample of 125,901 observations from
57,051 participants (2.4 person, ranging from 2 to 3) were included in the nal analysis. The study
received ethical approval from Nihon Fukushi University, Chiba University and the National Center for
Geriatrics and Gerontology, and all participants provided informed consent.
Page 11/22
HRS.
HRS is a nationally representative study of more than 37,000 individuals over the age of 50 in the
US37. The study was initiated by the National Institute on Aging and conducted by the Institute for Social
Research at the University of Michigan to track the Baby Boom generation’s transition from work to
retirement. The initial HRS cohort was interviewed for the rst time in 1992 and followed-up every two
years, with other studies and younger cohorts merged with the initial sample. Together, these studies
create a fully representative sample of individuals over the age of 50 in the United States. Further details
on study design are reported elsewhere37. We used data from HRS waves at which participation in a
hobby was measured (2008-2018). At each wave, a rotating random 50% subsample of participants were
invited to an enhanced interview and given a Leave-Behind Psychosocial and Lifestyle Questionnaire to
complete and return by mail, which included questions on participation in community arts groups and
mental health38. Participants were eligible to complete this psychosocial questionnaire every four years.
Response rates in each year varied from 62% to 85%. We restricted the sample to participants aged 65
years and above, with complete data on hobby engagement and mental health outcomes in at least two
waves, and no missing data on time-varying covariates. This resulted in a nal analytical sample of
14,989 observations from 6,204 participants (2.4 observations per person, range 2-3).
SHARE.
The Survey of Health, Ageing and Retirement (SHARE is the largest pan-European social science
panel study providing internationally comparable longitudinal micro data on the population aged 50 and
over and currently includes eight waves with data collection starting in 2004. SHARE contains an original
sample as well as a refreshment sample to account for sample size reduction due to panel attrition.
SHARE has original core questionnaires as well as retrospective questionnaires (SHARELIFE, in Waves 3
and 8). In Waves 3 and 8, respondents answering the retrospective questionnaire were asked to answer a
reduced core questionnaire with fewer information, justifying the use of Waves 5, 6 and 7 in this study.
Data information for these three waves is available for twelve countries. Data was not available over
these three waves for Croatia, Greece, Hungary, Israel, Luxembourg, the Netherlands, Poland and Portugal.
The analytical sample by country, including non-response at baseline, is 2,524 in Austria; 2,304 in
Belgium; 2,664 in Czech Republic; 1,006 in Denmark; 3,584 in Estonia, 2,705 in France, 966 in Germany;
1,915 in Italy; 1,272 in Slovenia; 2,099 in Spain; 1,315 in Sweden and 1,776 in Switzerland. The study
received ethical approval from the Ethics Council of the Max Planck Society and all participants provided
informed consent.
CHARLS
. The China Health and Retirement Longitudinal Study (CHARLS) is a nationally representative
sample of Chinese residents aged 45 and above39. The baseline survey started in 2011 and has been
followed up every two years (in 2013 and 2015). Multistage probability sampling was used for sample
selection. The baseline included 17,708 individuals, and the response rates were over 80% in all three
waves (wave 1= 80.5%, wave 2=82.6% and wave 3=82.1%). The study only considered participants who
responded to all measures, with a nal analytical sample of 3,440 observations from 1,611 participants
(2.1 observations per person, range 2-3). The study received ethical approval from the Biomedical Ethics
Review Committee of Peking University, and all participants provided informed consent.
Measures
Page 12/22
Hobby and mental health.
Our measures of hobby and mental health are given in Table S5, which
presents the exact question-wording and item responses across datasets. The analysis will focus on four
types of mental health: (a) self-reported health, (b) life satisfaction, (c) happiness, and (d) depressive
symptoms. The measure items and response categories vary somewhat by country, reecting the cultural
differences across 16 nations. Therefore, to ensure the data were comparable, we harmonised and
recoded all variables, and standardised the outcome variables. We created a binary indicator of hobby
engagement (yes, no) in each country. Nonetheless, care needs to be taken in comparing the proportion of
hobby engaged and levels of various mental health outcomes across countries.
Time-varying covariates.
Nine time-varying variables that might confound observational associations
between hobby and mental health were identied for the analysis. These included
demographic
characteristics:
age (a continuous variable), partnership status (living with a partner/spouse vs not living
with a partner/spouse), number of people living in the household (a continuous variable);
socio-economic
position (SEP):
employment status (working vs not working), household income (a continuous variable),
housing tenure (home-owner vs not a home-owner); and
health proles:
long-standing mental/physical
conditions (yes vs no), diculties with daily activities (ADLs) (with diculties vs without diculties), and
diculties with instrumental activities of daily living (IADLs) (a continuous variable).
Statistical analysis
Due to data access restrictions, data were analysed separately for each country using xed-effects
regression, a longitudinal data method that controls for all time-invariant variables. For this reason, xed-
effects regression is considered to be more robust than traditional regression models such as OLS
regressions in exploring how changes in the predictor are associated with the changes in the outcomes.
Listwise deletion was applied to handle missing data. The proportion ofmissingness was as
follows:Austria (20.6%), Belgium (39.6%), China (69.7%; mainly due to missingness in household wealth,
hobby and life satisfaction), Czech Republic (34.8%),Denmark (56.6% ), England (15.1%), Estonia
(20.8%),France (21.2%),Germany (69.1%),Italy (51.3%),Japan (54.0%), Slovenia (54.8%),Spain
(53.8%),Sweden (62.5%),Switzerland (19.3%), and the US (15.7%). In our main analysis, we present
coecients and 95% condence intervals (CIs) to show the relationship between hobby engagement and
the outcomes across countries after adjustment for time-varying covariates.
In addition to the main analysis, we pooled our ndings into novel multi-national meta-analyses using the
random effect model to estimate the overall effect sizes for all outcomes. Pooled effect sizes and 95%CI
were reported. Between-study heterogeneity was estimated using the algorithm of the restricted
maximum likelihood and was assessed using I2 and H2 statistics. I2 indicates the percentage of
variability in the effect size that is caused by between-study heterogeneity, rather than by sampling
error40. A value of I2 greater than 50% indicates heterogeneity40. Similarly, the H2 statistics describes the
ratio of the observed variation and the expected variance due to sampling error40. A value of H2 greater
than 1 indicates the presence of between-study heterogeneity40.We also conduct a subgroup analysis by
Page 13/22
hobbies measures to explore whether the differences in effect sizes might have been attributed to the
way questions on hobbies were asked in each cohort study (a binarymeasure vs index created from a list
options). We further ran meta-regressions to explore the heterogeneity variance in our meta-analysis
using ve country-level predictors: the prevalence of hobby engagement, country wealth,29 world
happiness index score,27 life expectancy,28 and the Gini index.30 Finally, we performed a set of sensitivity
analyses to explore the robustness of the associations between hobby engagement and mental health:
a. To check that missing data did not inuence our ndings, we re-ran the analysis after using multiple
imputation by chained equations to impute missing data on hobby engagement, mental health
outcomes, and time-varying covariates across all included waves.
b. The main analysis only considered respondents aged 65 years and above to allow for comparison
across all datasets, but this signicantly restricted the sample size in the ELSA, HRS, SHARE, and
CHARLS data. This might reduce statistical power. To check the robustness of our main results, we
replicated the analysis using these four datasets and extended the sample to those who were aged
55 years and above.
c. To test for the consistency of the association between hobby engagement and mental health across
different population groups, we stratied our samples by gender (female vs male).
Declarations
Author Contributions
HWM, TN, JKB, JW, KK, TS and DF designed the study. HWM analysed data from ELSA, led the study, and
drafted the manuscript. TN analysed data from JAGES, JKB analysed data from HRS, JW analysed data
from SHARE, and QG analysed data from CHARLS. TN, JKB, JW, QG, KK, TS and DF assisted with
analytical issues, contributed to the writing and made critical revisions. All authors approved the nal
manuscript.
Data availability
The English Longitudinal Study of Ageing can be accessed via the UK Data Service:
https://beta.ukdataservice.ac.uk/datacatalogue/series/series?id=200011
The Health and Retirement Study can be accessed via the RAND Center for the Study of Aging:
https://hrsdata.isr.umich.edu/data-products/rand
The Survey of Health, Ageing and Retirement in Europe can be accessed via the SHARE Research Data
Center: http://www.share-project.org/data-access.html
The China Health and Retirement Longitudinal Study can be accessed via the National School of
Development, Peking University: https://charls.pku.edu.cn/en/
Page 14/22
Data of Japan Gerontological Evaluation Study are not open for public due to ethical concerns.
Code availability
All code used for these analyses is publicly available online: https://osf.io/84xzu/
Funding resources
The EpiArts Lab, a National Endowment for the Arts Research Lab at the University of Florida, is
supported in part by an award from the National Endowment for the Arts (Award: 1862896-38-C-20). The
opinions expressed are those of the authors and do not represent the views of the National Endowment
for the Arts Oce of Research & Analysis or the National Endowment for the Arts. The National
Endowment for the Arts does not guarantee the accuracy or completeness of the information included in
this material and is not responsible for any consequences of its use. The EpiArts Lab is also supported by
the University of Florida, the Pabst Steinmetz Foundation, and Bloomberg Philanthropies. DF is supported
by the Wellcome Trust (205407/Z/16/Z). JW is funded by the Belgian National Scientic Fund (FNRS)
Research Associate Fellowship (CQ) n° 40010931. The Japan Gerontological Evaluation Study (JAGES)
is supported by the Japan Society for the Promotion Science (JSPS) KAKENHI Grant Number
(JP15H01972), Health Labour Sciences Research Grant (H28-Choju-Ippan-002), Japan Agency for
Medical Research and Development (AMED) (JP18dk0110027, JP18ls0110002, JP18le0110009,
JP20dk0110034, JP21lk0310073, JP21dk0110037, JP22lk0310087), Open Innovation Platform with
Enterprises, Research Institute and Academia (OPERA, JPMJOP1831), and the Research Founding for
Longevity Sciences from National Center for Geriatrics and Gerontology (29-42, 30-22, 20-19, 21-20).
References
1. Holt-Lunstad, J. Why Social Relationships Are Important for Physical Health: A Systems Approach to
Understanding and Modifying Risk and Protection. Annu. Rev. Psychol. 69, 437–458 (2018).
2. Ageing: a 21st century public health challenge? Lancet Public Heal. 2, e297 (2017).
3. Steptoe, A. & Fancourt, D. Protecting physical health in people with mental illness. The Lancet
Psychiatry 6, 889–890 (2019).
4. Fancourt, D. & Finn, S.
What is the evidence on the role of the arts in improving health and well-
being? A scoping review
. Health Evidence Network synthesis report vol.67 (2019).
5. Fancourt, D. & Finn, S.
WHO Health Evidence Synthesis Report- Cultural Contexts of Health: The role
of the arts in improving health and well-being in the WHO European Region
. (2019).
.
Older people: independence and mental wellbeing
. (2015).
7. Labour and Welfare: Long-Term Care, Health and Welfare Services for the Elderly. Ministry of Health,
Labour and Welfare of Japan https://www.mhlw.go.jp/english/policy/care-welfare/care-welfare-
elderly/index.html.
Page 15/22
. Participating in Activities You Enjoy As You Age.
National Institute on Aging
https://www.nia.nih.gov/health/participating-activities-you-enjoy-you-age (2022).
9. Moore, A.
et al.
Volunteer Impact on Health-Related Outcomes for Seniors: a Systematic Review And
Meta-Analysis. Can. Geriatr. J. 24, 72 (2021).
10. Coventry, P. A.
et al.
Nature-based outdoor activities for mental and physical health: Systematic
review and meta-analysis. SSM Popul. Heal. 16, (2021).
11. Kim, J. & Park, M. A Meta-Analysis of the Association between Leisure Participation and Depressive
Symptoms among Older Adults. Ann. Appl. Sport Sci. 9, 0–0 (2021).
12. Catalan-Matamoros, D., Gomez-Conesa, A., Stubbs, B. & Vancampfort, D. Exercise improves
depressive symptoms in older adults: An umbrella review of systematic reviews and meta-analyses.
Psychiatry Res. 244, 202–209 (2016).
13. Bone, J. K.
et al.
Engagement in leisure activities and depression in older adults in the United States:
Longitudinal evidence from the Health and Retirement Study. Soc. Sci. Med. 294, 114703 (2022).
14. Noguchi, T.
et al.
Art and cultural activity engagement and depressive symptom onset among older
adults: A longitudinal study from the Japanese Gerontological Evaluation Study. Int. J. Geriatr.
Psychiatry 37, (2022).
15. Fancourt, D., Opher, S. & De Oliveira, C. Fixed-Effects Analyses of Time-Varying Associations between
Hobbies and Depression in a Longitudinal Cohort Study: Support for Social Prescribing? Psychother.
Psychosom. 89, 111–113 (2020).
1. Xiao, S.
et al.
Impact of Different Type and Frequency of Social Participation on Depressive
Symptoms Among Older Chinese Adults: Is There a Gender Difference? Front. Psychiatry 12, (2021).
17. Soga, M., Gaston, K. J. & Yamaura, Y. Gardening is benecial for health: A meta-analysis. Prev. Med.
Reports 5, 92–99 (2016).
1. Clifford, A. M.
et al.
The effect of dance on physical health and cognition in community dwelling
older adults: a systematic review and meta-analysis. Arts Health 15, 1–29 (2022).
19. McCrary, J. M., Altenmüller, E., Kretschmer, C. & Scholz, D. S. Association of Music Interventions With
Health-Related Quality of Life: A Systematic Review and Meta-analysis. JAMA Netw. open 5, (2022).
20. Daykin, N.
et al.
What works for wellbeing? A systematic review of wellbeing outcomes for music and
singing in adults. Perspect. Public Health 138, 39–46 (2018).
21. Balbag, M. A., Pedersen, N. L. & Gatz, M. Playing a musical instrument as a protective factor against
dementia and cognitive impairment: A population-based twin study.
Int. J. Alzheimers. Dis.
2014,
(2014).
22. Mak, H. W., Coulter, R. & Fancourt, D. Relationships between Volunteering, Neighbourhood Deprivation
and Mental Wellbeing across Four British Birth Cohorts: Evidence from 10 Years of the UK Household
Longitudinal Study.
Int. J. Environ. Res. Public Heal.
2022,
Vol.19, Page 1531
19, 1531 (2022).
23. Tomioka, K., Kurumatani, N. & Hosoi, H. Social Participation and Cognitive Decline Among
Community-dwelling Older Adults: A Community-based Longitudinal Study. Journals Gerontol. Ser. B
Page 16/22
73, 799–806 (2018).
24. Bone, J. K.
et al.
Associations between participation in community arts groups and aspects of
wellbeing in older adults in the United States: a propensity score matching analysis. Aging Ment.
Heal. (2022)
doi:10.1080/13607863.2022.2068129/SUPPL_FILE/CAMH_A_2068129_SM5404.DOCX.
25. Warran, K., Burton, A. & Fancourt, D. What are the active ingredients of ‘arts in health’ activities?
Development of the INgredients iN ArTs in hEalth (INNATE) Framework [version 2; peer review: 2
approved]. Wellcome Open Res. 7, 1–22 (2022).
2. Fancourt, D., Aughterson, H., Finn, S., Walker, E. & Steptoe, A. How leisure activities affect health: a
narrative review and multi-level theoretical framework of mechanisms of action. The Lancet
Psychiatry (2021) doi:10.1016/S2215-0366(20)30384-9.
27. Helliwell, J.
et al. World Happiness Report
. https://happiness-
report.s3.amazonaws.com/2022/WHR+22.pdf (2022).
2. Life expectancy at birth, total (years).
World Bank Group
https://data.worldbank.org/indicator/SP.DYN.LE00.IN.
29.
Gross domestic product 2021, World Development Indicators database
. (2022).
30. Gini index | Data.
The World Bank
https://data.worldbank.org/indicator/SI.POV.GINI (2022).
31. Boulton, M. Quality of Life. in
Key Concepts in Medical Sociology
(eds. Monaghan, L. & Gabe, J.)
180–186 (SAGE Publications Ltd, 2022). doi:10.2/JQUERY.MIN.JS.
32. Fluharty, M.
et al.
Difference in predictors and barriers to arts and cultural engagement with age in
the United States: A cross-sectional analysis using the Health and Retirement Study. PLoS One 16,
(2021).
33. Han, J., Zhang, X. & Meng, Y. The impact of old-age pensions on the happiness level of elderly people
– evidence from China. Ageing Soc. 42, 1079–1099 (2022).
34. Higgs, P. Ageing and the Lifecourse. in
Key Concepts in Medical Sociology
(eds. Monaghan, L. &
Gabe, J.) 77–83 (SAGE Publications Ltd, 2022). doi:10.2/JQUERY.MIN.JS.
35. Steptoe, A., Breeze, E., Banks, J. & Nazroo, J. Cohort prole: the English longitudinal study of ageing.
Int. J. Epidemiol. 42, 1640–1648 (2013).
3. Kondo, K., Rosenberg, M. & World Health Organization.
Advancing universal health coverage through
knowledge translation for healthy ageing: lessons learnt from the Japan gerontological evaluation
study
. https://apps.who.int/iris/handle/10665/279010 (2018).
37. Sonnega, A.
et al.
Cohort Prole: the Health and Retirement Study (HRS). Int. J. Epidemiol. 43, 576–
585 (2014).
3. Smith, J., Ryan, L., Fisher, G. G., Sonnega, A. & Weir, D.
Psychosocial and Lifestyle Questionnaire
2006–2016 Documentation Report Core Section LB
. (2017).
39. Zhao, Y., Hu, Y., Smith, J. P., Strauss, J. & Yang, G. Cohort Prole: The China Health and Retirement
Longitudinal Study (CHARLS). Int. J. Epidemiol. 43, 61–68 (2014).
Page 17/22
40. Higgins, J. P. T. & Thompson, S. G. Quantifying heterogeneity in a meta-analysis. Stat. Med. 21,
1539–1558 (2002).
Figures
Figure 1
Levels of hobby engagement amongst older adults aged 65 and above across 16 nations
Page 18/22
Figure 2
See image above for gure legend.
Page 19/22
Figure 3
See image above for gure legend.
Page 20/22
Figure 4
See image above for gure legend.
Page 21/22
Figure 5
See image above for gure legend.
Page 22/22
Figure 6
Hobbies engagement and country wealth, world happiness index, life expectancy, and Gini index
(Country abbreviations were used: AT=Austria, BE=Belgium, CN=China, CZ=Czech Republic, DK=Denmark,
EN=England, EE=Estonia, FR=France, DE=Germany, IT=Italy, JP=Japan, SI=Slovenia, ES=Spain,
SE=Sweden, CH=Switzerland, and US=United States.)
Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.
KMhobbyandmentalwellbeingSupplementaryv03.docx
