Access to this full-text is provided by Springer Nature.
Content available from Nature Medicine
This content is subject to copyright. Terms and conditions apply.
Nature Medicine | Voume 29 | September 2023 | 2233–2240 2233
nature medicine
https://doi.org/10.1038/s41591-023-02506-1
Article
Hobby engagement and mental wellbeing
among people aged 65 years and older in
16 countries
Hei Wan Mak 1, Taiji Noguchi 2,3, Jessica K. Bone 1, Jacques Wels 4,5,
Qian Gao 1, Katsunori Kondo6,7, Tami Saito2 & Daisy Fancourt 1
Growing aging populations pose a threat to global health because of the
social and psychological challenges they experience. 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
dierent national settings. We harmonized measures of hobby engagement
and multiple aspects of mental wellbeing across 16 nations represented
in ve longitudinal studies (N = 93,263). Prevalence of hobby engagement
varied substantially across countries, from 51.0% of Spanish respondents
to 96.0% of Danish respondents. Fixed eects models and multinational
meta-analyses were applied to compare the longitudinal associations
between hobbies and mental wellbeing. Independent of confounders,
having a hobby was associated with fewer depressive symptoms (pooled
coecient = −0.10; 95% condence intervals (CI) = −0.13, −0.07), and higher
levels of self-reported health ( po ol ed c oe c ient = 0.06; 95% CI = 0.03, 0.08),
happiness (pooled c o e c ient = 0.09; 95% CI = 0.06, 0.13) and life satisfaction
(pooled c oe cient = 0.10; 95% CI = 0.08, 0.12). Further analyses suggested
a temporal relationship. The strength of these associations, and prevalence
of hobby engagement, were correlated with macrolevel factors such as
life expectancy and national happiness levels but overall, little variance in
ndings was explained by country-level factors (<9%). Given the relative
universality of ndings, ensuring equality in hobby engagement within and
between countries should be a priority for promoting healthy aging.
Aging populations are an increasing global concern given the social
and psychological challenges they can experience, including loneli-
ness, social isolation and worsening mental health, all of which are
associated with increasing physical multimorbidity and mortality
1,2
.
Globally, the population aged 65 years and older (65+) is growing at a
faster rate than all other age groups3. According to data from the United
Nations, 1 in 11 people were aged 65+ in 2019, which is expected to rise
to 1 in 6 people by 2050 (ref. 3). Although advancements in healthcare
Received: 6 December 2022
Accepted: 18 July 2023
Published online: 11 September 2023
Check for updates
1Department of Behavioural Science and Health, Institute of Epidemiology & Health Care, University College London, London, UK. 2Department of Social
Science, Center for Gerontology and Social Science, Research Institute, National Center for Geriatrics and Gerontology, Obu, Japan. 3Japan Society for
the Promotion of Science, Tokyo, Japan. 4MRC Unit for Lifelong Health and Ageing, University College London, London, UK. 5Centre Metices, Université
libre Bruxelles, Brussels, Belgium. 6Department of Social Preventive Medical Sciences, Center for Preventive Medical Sciences, Chiba University, Chiba,
Japan. 7Department of Gerontological Evaluation, Center for Gerontology and Social Science, Research Institute, National Center for Geriatrics and
Gerontology, Obu, Japan. e-mail: d.fancourt@ucl.ac.uk
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine | Voume 29 | September 2023 | 2233–2240 2234
Article https://doi.org/10.1038/s41591-023-02506-1
Results
Participants
We undertook fixed effect analyses and multinational meta-analyses
of longitudinal data from the English Longitudinal Study of Ageing
(ELSA, Waves 7–9), Japan Gerontological Evaluation Study (JAGES,
Waves 2–4), US Health and Retirement Study (HRS, Waves 9–14), Survey
of Health, Ageing and Retirement in Europe (SHARE, Waves 4–6) and
China Health and Retirement Longitudinal Study (CHARLS, Waves 1–3).
ELSA, JAGES, HRS and CHARLS follow participants living in England,
Japan, the USA and China, respectively. SHARE follows participants liv-
ing in 28 European countries and Israel but, for this study, we focused
only on participants living in Austria, Belgium, Czech Republic, Den-
mark, Estonia, France, Germany, Italy, Slovenia, Spain, Sweden and
Switzerland, where data were available for the analysis. We followed
participants for three consecutive waves (4–8 years).
To allow for comparison across all datasets, we limited participants
to those aged 65+. To explore how changes in hobby engagement were
associated with changes in mental wellbeing over time, a total of 93,263
respondents who provided data across all study measures were ana-
lyzed: 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 USA (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 propor-
tion 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. More than 60% of the partici
-
pants experienced long-standing mental or physical health conditions
(Table 1). 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%). Italy (54.0%),
Spain (51.0%) and China (37.6%; albeit focusing exclusively on social
hobbies) had the lowest engagement levels (Fig. 1 and Table 1).
Longitudinal associations between hobby engagement and
mental wellbeing
Fixed effects models tested the longitudinal associations of how
changes in engagement in hobbies were associated with changes in
mental wellbeing, simultaneously accounting for all time-constant
factors (regardless of whether they were observed; for example, genet-
ics, past leisure behaviors, medical histories and psychological traits)
and identified time-varying factors (for example, sociodemographic
backgrounds, clinical conditions and difficulties with activities of
daily living). We then pooled our findings into novel multinational
meta-analyses.
Overall, hobby engagement was negatively associated with depres-
sive symptoms (pooled coefficient = −0.10; 95% CI = −0.13, −0.07;
I
2
= 69.5%; H
2
= 3.28; where I
2
is the percentage of variability in the effect
size that is caused by between-study heterogeneity, rather than by
sampling error, and the H2 statistic describes the ratio of the observed
variation and the expected variance due to sampling error), positively
associated with self-reported health (pooled coefficient = 0.06; 95%
CI = 0.03, 0.08; I2 = 48.1%; H2 = 1.93), positively associated with hap-
piness (pooled coefficient = 0.09; 95% CI = 0.06, 0.13; I2 = 67.0%;
H2 = 3.03), and positively associated with life satisfaction (pooled
coefficient = 0.10; 95% CI = 0.08, 0.12; I2 = 33.6%; H2 = 1.51) (Fig. 2).
Directionality
Although fixed effects regression showed the nature of the relation-
ship between hobby and mental wellbeing, the directionality of this
relationship required further investigation. So we ran ordinary least
squares (OLS) regressions estimating the associations between hob-
bies measured at Time 1 and the outcomes measured at Time 2, while
have helped people live longer, healthy life expectancy (the average
number of years that a person is expected to live with good health and
without any disability, physical or psychological illnesses or injuries)
is often not matched with the increase in life expectancy, and there is
a growing prevalence of long-term mental health conditions. This is
placing untenable burdens on global health and social care services,
providing financial and workforce planning predicaments. To help
meet older adults’ needs and to support the sustainability of health and
social care systems globally, it is important to explore cost-effective
strategies to enhance older adults’ mental health and wellbeing.
There is increasingly global interest in how engagement in psycho-
social activities could address these challenges4,5. Hobbies (defined as
activities that people engage in during their leisure time for pleasure,
such as the arts, crafts, reading, playing games, sports, gardening,
volunteering and participating in societies/clubs) involve imagina-
tion, novelty, creativity, sensory activation, self-expression, relaxation
and cognitive stimulation, all of which are positively related to mental
health and wellbeing via psychological, biological, social and behavio-
ral pathways5. Participation in hobby groups can additionally provide
social support and reduce loneliness and social isolation5. For this rea-
son, many countries including the UK6, Japan7 and the USA8 have been
promoting hobbies and leisure activities as part of their policies and
recommendations to support and improve mental health and wellbeing,
with a particular focus on increasing participation among older adults.
These policies are underpinned by a large body of research that has
shown how hobbies can enhance multidimensional aspects of mental
health and wellbeing, including negative symptomatology and clinical
diagnoses of depression and psychiatric conditions, experiential well-
being (for example, positive and negative affect), evaluative wellbeing
(for example, life satisfaction) and eudemonic wellbeing (for example,
purpose in life) for older adults. Meta-analyses of both observational
and interventional studies involving engagement in hobbies such as
nature-based activities and volunteering have shown protective asso-
ciations with depressive symptoms9–12. These findings are supported by
individual studies showing concurrent and longitudinal relationships
(3–12 years of follow-up) between other types of hobbies such as com-
munity groups, arts and social clubs, and a lower incidence and preva-
lence of depression in adults aged 50 years and older (50+) in the USA
13
,
Japan
14
, the UK
15
and China
16
. Similarly, meta-analyses of various types
of leisure activities, such as dancing, nature-based activities and gar-
dening, have reported benefits for positive aspects of wellbeing
10,17–20
.
Again, these findings are supported by individual studies focusing on
broader activities such as volunteering, arts, cultural engagement
and indoor gardening from Sweden21, the UK22, Japan23 and the USA24.
However, the literature to date is hampered by several limita-
tions. First, studies have focused on single countries at a time, so given
differences in definitions, outcome measures and methodological
approaches between studies, it is unclear whether there is consistency
in results across different cultural settings, and thus whether findings
from one country population could be applied to populations in other
countries. Second, many studies have focused on specific subcat-
egories of hobbies (for example, volunteering versus nature-based
activities versus arts participation versus cultural engagement), often
applying conflicting definitions. 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. Individual
meta-analyses focusing on specific hobby definitions thus present
only a fraction of the literature available on the topic and provide an
incomplete picture to policymakers.
This study was therefore designed to harmonize measures of
hobby engagement and mental wellbeing in adults aged 65+ across
16 nations represented in five longitudinal studies, and explore the
relationship with mental wellbeing, the direction of association and
the variation in findings by country.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine | Voume 29 | September 2023 | 2233–2240 2235
Article https://doi.org/10.1038/s41591-023-02506-1
controlling for identified confounders and baseline outcomes. Results
were then pooled into meta-analyses.
Hobby engagement was associated with subsequently fewer
depressive symptoms (pooled coefficient = −0.14; 95% CI = −0.19,
−0.09; I
2
= 70.7%; H
2
= 3.41) and greater self-reported health (pooled
coefficient = 0.09; 95% CI = 0.07, 0.12; I2 = 18.0%; H2 = 1.22), happiness
(pooled coefficient = 0.11; 95% CI = 0.08, 0.14; I2 = 17.7%; H2 = 1.21) and
life satisfaction (pooled coefficient = 0.10; 95% CI = 0.07, 0.13; I2 = 21.8%;
H2 = 1.28) (Extended Data Fig. 1).
We tested the consistency of these findings using a different sta-
tistical approach—lagged fixed effects models using an Arellano–Bond
estimator model—on the ELSA dataset where there were sufficient
repeated waves (nine available). Results confirmed that hobby engage-
ment was still associated with subsequent changes in depressive
symptoms (coefficient = −0.38; 95% CI = −0.63, −0.12), self-reported
health (coefficient = 0.73; 95% CI = 0.47, 0.99) and happiness (coeffi-
cient = 0.36; 95% CI = 0.01, 0.71), with marginal effects on life satisfac-
tion (coefficient = 0.19; 95% CI = −0.03, 0.41) (Supplementary Table 1).
Country-level factors
To ascertain how much of the variance in the relationship with men-
tal wellbeing was explained by country, we merged the datasets and
ran multilevel models. After adjusting for confounders, associations
between hobbies and the outcomes remained, and the country variance
explained <9% of the total variance (Extended Data Fig. 2).
We then explored which country-level factors might explain this
variance. Prevalence of hobby engagement was positively correlated
with the world happiness index score
27
(r = 0.63), country wealth meas-
ured by gross domestic product by capita
28
(r = 0.49) and life expec-
tancy
29
(r = 0.39), and was negatively correlated with the Gini index
measuring income inequality within a nation30 (r = −0.63) (Fig. 3).
These same country-level factors were also used in meta-
regressions as potential predictors of between-study heterogeneity in
outcomes. For the prevalence of hobby engagement (Extended Data
Fig. 3), country wealth (Extended Data Fig. 4) and Gini index (Extended
Data Fig. 5), no differences in effect sizes were found according to
these predictors. For the world happiness index score, no associations
were shown between effect sizes and index score, except marginally
for life satisfaction (Extended Data Fig. 6). With a confidence level of
90%, for every additional unit in the world happiness index score, the
effect size of a study rose by 0.05 (90% CI = −0.01, 0.11). For life expec-
tancy, a positive correlation was shown between life expectancy and
self-reported health effect sizes: for every year increase in life expec-
tancy across countries, the association between hobby engagement
and self-reported health was 0.01 points larger (coefficient = 0.01;
95% CI = 0.01, 0.02). No associations were found for other outcomes
(Extended Data Fig. 7).
Finally, we explored whether these country-level factors could
moderate the relationship between hobby engagement and mental
wellbeing. When interacting hobby engagement with country-level
factors in multilevel models, there was a small moderating effect
of Gini index, the world happiness index, country wealth and life
expectancy on the associations between hobbies and depression, life
satisfaction and self-reported health, but not on happiness (Extended
Data Fig. 2).
Sensitivity analyses
When using multiple imputation to account for missing data, results
were largely replicated (Supplementary Table 2). When including
respondents aged 55+ (except for Japan where all participants were
aged 65+), the evidence for longitudinal associations between hobby
engagement and the outcomes across countries became stronger,
Table 1 | Basic demographics by country in percentages or mean (s.d.)
Austria Belgium China Czech
Republic
Denmark England Estonia France Germany Italy Japan Slovenia Spain Sweden Switzerland USA
Hobby (%)
With
hobby
90.0 88.9 37.6 89.1 96.0 78.1 88.3 82.5 91.0 54.0 90.0 70.5 51.0 95.8 94.4 56.2
Without
hobby
9.1 11.1 62.4 10.9 4.0 21.9 11.7 17.5 9.0 46.0 10.0 29.5 49.0 4.20 5.60 43.8
n2,524 2,304 1,611 2,664 1,006 4,267 3,584 2,705 966 1,915 57,051 1,272 2,099 1,315 1,776 6,204
Gender (%)
Female 58.1 55.8 45.5 5 7.6 54.2 53.7 61.6 58.0 48.7 51.4 45.1 56.6 54.3 52.8 53.0 58.4
Male 41.9 44.2 54.5 42.4 45.8 46.3 38.4 42.0 51.3 48.6 54.9 43.4 4 5.7 47.2 47.0 41.6
n2,524 2,304 1,611 2,664 1,006 4,267 3,584 2,705 966 1,915 57,051 1,272 2,099 1,315 1,776 6,204
Age (s.d.) 74. 2
(6.53)
75.5
(7.25)
71.7
(4.98)
73.5
(6.52)
74.8
(7.42)
73.8
(6.65)
74.6
(6.16)
75.8
(7.17)
74.0
(6.33)
74.1
(6.46)
73.5
(5.60 )
74.7
(6.41)
75.9
(7.16)
74.7
(7.22)
74.2
(6.82)
72.6
(6.06)
Range 65–98 65–101 65–92 65–98.9 65–99 65–99 65–101 65–103 65–100 65–100 65–99 65–99 65–101 65–99 65–101 65–101
n2,524 2,304 1,611 2,664 1,006 4,267 3,584 2,705 966 1,915 57,051 1,272 2,099 1,315 1,776 6,204
Employment status (%)
Working 1.5 1.0 42.9 1.0 4.3 11.2 9.6 0.9 3.2 1.6 2 7.6 0.5 1.0 4.4 5.2 14.6
Not
working
13.5 16.3 0.9 0.2 3.2 4.27 1.2 7.1 6.4 23.1 6.5 10.5 3 7.6 0.2 8.6 7.0
Retired 85.0 82.7 56.2 98.8 92.5 84.5 89.2 92.0 90.4 75.3 66.0 89.0 61.4 95.4 86.2 78.3
n2,458 2,216 1,611 2,612 973 4,267 3,568 2,620 950 1,906 57,051 1,259 2,073 1,293 1,742 6,204
Long-standing mental/physical health conditions (%)
Yes 73.7 7 7.0 84.1 83.8 68.4 60.2 83.0 75.9 75.1 74.9 81.4 76.7 78.0 67.3 66.1 93.6
No 26.3 23.0 15.9 16.2 31.6 39.8 1 7.0 24.1 24.9 2 5.1 18.6 23.3 22.0 32.7 33.9 6.4
n2,504 2,295 1,611 2,643 1,004 4,267 3,576 2,665 966 1,912 57,051 1,260 2,091 1,313 1,770 6,204
Note: The table shows baseline demographics where baseline indicates the irst wave at which each participant completed the survey, and therefore does not relate to a single year of data.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine | Voume 29 | September 2023 | 2233–2240 2236
Article https://doi.org/10.1038/s41591-023-02506-1
likely because of the increase in the number of respondents (Supple-
mentary Table 3).
When analyses were stratified by gender, we found some variations
between countries. However, pooled effect sizes from meta-analysis
showed that engagement in hobbies remained beneficial for both
females (depressive symptoms: pooled coefficient = −0.10; 95%
CI = −0.15, −0.06; self-reported health: pooled coefficient = 0.06; 95%
CI = 0.03, 0.08; happiness: pooled coefficient = 0.08; 95% CI = 0.03,
0.12; life satisfaction: pooled coefficient = 0.10; 95% CI = 0.07, 0.14)
and males (depressive symptoms: pooled coefficient = −0.09; 95%
CI = −0.12, −0.06; self-reported health: pooled coefficient = 0.06; 95%
CI = 0.03, 0.09; happiness: pooled coefficient = 0.10; 95% CI = 0.07,
0.12; life satisfaction: pooled coefficient = 0.09; 95% CI = 0.06, 0.11)
(Extended Data Fig. 8 for female and Extended Data Fig. 9 for male).
The potential positive effects of hobby engagement remained
when only considering respondents who were retired (depres-
sive symptoms: pooled coefficient = −0.09; 95% CI = −0.12, −0.06;
self-reported health: pooled coefficient = 0.06; 95% CI = 0.03, 0.09;
happiness: pooled coefficient = 0.09; 95% CI = 0.05, 0.13; life satisfac-
tion: pooled coefficient = 0.09; 95% CI = 0.07, 0.11) (Extended Data
Fig. 10). Further, multilevel model analyses also showed no moderating
effects of national pension age (Extended Data Fig. 2f).
To assess whether the type of hobbies measured (binary measure
or index created from a list of options) was responsible for differences
in effect sizes between studies, a subgroup meta-analysis was con-
ducted. Pooled analyses showed no subgroup differences between the
measures; hobbies continued to associate with all outcomes (P > 0.05;
Supplementary Table 4). We also included a variable capturing the
type of hobby measure within the multilevel models of the merged
datasets, but results were unaffected, suggesting measurement bias did
not underlie the findings (Extended Data Fig. 2). Even when excluding
CHARLS data (which focused exclusively on social hobbies rather than
solitary ones) in our meta-analyses of analyses exploring directionality,
the results were consistent (Extended Data Fig. 1).
Discussion
This study compared longitudinal associations between hobby engage-
ment and multidimensional aspects of mental wellbeing across 16
countries. The prevalence of hobby engagement varied substantially
across countries, from countries where only one in two people had a
hobby (for example, 51.0% of the Spanish respondents) to countries
where hobby engagement was ubiquitous (for example, 96.0% of the
Danish respondents). Meta-analysis of the findings revealed that hav-
ing 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. Looking at the
direction of these associations, increased hobby engagement pre-
dicted subsequent decreases in depressive symptoms and increased
self-reported health, happiness and life satisfaction. There was little
variance in findings among countries, suggesting a relative universality
of response. However, on average, more adults aged 65+ had hobbies in
countries with higher world happiness index score and life expectancy,
and the relationship between hobby engagement and life satisfaction
and self-reported health was slightly stronger in such countries. Sensi-
tivity analysis showed that findings did not vary by gender or retirement
status, nor by country-level retirement age.
Our findings are in line with various cross-disciplinary interna-
tional literature indicating that having a hobby may enhance mental
wellbeing among adults aged 65+, but they present an advance on past
literature in several ways. First, the results provide evidence for the
consistency in such findings across cultural settings and countries,
highlighting the relevance to global public health policies and prac-
tices. Of the four outcomes, hobby engagement has the most consist-
ent association with life satisfaction; a subjective evaluation of one’s
social, emotional and physical wellbeing that can be independent of
‘objective’ health status or functional ability, which tend to decline with
age31. Hobbies could contribute to older adults’ life satisfaction through
many mechanisms, including feeling in control of their minds and bod-
ies, finding a purpose in life and feeling competent in tackling daily
issues
26
. Our temporal analyses showed that these associations were
not merely the result of good psychological health predicting hobby
engagement. In actuality, the relationship between hobbies and mental
wellbeing is likely bidirectional, because theoretical work applying
lenses from complex adaptive systems science to leisure engagement
and health has posited constant positive and negative feedback loops
between leisure behaviors and health outcomes
26
. But our directional-
ity findings are encouraging because they suggest that experimental
efforts to increase hobby engagement may have the potential to alter
United States Europe China
Hobby engagement (%) 37.60% 96.00%
England Japan
Fig. 1 | Levels of hobby engagement. Levels of hobby engagement among older adults aged 65 and above across 16 nations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine | Voume 29 | September 2023 | 2233–2240 2237
Article https://doi.org/10.1038/s41591-023-02506-1
subsequent mental wellbeing. Indeed, the association of hobbies with
life satisfaction is particularly promising given that it was seen not
only in healthier respondents, but also in respondents such as those in
the USA where a very high proportion of the respondents were living
with long-standing mental or physical health conditions and where
psychosocial interventions could be even more relevant.
However, it is also relevant to consider why, even in the context of
notable meta-analytic findings, associations between hobby engage-
ment and mental wellbeing showed some variation across countries.
The results from meta-regression analyses showed 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 for self-reported health were also generally
larger for countries with higher life expectancy. Individuals living
in countries with higher life expectancies or happiness levels may
be more likely to have a hobby (for example, Denmark, Sweden and
Switzerland; Fig. 3c,d), which may have inflated the size of the coef-
ficients. However, there are some exceptions. For example, although
people living in Spain had a lower engagement rate comparatively,
the strength of the association between hobby engagement and life
satisfaction was similar to that for countries with much higher engage-
ment rates (including Austria, Czech Republic and Switzerland). This
suggests the health benefits of hobbies (at least for life satisfaction) are
not simply driven by the high prevalence of engagement rates, but can
also be found in countries where hobbies are less popular. Similarly,
Austria
Belgium
China
Czech Republic
Denmark
England
Estonia
France
Germany
Italy
Japan
Slovenia
Spain
Sweden
Switzerland
USA
Overall
Heterogeneity: τ2 = 0.00, I2 = 48.09%, H2 = 1.93
Test of θi = θj: Q(15) = 35.14, P = 0.00
Test of θ = 0: z = 4.78, P = 0.00
Study
Negative eect Positive eect
–0.4 –0.2 0 0.2 0.4
Eect size
with 95% CI
0.15 [0.05, 0.25]
0.07 [–0.01, 0.15]
0.00 [–0.10, 0.11]
0.05 [–0.03, 0.13]
–0.14 [–0.35, 0.07]
0.02 [–0.03, 0.07]
0.01 [–0.04, 0.06]
0.06 [–0.01, 0.13]
0.01 [–0.10, 0.12]
0.03 [–0.03, 0.09]
0.10 [0.07, 0.13]
0.06 [–0.02, 0.14]
0.08 [0.03, 0.13]
0.22 [0.07, 0.37]
0.20 [0.08, 0.32]
0.03 [–0.00, 0.06]
0.06 [0.03, 0.08]
3.81
5.58
3.81
5.58
1.17
9.16
9.16
6.56
3.55
7.75
12.44
5.58
9.16
2.01
3.09
11.60
Weight
(%)
Random-eects REML model
b
Austria
Belgium
China
Czech Republic
Denmark
England
Estonia
France
Germany
Italy
Japan
Slovenia
Spain
Sweden
Switzerland
USA
Overall
Heterogeneity: τ2 = 0.00, I2 = 69.49%, H2 = 3.28
Test of θi = θj: Q(15) = 41.55, P = 0.00
Test of θ = 0: z = –5.73, P = 0.00
Study
Negative eect Positive eect
–0.4 –0.2 0 0.2
Eect size
with 95% CI
–0.24 [–0.35, –0.13]
–0.06 [–0.16, 0.04]
–0.00 [–0.10, 0.09]
–0.07 [–0.16, 0.02]
–0.16 [–0.34, 0.02]
–0.03 [–0.08, 0.02]
–0.14 [–0.22, –0.06]
–0.23 [–0.31, –0.15]
–0.04 [–0.16, 0.08]
–0.06 [–0.13, 0.01]
–0.13 [–0.15, –0.11]
–0.08 [–0.17, 0.01]
–0.12 [–0.19, –0.05]
–0.20 [–0.34, –0.06]
0.02 [–0.10, 0.14]
–0.09 [–0.13, –0.05]
–0.10 [–0.13, –0.07]
5.05
5.57
5.84
5.84
2.67
8.40
6.74
6.43
4.59
7.39
10.18
6.13
7.39
3.80
4.59
9.37
Weight
(%)
Random-eects REML model
a
Austria
Belgium
China
Czech Republic
Denmark
England
Estonia
France
Germany
Italy
Japan
Slovenia
Spain
Sweden
Switzerland
USA
Overall
Heterogeneity: τ2 = 0.00, I2 = 66.96%, H2 = 3.03
Test of θi = θj: Q(15) = 34.31, P = 0.00
Test of θ = 0: z = 4.94, P = 0.00
Study
Negative eect Positive eect
–0.2 0 0.2 0.4 –0.2 0 0.2 0.4
Eect size
with 95% CI
0.16 [0.05, 0.27]
0.03 [–0.09, 0.15]
0.21 [0.10, 0.32]
0.05 [–0.07, 0.17]
0.04 [–0.12, 0.20]
0.10 [0.04, 0.16]
0.07 [–0.03, 0.17]
0.03 [–0.06, 0.12]
0.18 [0.03, 0.33]
0.21 [0.14, 0.28]
0.10 [0.08, 0.12]
0.01 [–0.10, 0.12]
–0.02 [–0.10, 0.06]
0.21 [0.04, 0.38]
–0.02 [–0.16, 0.12]
0.11 [0.07, 0.15]
0.09 [0.06, 0.13]
5.42
4.93
5.42
5.17
3.60
8.93
6.24
6.54
3.93
7.86
11.17
5.68
7.51
3.17
4.11
10.32
Weight
(%)
Random-eects REML model
c
Austria
Belgium
China
Czech Republic
Denmark
England
Estonia
France
Germany
Italy
Japan
Slovenia
Spain
Sweden
Switzerland
USA
Overall
Heterogeneity: τ2 = 0.00, I2 = 33.62%, H2 = 1.51
Test of θi = θj: Q(15) = 26.20, P = 0.04
Test of θ = 0: z = 8.25, P = 0.00
Study
Negative eect Positive eect
Eect size
with 95% CI
0.18 [0.07, 0.29]
0.05 [–0.02, 0.12]
–0.04 [–0.16, 0.08]
0.17 [0.06, 0.28]
0.26 [0.07, 0.45]
0.07 [0.02, 0.12]
0.14 [0.04, 0.24]
0.11 [0.03, 0.19]
–0.03 [–0.16, 0.10]
0.07 [–0.00, 0.14]
0.09 [0.07, 0.11]
0.07 [–0.03, 0.17]
0.17 [0.11, 0.23]
0.12 [–0.02, 0.26]
0.18 [0.07, 0.29]
0.10 [0.06, 0.14]
0.10 [0.08, 0.12]
3.56
6.92
3.32
3.84
1.38
10.20
4.15
5.77
2.70
7.32
18.16
4.15
8.37
2.53
3.84
13.79
Weight
(%)
Random-eects REML model
d
Fig. 2 | Meta-analysis of the findings from fixed effects models (n study = 16).
Data were first analyzed separately for each country using fixed effects
regression. The findings were then pooled into multinational meta-analyses
using the random effects model to estimate the overall effect sizes for all
outcomes. Between-study heterogeneity was estimated using the algorithm of
the restricted maximum likelihood and was assessed using I2 and H2 statistics.
I2 is the percentage of variability in the effect size that is caused by between-study
heterogeneity, rather than by sampling error. The H2 statistic describes the ratio
of the observed variation and the expected variance due to sampling error. Given
that some of the analyses had more participants than others and thus had lower
sampling variability and more precise estimates, the meta-analysis was weighted.
Studies with a greater number of respondents were given more weight than
studies with a small number of respondents. These were relative weights that
summed to 100. Data are presented as fixed effects coefficients and 95% CI. The
overall effect size and its width should have accounted for the between-study
variance, the number of studies, the precision of the study-specific estimates
(or ‘effect sizes’) and the significance level. a, Hobbies and depressive symptoms.
b, Hobbies and self-reported health. c, Hobbies and happiness. d, Hobbies and
life satisfaction.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine | Voume 29 | September 2023 | 2233–2240 2238
Article https://doi.org/10.1038/s41591-023-02506-1
the longitudinal associations between hobbies and the outcomes
continued to be found in countries with lower world happiness index
scores (for example, Japan and China) as well as countries with lower
life expectancy (for example, the USA). In countries with higher hap-
piness levels and life expectancy, there may be fewer psychological
barriers to hobby engagement in the first place and stronger positive
feedback loops supporting the translation of this engagement into
causal mechanisms that both support mental wellbeing and addition-
ally contribute to the maintenance of and even increase in the original
salutogenic hobby behavior. However, less than 9% of the variance in
findings was explained by country. So, taken together, our results sug-
gest that having a hobby may have the potential to be associated with
improvements in health among the older population cross-culturally.
This supports theories from anthropology, evolutionary psychology
and sociology that have focused on the potential adaptive benefits of
hobby engagement, seeing pleasure as a by-product but other func-
tions such as developing attention and cognition, social bonding and
societal cohesion, communication and knowledge, and adaptation as
important to species survival32–34.
The differential participation rates in hobbies across countries
must be cautiously interpreted given that questions about hobbies
varied in style and in length of time asked about within the datasets.
In particular, China’s lower hobby rate may be partly influenced by the
questions focusing largely on social hobbies (hence our additional sen-
sitivity analyses excluding China’s data from meta-analysis). Nonethe-
less, even among countries with identical questions on hobbies (such
as the 12 countries within the SHARE dataset), there was substantial
variation in participation rates. This may be a result of greater barriers
to engagement in some countries. 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 influence the likelihood of engagement
among adults aged 50+ (ref. 35). Between-country comparisons
found greater hobby engagement rates in more affluent countries.
Differences in hobby participation are thus concerning, because they
could contribute to or exacerbate health inequalities both within and
between countries. As a result, in working to capitalize on the findings
presented here, a systematic approach should be taken, considering
both how to address individual-level barriers to hobby engagement
that adults aged 65+ 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 wellbeing
outcomes. Public health strategies such as social prescribing schemes
including in the UK, USA, Japan and parts 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 influencing motivations
and propensity to engage among older populations, and in turn pro-
viding opportunities to strengthen the associations between hobby
and health outcomes.
Our findings have policy and health implications for adults aged
65+, especially those who are retired (between 56.2% and 98.8% of our
respondents). Contemporary life-course research has demonstrated
that the concept of aging has shifted from seniority to an emphasis
on lifestyle and consumption including expenditures for services and
healthy goods
36
. This aligns with the idea of ‘the third age’ emphasized
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 dependency
36
. As suggested in our
findings, hobbies such as physical activity, arts and cultural engage-
ment, and social and community participation have the potential to
lengthen ‘the third age’ period and make it one of ‘productive aging’
AT
BE
CN
CZ
DK
EN
EE
FR
DE
IT
JP
SI
ES SE
CH
USA
76
78
80
82
84
Life expectancy (years)
d
AT
BE
CN
CZ
DK
EN
EE
FR
DE
IT
JP
SI
ES
SE
CH
USA
5.5
6.0
6.5
7.0
7.5
8.0
World happiness index score
c
AT
BE
CN
CZ
DK
EN
EE
FR DE
IT
JP
SI
ES
SE
CH
USA
20
40
60
Gini index (%)
b
AT
BE
CN
CZ
DK
EN
EE
FR
DE
IT JP
SI
ES
SE
CH
USA
0
30,000
60,000
90,000
Country wealth (GDP per capita)
40 60 80 100
Hobbies engagement (%)
40 60 80 100
Hobbies engagement (%)
40 60 80 100
Hobbies engagement (%)
40 60 80 100
Hobbies engagement (%)
a
Fig. 3 | Correlations between hobby engagement rate and country-level
factors (n study = 16). a, Hobbies and country wealth. b, Hobbies and Gini
index. c, Hobbies and world happiness index. d, Hobbies and life expectancy.
Data are presented as mean values. 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; USA,
United States of America.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine | Voume 29 | September 2023 | 2233–2240 2239
Article https://doi.org/10.1038/s41591-023-02506-1
through protecting against age-related declines in mental health and
enhancing wellbeing, which have profound consequences for morbid-
ity and mortality.
There are many strengths in this study including the use of five
national longitudinal studies containing data from 16 nations. The
study also uses population surveys to compare hobby engagement
rates internationally, as well as assessing the strengths of the asso-
ciations against population statistics relating to country wealth, Gini
index, world happiness index score and life expectancy. In addition,
fixed effects analyses allowed us to explore how changes in hobbies
were associated with changes in the mental wellbeing outcomes, while
adjusting for all time-constant variables (regardless of whether they
were observed) and important time-varying variables.
However, the study is not without limitations. Because of the
use of observational data, causality cannot be established even with
sophisticated longitudinal data analysis modeling. Further, although
there was overall relative homogeneity in the way questions about
hobbies were asked and sub-questions were collapsed into a binary
indicator, some countries chose to list hobby examples, whereas oth-
ers did not, which may have led to differences in interpretation of the
question by the respondents. However, no differences in associations
between hobbies and outcomes were found with different measures
(as shown in Supplementary Table 4) and the use of only 5 studies for
16 countries limited the amount of heterogeneity in the measures.
Relatedly, the reference period measuring hobby engagement rate
varied across the longitudinal datasets, although we still found some
engagement variations between countries with the same reference
period measure.
Future research is needed to consider the types, frequency and
length of hobby engagement in different countries, as well as whether
modulation of specific types of hobby engagement (such as the pres-
ence or absence of physical activity or social interaction) differentially
affect outcomes25. It will also be necessary to examine further whether
key benefits of hobby engagement are derived from the activities
themselves or additionally from time spent on hobbies displacing time
that otherwise could be spent on less salutogenic activities includ-
ing chores, work or procrastination. In addition, our analysis did not
explore other intraindividual factors that are largely time-constant
but may have some limited variability. Future studies may wish to use
datasets with more interview waves that might capture this variability
over time to explore the role of such factors as moderators of effects.
Finally, natural experiments such as changes in leisure or retirement
policies or behaviors (for example, as the result of major financial
upheavals within countries) are encouraged to explore potential causal
effects of hobbies on mental wellbeing in more detail.
The cross-national mental wellbeing benefits of hobby engage-
ment reported here suggest that facilitating greater opportunities
for engagement across demographic groups and between countries
should be a priority in efforts to increase healthy life expectancy and
relieve the increased burden of aging populations on healthcare
systems internationally. Results from this study could also be used
as evidence when formulating and developing schemes to increase
equity of access to leisure activities among older adults across demo-
graphic groups and between countries, as well as in integrating psy-
chosocial interventions into health services or public health strategies
(for example, through social prescribing schemes) to reduce morbid-
ity, mortality and healthcare burden, and enhance aging experiences
among older adults.
Online content
Any methods, additional references, Nature Portfolio reporting sum-
maries, source data, extended data, supplementary information,
acknowledgements, peer review information; details of author contri-
butions and competing interests; and statements of data and code avail-
ability are available at https://doi.org/10.1038/s41591-023-02506-1.
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. The Lancet Public Health. Ageing: a 21st century public health
challenge? Lancet Public Heal. 2, e297 (2017).
3. Global issues – Ageing. United Nations https://www.un.org/en/
global-issues/ageing (accessed 15 March 2023).
4. Steptoe, A. & Fancourt, D. Protecting physical health in people
with mental illness. Lancet Psychiatry 6, 889–890 (2019).
5. 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 67 (WHO Regional Oice for
Europe, 2019).
6. Older people: Independence and mental wellbeing. NICE
https://www.nice.org.uk/guidance/ng32 (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 (accessed 19 October 2022).
8. 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,
44–72 (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, 100934 (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. http://aassjournal.com/article-1-
960-en.html (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 https://doi.org/10.1002/gps.5685 (2022).
15. Fancourt, D., Opher, S. & De Oliveira, C. Fixed-eects analyses
of time-varying associations between hobbies and depression
in a longitudinal cohort study: support for social prescribing?
Psychother. Psychosom. 89, 111–113 (2020).
16. Xiao, S. et al. Impact of dierent type and frequency of social
participation on depressive symptoms among older Chinese
adults: is there a gender dierence? Front. Psychiatry 12,
758105 (2021).
17. Soga, M., Gaston, K. J. & Yamaura, Y. Gardening is beneicial for
health: a meta-analysis. Prev. Med. Rep. 5, 92–99 (2016).
18. Cliord, A. M. et al. The eect of dance on physical health and
cognition in community dwelling older adults: a systematic
review and meta-analysis. Arts Health 15, 200–228 (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,
e223236 (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).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine | Voume 29 | September 2023 | 2233–2240 2240
Article https://doi.org/10.1038/s41591-023-02506-1
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, 836748 (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. 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. J. Gerontol. Ser. B 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. Health
27, 1163–1172 (2023).
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). Wellcome Open Res. 7,
10 (2022).
26. Fancourt, D., Aughterson, H., Finn, S., Walker, E. & Steptoe, A. How
leisure activities aect health: a narrative review and multi-level
theoretical framework of mechanisms of action. Lancet Psychiatry
8, 329–339 (2021).
27. Helliwell, J. et al. World Happiness Report (Sustainable
Development Solutions Network, 2022).
28. DataBank – world development indicators. The World Bank
https://databank.worldbank.org/reports.aspx?source=2&series=
NY.GDP.PCAP.CD&country= (accessed 15 March 2023).
29. Life expectancy at birth, total (years). The World Bank Group
https://data.worldbank.org/indicator/SP.DYN.LE00.IN (2022)
(accessed 6 October 2022).
30. Gini index | Data. The World Bank https://data.worldbank.org/
indicator/SI.POV.GINI (accessed 3 November 2022).
31. Boulton, M. in Key Concepts in Medical Sociology (eds Monaghan,
L. & Gabe, J.) 180–186 (SAGE, 2022).
32. Csikszentmihalyi, M. & Nakamura, J. in Eortless Attention
(ed. Buuya, B.) 179–189 (MIT Press, 2013).
33. Dissanayake, E. Homo aestheticus: Where Art Comes from and Why
(Univ. Washington Press, 1995).
34. Carroll, J. The adaptive function of literature and the other arts.
On the Human. A Project of the National Humanities Center (2009);
https://nationalhumanitiescenter.org/on-the-human/2009/06/
the-adaptive-function-of-literature-and-the-other-arts/
35. Fluharty, M. et al. Dierence 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, e0261532 (2021).
36. Higgs, P. in Key Concepts in Medical Sociology (eds Monaghan, L.
& Gabe, J.) 77–83 (SAGE, 2022).
Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional ailiations.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format,
as long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons license, and indicate
if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not
included in the article’s Creative Commons license and your intended
use is not permitted by statutory regulation or exceeds the permitted
use, you will need to obtain permission directly from the copyright
holder. To view a copy of this license, visit http://creativecommons.
org/licenses/by/4.0/.
© The Author(s) 2023
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Methods
Data
ELSA. ELSA started in 2002–2023 and follows over 11,000 participants
aged 50+ living in England every 2 years
37
. In this study, to be in line
with the other datasets, we extracted a pool of respondents aged 65+
who responded in Waves 7 (2014–2015; response rate = 78.3%–81.4%),
8 (2016–2017; response rate = 82.4%) and 9 (2018–2019; response
rate = 79.5%) where hobby engagement and outcome variables were
measured. We considered only respondents who provided data across
all measures. This resulted in 10,876 observations from 4,267 partici-
pants (2.5 per person, ranging from 2 to 3).
JAGES. JAGES is a large-scale population-based longitudinal
study about aging established in 2010, mainly collected through
self-administered mail surveys, targeting older people aged 65+ who
do not receive long-term care insurance benefits38. JAGES has con-
ducted a joint survey with municipalities that are the public insurers
of long-term care insurance every 3–4 years: Wave 1 (2010–2011) to
Wave 4 (2019–2020). This study used data from Waves 2, 3 and 4 (30–64
municipalities; response rate = 52.4%–71.1%). Of the respondents, those
with complete data on hobby engagement and health outcomes in at
least two waves were considered. This resulted in 125,901 observations
from 57,051 participants (2.4 person, ranging from 2 to 3).
HRS. HRS is a national cohort study of more than 37,000 individuals
over the age of 50 in the USA39. 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 inter-
viewed for the first time in 1992 and followed up every 2 years, with
other studies and younger cohorts merged with the initial pool of
respondents. Together, these studies create a group of fully repre-
sentative respondents aged over 50 in the USA. Further details on
study design are reported elsewhere39. We used data from HRS Waves
9–14 at which participation in a hobby was measured (2008–2018).
At each wave, a rotating random 50% subgroup of respondents was
invited to an enhanced interview and given a Leave Behind Psycho-
social and Lifestyle Questionnaire to complete and return by mail,
which included questions on participation in community arts groups
and mental wellbeing
40
. Participants were eligible to complete this
psychosocial questionnaire every 4 years. Response rates in each year
varied from 62% to 85%. We restricted the respondents to those aged
65+, with complete data on hobby engagement and mental wellbeing
outcomes in at least two waves and no missing data on time-varying
covariates. This resulted in 14,989 observations from 6,204 participants
(2.4 observations per person, range 2–3).
SHARE. SHARE is the largest pan-European social science panel study
providing internationally comparable longitudinal micro data on the
population aged 50+ and currently includes eight waves with data
collection starting in 2004. SHARE contains both the participation of
respondents in their baseline and refreshment interview to account for
a reduction in the number of respondents 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 less information, justifying the use of Waves 4, 5
and 6 in this study. Data information for these three waves is available
for twelve countries. Data were not available over these three waves for
Croatia, Greece, Hungary, Israel, Luxembourg, the Netherlands, Poland
and Portugal. The analytical pool of respondents by country, including
nonresponse 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.
CHARLS. CHARLS is a national cohort study of Chinese residents aged
45+ (ref. 41). The baseline survey started in 2011 and has been followed
up every 2 years (in 2013 and 2015). Multistage probability sampling
was used for a selection of respondents. 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 con-
sidered only participants who responded to all measures, resulting
in 3,440 observations from 1,611 participants (2.1 observations per
person, range 2–3).
Measures
Hobby and mental wellbeing. Our measures of hobby and mental
wellbeing are shown in Supplementary Table 5, which presents the
exact question-wording and item responses across datasets. Hobbies
and mental wellbeing outcomes were time-varying variables. The analy-
sis will focus on four types of mental wellbeing: depressive symptoms,
self-reported health, happiness and life satisfaction. The measure
items and response categories vary somewhat by country, reflecting
cultural differences across the 16 nations. Therefore, to ensure the
data were comparable, we harmonized and recoded all variables, and
standardized 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 wellbeing outcomes across countries.
Time-varying covariates. Nine time-varying variables that might
confound observed associations between hobby and mental wellbeing
were identified for the analysis. These included demographic charac-
teristics: age (a continuous variable), partnership status (living with a
partner/spouse versus not living with a partner/spouse), number of
people living in the household (a continuous variable); socioeconomic
position: employment status (working versus not working), household
income (a continuous variable), housing tenure (homeowner versus
not a homeowner); and health profiles: long-standing mental/physical
conditions (yes versus no), difficulties with daily activities (ADL) (with
difficulties versus without difficulties) and difficulties with instrumen-
tal activities of daily living (IADL; a continuous variable).
Statistical analysis
In the first instance, data were analyzed separately for each country
using fixed effects regression. Fixed effects regression is a longitudi-
nal data method that tests within-individual variation, meaning that
each individual is compared with themselves over time. Such a model
automatically controls for all time-invariant variables such as age,
gender, genetics, personality, socioeconomic status, education, area
of dwelling, past life experiences, past mental health and medical his-
tory, even if they are unobserved, as well as controlling for identified
time-varying covariates. For this reason, fixed effects regression is
considered to be more robust than traditional regression models in
exploring how changes in the predictor are associated with the changes
in the outcomes.
We pooled our findings into multinational meta-analyses using
the random effects 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 error
42
. A
value of I
2
> 50% indicates heterogeneity
42
. Similarly, the H
2
statistic
describes the ratio of the observed variation and the expected variance
caused by sampling error42. A value of H2 > 1 indicates the presence of
between-study heterogeneity
42
. Given that some of the analyses had
more participants than others and thus had lower sampling variability
and more precise estimates, the meta-analysis was weighted. Stud-
ies with a greater number of respondents were given more weight
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
than studies with a small number of respondents. These were relative
weights that summed to 100. The overall effect size and its width should
have accounted for the between-study variance, the number of studies,
the precision of the study-specific estimates (or ‘effect sizes’) and the
significance level. We also conducted a subgroup analysis by hobbies
measures to explore whether the differences in effect sizes might have
been attributed to the way in which questions on hobbies were asked in
each longitudinal study (a binary measure versus index created from
a list options). We further ran meta-regressions to explore the hetero-
geneity variance in our meta-analysis using five country-level factors:
the prevalence of hobby engagement, country wealth measured by
gross domestic product per capita
28
, world happiness index score
27
,
life expectancy29 and the Gini index measuring income inequality
within a nation30.
Although fixed effects analyses explored the longitudinal associa-
tions between changes in hobby engagement and changes in mental
wellbeing outcomes, they cannot test the direction of these relation-
ships. We therefore performed two further sets of analyses. First, we ran
OLS regressions estimating the association between hobbies measured
at Time 1 and outcomes measured at Time 2, while controlling for
baseline outcomes and covariates. The covariates included age, gen-
der, the number of people living in the household, partnership status,
household income, housing tenure, employment status, educational
level, long-term mental/physical conditions, ADLs and IADLs. Results
from these analyses were then pooled into meta-analysis as described
above. Respondents from China were dropped because of the inclusion
of only social hobbies. Second, we tested the consistency of these find-
ings using a different statistical approach—lagged fixed effects models
using an Arellano–Bond estimator model—on the ELSA dataset where
there were sufficient repeated waves (nine available). The Arellano–
Bond estimator is considered an extension of the fixed effects model,
which uses a first-difference model and includes lags of the outcome
variable as instruments for the first difference43. This model takes
account of previous changes in mental wellbeing outcomes over time
to estimate the effect of hobby on subsequent changes in the outcomes,
while accounting for differences in individual characteristics. However,
the model requires multiple waves of data and consistency in measures
across every wave. Not all of the datasets in our analyses met these
requirements, so we performed the Arellano–Bond estimator analyses
solely on ELSA, which is one of the earliest aging longitudinal studies
with a longer follow-up period and more consistent measures than
many of the other datasets. By applying the Arellano–Bond estimator
to this dataset, we were able to ascertain whether the findings matched
those from the OLS regressions and confirm that any findings from
the OLS regressions were not merely the result of a less sophisticated
statistical approach. In the analysis, all models were fully adjusted
without any age restriction to allow sufficient statistical power for use
of the Arellano–Bond estimator.
For both fixed effects and regression analyses, listwise deletion
was applied to handle missing data. The proportion of missingness
was as follows: Austria (20.6%), Belgium (39.6%), China (69.7%; mainly
because of missingness in household wealth, hobby and life satis-
faction), 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 USA (15.7%). In our main analysis, we present coeffi-
cients and 95% CIs to show the relationship between hobby engagement
and the outcomes across countries after adjustment for time-varying
covariates. Stata v.17 was used for the analyses.
To explore whether country-level factors could moderate the
relationship between hobby engagement and mental wellbeing, we
pooled data from four longitudinal datasets and undertook multilevel
analyses (JAGES was not available because of data restriction). The
models were adjusted for interview waves, type of hobby measure,
age, gender, the number of people living in the household, partnership
status, household income, housing tenure, employment status, edu-
cational level, long-term mental/physical conditions, ADLs and IADLs.
Finally, we performed a set of sensitivity analyses to explore the
robustness of the associations between hobby engagement and mental
wellbeing:
(1) 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 wellbeing
outcomes and time-varying covariates across all included waves.
(2) The main analysis considered only respondents aged 65+ to
allow for comparison across all datasets, but this signicantly
restricted the number of respondents 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 pool of respondents
to those who were aged 55+.
(3) To test for the consistency of the association between hobby
engagement and mental wellbeing across dierent population
groups, we stratied our respondents by gender (female and
male) and restricted our respondents to those who were retired.
Ethics and inclusion statement
This research analyzed five large and longitudinal datasets across
England, the USA, Europe, Japan and China, and collaborated with
local researchers throughout the research process to ensure its local
relevance. H.W.M. and D.F. are from the UK, J.K.B. is also from the UK
but her work has largely focused on the US context; J.W. is based in Bel-
gium; T.N., K.K. and T.S. are from Japan; and Q.G. is based in the UK and
originally from China. Roles and responsibilities were agreed among
authors ahead of the research.
This research is locally relevant to all studied countries given
that it shows individual findings by country, while aggregating them
to provide more conclusive evidence on the psychological benefits of
hobby engagement for older adults. These findings can provide local
decision-makers with data that could support the drafting of recom-
mendations on supporting healthy aging though encouraging hobby
uptake. The research result does not result in stigmatization, incrimi-
nation, discrimination or otherwise personal risk to participants. The
research did not involve any health, safety, security or other risk to
researchers. No biological materials, cultural artifacts or associated
traditional knowledge were transferred out of any country. The authors
have undertaken research relevant to the study.
Ethics approval
ELSA. ELSA Wave 9 received ethical approval from the South Central—
Berkshire Research Ethics Committee on 10 May 2018 (17/SC/0588).
ELSA Wave 8 received ethical approval from the South Central—Berk-
shire Research Ethics Committee on 23 September 2015 (15/SC/0526).
ELSA Wave 7 received ethical approval from the National Research Eth-
ics Service (NRES) Committee South Central—Berkshire on 28 Novem-
ber 2013 (13/SC/0532). ELSA Wave 6 received ethical approval from the
NRES Committee South Central—Berkshire on 28 November 2012 (11/
SC/0374). ELSA Wave 5 received ethical approval from the Berkshire
Research Ethics Committee on 21 December 2009 (09/H0505/124).
ELSA Wave 4 received ethical approval from the National Hospital for
Neurology and Neurosurgery and Institute of Neurology Joint Research
Ethics Committee on 12 October 2007 (07/H0716/48). ELSA Wave 3
received ethical approval from the London Multi-Centre Research
Ethics Committee on 27 October 2005 (05/MRE02/63). ELSA Wave
2 received ethical approval from the London Multi-Centre Research
Ethics Committee on 12 August 2004 (MREC/04/2/006). ELSA Wave
1 received ethical approval from the London Multi-Centre Research
Ethics Committee on 7 February 2002 (MREC/01/2/91). All participants
provided informed written consent.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
JAGES. JAGES received ethical approval from Nihon Fukushi University
(no. 10-05), Chiba University (no. 2493) and the National Center for
Geriatrics and Gerontology (no. 992), and all participants provided
informed written consent.
HRS. Ethical approval for HRS was obtained from the University of
Michigan Institutional Review Board. All participants gave informed
written consent.
SHARE. SHARE received ethical approval from the Ethics Council of
the Max Planck Society and all participants provided informed writ-
ten consent.
CHARLS. CHARLS received ethical approval from the Biomedical Ethics
Review Committee of Peking University (IRB00001052-11015) and all
participants provided informed written consent.
Reporting summary
Further information on research design is available in the Nature Port-
folio Reporting Summary linked to this article.
Data availability
The English Longitudinal Study of Ageing (ELSA) can be accessed via
the UK Data Service: https://beta.ukdataservice.ac.uk/datacatalogue/
series/series?id=200011. The Health and Retirement Study (HRS) 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 (SHARE) can be accessed via the SHARE Research
Data Center: http://www.share-project.org/data-access.html. The China
Health and Retirement Longitudinal Study (CHARLS) can be accessed via
the National School of Development, Peking University: https://charls.
charlsdata.com/pages/data/111/en.html. Restrictions to access data of
Japan Gerontological Evaluation Study (JAGES) applied. For researchers
who wish to use the data, please contact the JAGES Data Administration
Office at dataadmin.ml@jages.net. Non-JAGES research members may
be required to include JAGES members in their project or co-authors
in research papers depending on the study topic or data used.
Code availability
All code used for these analyses is publicly available online:
https://osf.io/84xzu/.
References
37. Steptoe, A., Breeze, E., Banks, J. & Nazroo, J. Cohort proile:
the English longitudinal study of ageing. Int. J. Epidemiol. 42,
1640–1648 (2013).
38. Kondo, K. & Rosenberg, M. (eds) Advancing Universal Health
Coverage through Knowledge Translation for Healthy Ageing:
Lessons Learnt from the Japan Gerontological Evaluation Study
(World Health Organization, 2018).
39. Sonnega, A. et al. Cohort proile: The Health and Retirement
Study (HRS). Int. J. Epidemiol. 43, 576–585 (2014).
40. Smith, J., Ryan, L., Fisher, G. G., Sonnega, A. & Weir, D. HRS
Psychosocial and Lifestyle Questionnaire 2006–2016 (Survey
Research Center, Institute for Social Research, Univ. Michigan, 2017).
41. Zhao, Y., Hu, Y., Smith, J. P., Strauss, J. & Yang, G. Cohort proile:
The China Health and Retirement Longitudinal Study (CHARLS).
Int. J. Epidemiol. 43, 61–68 (2014).
42. Higgins, J. P. T. & Thompson, S. G. Quantifying heterogeneity in a
meta-analysis. Stat. Med. 21, 1539–1558 (2002).
43. Arellano, M. & Bond, S. Some tests of speciication for panel
data: Monte Carlo evidence and an application to employment
equations. Rev. Econ. Stud. 58, 277–297 (1991).
Acknowledgements
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 no. 1862896-38-C-20
to D.F.). The opinions expressed are those of the authors and do not
represent the views of the National Endowment for the Arts Oice
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. D.F. is additionally
supported by the Wellcome Trust (grant no. 205407/Z/16/Z). J.W.
is funded by the Belgian National Scientiic Fund (FNRS) Research
Associate Fellowship (CQ) no. 40010931. The Japan Gerontological
Evaluation Study (JAGES) is supported by the Japan Society
for the Promotion of Science (JSPS) KAKENHI Grant (grant no.
JP15H01972), Health Labour Sciences Research Grant (grant no.
H28-Choju-Ippan-002), Japan Agency for Medical Research and
Development (AMED) (grant nos. JP18dk0110027, JP18ls0110002,
JP18le0110009, JP20dk0110034, JP21lk0310073, JP21dk0110037,
JP22lk0310087), Open Innovation Platform with Enterprises,
Research Institute and Academia (OPERA, grant no. JPMJOP1831)
and the Research Founding for Longevity Sciences from National
Center for Geriatrics and Gerontology (grant nos. 29-42, 30-22, 20-
19, 21-20). All funders supported the study design, study analysis and
writing of the manuscript.
Author contributions
H.W.M., T.N., J.K.B., J.W., K.K., T.S. and D.F. designed the study.
H.W.M. and D.F. led the study. H.W.M. analyzed data from the English
Longitudinal Study of Ageing and ran pooled analyses and multilevel
models. T.N. analyzed data from the Japan Gerontological Evaluation
Study. J.K.B. analyzed data from the Health and Retirement Study.
J.W. analyzed data from the Survey of Health, Ageing and Retirement
in Europe. Q.G. analyzed data from the China Health and Retirement
Longitudinal Study. H.W.M. and D.F. drafted the manuscript. T.N.,
J.K.B., J.W., Q.G., K.K. and T.S. assisted with analytical issues,
contributed to the writing and made critical revisions. All authors
approved the inal manuscript.
Competing interests
The authors declare no competing interests.
Additional information
Extended data is available for this paper at
https://doi.org/10.1038/s41591-023-02506-1.
Supplementary information The online version
contains supplementary material available at
https://doi.org/10.1038/s41591-023-02506-1.
Correspondence and requests for materials should be addressed to
Daisy Fancourt.
Peer review information Nature Medicine thanks Fang Fang,
Alexandru Dregan, Sophie Wickham and the other, anonymous,
reviewer(s) for their contribution to the peer review of this work.
Primary Handling Editor: Ming Yang, in collaboration with the
Nature Medicine team.
Reprints and permissions information is available at
www.nature.com/reprints.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 1 | Directionality testing using Ordinary Least Squares
(OLS) regressions. N study = 15. Data were first analysed separately for each
country using OLS regression. OLS regressions were applied to estimate the
association between hobbies measured at Time 1 and outcomes measured at
Time 2, while controlling for baseline outcomes measured, age, gender, the
number of people living in the household, partnership status, household income,
housing tenure, employment status, educational level, long-term mental/
physical conditions, difficulties with daily activities (ADLs), and difficulties with
instrumental activities of daily living (IADLs). The findings were then pooled
into multi-national meta-analyses using the random effect model to estimate the
overall effect sizes for all outcomes. 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 error. H2
statistics describes the ratio of the observed variation and the expected variance
due to sampling error. Given that some of the analyses had more participants
than others and thus had lower sampling variability and more precise estimates,
the meta-analysis was weighted. Studies with a greater number of respondents
were given more weight than studies with a small number of respondents. These
were relative weights that summed to 100. Data are presented as OLS coefficients
and 95% confidence intervals. The overall effect size and its width should have
accounted for the between-study variance, the number of studies, the precision
of the study-specific estimates (or ‘effect sizes’) and the significance level.
a. Hobbies and depressive symptoms. b. Hobbies and self-reported health.
c. Hobbies and happiness. d. Hobbies and life satisfaction.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 2 | Multi-level models testing the moderating effects of
country-level factors. N study = 15. Multi-level analyses were performed. All
models controlled for interview waves, type of hobby measure, age, gender, the
number of people living in the household, partnership status, household income,
housing tenure, employment status, educational level, long-term mental/
physical conditions, difficulties with daily activities (ADLs), difficulties with
instrumental activities of daily living (IADLs). Data are presented as coefficients
and 95% confidence intervals. a. No interaction. b. Interacting with country
wealth. c. Interacting with Gini index. d. Interacting with world happiness index.
e. Interacting with life expectancy. f. Interacting with national pension age.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 3 | Bubble plot with fitted meta regression line of the
effect size for four outcomes and the prevalence of hobby engagement.
N study = 16. Data are presented as fixed effects coefficients (bubbles), 95%
confidence intervals (shaded area) and the linear prediction (red line).
a. Depressive symptoms. b. Self-reported health. c. Happiness. d. Life satisfaction.
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.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 4 | Bubble plot with fitted meta regression line of the
effect size for four outcomes and country wealth (GDP per capita of USD).
N study = 16. Data are presented as fixed effects coefficients (bubbles), 95%
confidence intervals (shaded area) and the linear prediction (red line).
a. Depressive symptoms. b. Self-reported health. c. Happiness. d. Life satisfaction.
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.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 5 | Bubble plot with fitted meta regression line of the
effect size for four outcomes and Gini index. N study = 16. Data are presented as
fixed effects coefficients (bubbles), 95% confidence intervals (shaded area) and
the linear prediction (red line). a. Depressive symptoms. b. Self-reported health.
c. Happiness. d. Life satisfaction. 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.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 6 | Bubble plot with fitted meta regression line of the
effect size for four outcomes and world happiness index. N study = 16. Data
are presented as fixed effects coefficients (bubbles), 95% confidence intervals
(shaded area) and the linear prediction (red line). a. Depressive symptoms.
b. Self-reported health. c. Happiness. d. Life satisfaction. 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.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 7 | Bubble plot with fitted meta regression line of the
effect size for four outcomes and life expectancy. N study = 16. Data are
presented as fixed effects coefficients (bubbles), 95% confidence intervals
(shaded area) and the linear prediction (red line). a. Depressive symptoms.
b. Self-reported health. c. Happiness. d. Life satisfaction. 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.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 8 | Fixed effects analyses of hobbies and mental wellbeing
outcomes for females. N study = 16. Data were first analysed separately for each
country using fixed effects regression. All models controlled for all time-constant
variables and time-varying variables including age, partnership status, number
of people living in the household, employment status, household income,
housing tenure, long-standing mental/physical conditions, difficulties with
daily activities (ADLs), and difficulties with instrumental activities of daily living
(IADLs). The findings were then pooled into multi-national meta-analyses using
the random effect model to estimate the overall effect sizes for all outcomes.
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 error. H2 statistics describes the ratio
of the observed variation and the expected variance due to sampling error.
Given that some of the analyses had more participants than others and thus had
lower sampling variability and more precise estimates, the meta-analysis was
weighted. Studies with a greater number of respondents were given more weight
than studies with a small number of respondents. These were relative weights
that summed to 100. Data are presented as fixed effects coefficients and 95%
confidence intervals. The overall effect size and its width should have accounted
for the between-study variance, the number of studies, the precision of the
study-specific estimates (or ‘effect sizes’) and the significance level. a. Depressive
symptoms. b. Self-reported health. c. Happiness. d. Life satisfaction.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 9 | Fixed effects analyses of hobbies and mental wellbeing
outcomes for males. N study = 16. Data were first analysed separately for each
country using fixed effects regression. All models controlled for all time-constant
variables and time-varying variables including age, partnership status, number
of people living in the household, employment status, household income,
housing tenure, long-standing mental/physical conditions, difficulties with
daily activities (ADLs), and difficulties with instrumental activities of daily living
(IADLs). The findings were then pooled into multi-national meta-analyses using
the random effect model to estimate the overall effect sizes for all outcomes.
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 error. H2 statistics describes the ratio
of the observed variation and the expected variance due to sampling error.
Given that some of the analyses had more participants than others and thus had
lower sampling variability and more precise estimates, the meta-analysis was
weighted. Studies with a greater number of respondents were given more weight
than studies with a small number of respondents. These were relative weights
that summed to 100. Data are presented as fixed effects coefficients and 95%
confidence intervals. The overall effect size and its width should have accounted
for the between-study variance, the number of studies, the precision of the
study-specific estimates (or ‘effect sizes’) and the significance level. a. Depressive
symptoms. b. Self-reported health. c. Happiness. d. Life satisfaction.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Nature Medicine
Article https://doi.org/10.1038/s41591-023-02506-1
Extended Data Fig. 10 | Fixed effects analyses of hobbies and mental
wellbeing outcomes for adults who were retired. N study = 16. Data were
first analysed separately for each country using fixed effects regression. All
models controlled for all time-constant variables and time-varying variables
including age, partnership status, number of people living in the household,
employment status, household income, housing tenure, long-standing mental/
physical conditions, difficulties with daily activities (ADLs), and difficulties with
instrumental activities of daily living (IADLs). The findings were then pooled
into multi-national meta-analyses using the random effect model to estimate the
overall effect sizes for all outcomes. 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 error. H2
statistics describes the ratio of the observed variation and the expected variance
due to sampling error. Given that some of the analyses had more participants
than others and thus had lower sampling variability and more precise estimates,
the meta-analysis was weighted. Studies with a greater number of respondents
were given more weight than studies with a small number of respondents.
These were relative weights that summed to 100. Data are presented as fixed
effects coefficients and 95% confidence intervals. The overall effect size and its
width should have accounted for the between-study variance, the number of
studies, the precision of the study-specific estimates (or ‘effect sizes’) and the
significance level. a. Depressive symptoms. b. Self-reported health. c. Happiness.
d. Life satisfaction.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-scale
personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By accessing,
sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these purposes, Springer
Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription (to
the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue, royalties,
rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal content cannot be
used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any other, institutional
repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or content on
this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature may revoke this
licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied with
respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law, including
merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed from
third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com