Content uploaded by Yang Hu
Author content
All content in this area was uploaded by Yang Hu on Nov 04, 2024
Content may be subject to copyright.
PNAS Nexus, 2024, 3, pgae465
https://doi.org/10.1093/pnasnexus/pgae465
Advance access publication 17 October 2024
Research Report
Art for health’s sake or health for art’s sake:
Disentangling the bidirectional relationships between
arts engagement and mental health
Hei Wan Mak
a,
*, Yang Hu
b
, Feifei Bu
a
, Jessica K. Bone
a
and Daisy Fancourt
a
a
Department of Behavioural Science and Health, Institute of Epidemiology and Health Care, University College London, 1-19 Torrington Place, London WC1E 7HB,
United Kingdom
b
Department of Sociology, Lancaster University, Lancaster LA1 4YN, United Kingdom
*To whom correspondence should be addressed: Email: hei.mak@ucl.ac.uk
Edited By Yannis Yortsos
Abstract
Increasing evidence links arts engagement with mental health, but the directionality of the link remains unclear. Applying a novel
approach to causal inference, we used nonrecursive instrumental variable models to analyze two waves of data from the United
Kingdom Household Longitudinal Study (n = 17,927). Our ndings reveal bidirectional causal relationships between arts engagement
(arts participation, cultural attendance, and heritage visits) and mental health (GHQ-12 mental distress and SF-12 Mental Component
Summary mental well-being). After adjusting for time 1 measures and identifying confounders, cultural attendance and heritage
visits were reciprocally associated with mental distress and mental well-being, while arts participation was only reciprocally
associated with mental well-being. The bidirectional effects between arts engagement and mental health are modest but clearly
demonstrated not only from mental health to arts but also from arts to mental health. Our ndings indicate that previous evidence of
an association between arts engagement and mental health is due to bidirectional causal effects. Interventions that boost arts
participation, cultural attendance, and heritage visits may help break the negative feedback loop and enhance mental health.
Keywords: nonrecursive model, arts engagement, mental health, GHQ-12 mental distress, SF-12 MCS mental well-being
Signicance Statement
This study addresses important theoretical and methodological questions surrounding the directionality of the relationship between
engagement with the arts, culture and heritage, and mental health. It shows that this relationship is causally bidirectional. After ad-
justing for time 1 measures and identifying confounders, cultural attendance and heritage visits are reciprocally associated with men-
tal distress and mental well-being, while arts participation is only reciprocally associated with mental well-being. We also found that
the bidirectional causal effects between arts engagement and mental health are similar in size.
Competing Interest: The authors declare no competing interests.
Received: May 22, 2024. Accepted: October 2, 2024
© The Author(s) 2024. Published by Oxford University Press on behalf of National Academy of Sciences. This is an Open Access article
distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits
unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Introduction
Over the past two decades, increasing recognition of the potential
of arts activities in helping improve mental health has been trans-
forming attitudes and approaches to mental health care. Current
developments include an increasing emphasis on “prescribing”
the arts to support the treatment and management of mental
health conditions, as well as embedding arts and mental health
reciprocally within the strategic objectives of public health and
cultural organizations to improve and maintain population men-
tal health (1, 2). Arts engagement broadly encompasses arts par-
ticipation (e.g. singing, dancing, and acting), cultural attendance
(e.g. opera, exhibitions, and galleries), and heritage engagement
(e.g. visiting historical sites, parks, and monuments). Despite dec-
ades of research demonstrating that arts activities are associated
with better mental health (3), the directionality of the association
remains unclear.
On the one hand, researchers have used various designs to as-
sess how arts engagement inuences multidimensional aspects of
mental health. To date, >3,000 studies have identied the role of
the arts in helping promote good mental health and prevent and
manage mental health conditions. For instance, intervention
studies that compare outcomes between treatment and control
groups have shown that activities, including singing, visiting mu-
seums, group drumming, dancing, photography, drama, drawing,
and painting, are associated with improved psychological well-
being and reduced depression and anxiety (4–8). These studies
are supported by larger scale observational research showing
that arts engagement is longitudinally and positively associated
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
with happiness, life satisfaction, self-esteem, resilience, positive
effect and purpose in life, and negatively correlated with loneli-
ness and social isolation (3).
In seeking to understand the mechanisms underlying these
ndings, hundreds of processes have been identied across vari-
ous elds (9, 10). Psychological research has demonstrated how
the arts can help regulate emotions, enhance a sense of self, and
support coping and resilience (11). Neuroscientic research has
identied the ability of the arts to modulate arousal, engage
emotion-related brain networks, activate reward networks
(including triggering dopamine release), and release neuropepti-
des such as oxytocin (12). Psychobiological research has identied
reductions in neuroendocrine markers of stress (including cate-
cholamines and glucocorticoids), regulation of inammatory bio-
markers (such as cytokines), and regulation of cardiovascular
and electrodermal activity (13). Social science research has shown
how the arts can facilitate social bonding, increase social capital,
and reduce social isolation, all of which are linked to better mental
health (3). Further behavioral processes have also been identied,
including those related to the maintenance of healthy lifestyles
and health literacy that reduce risk factors associated with mental
ill health (10).
On the other hand, however, arts engagement is far from equal.
Those with poorer mental health, long-standing mental health
conditions, and those experiencing low levels of happiness are
less likely to engage in cultural activities (e.g. live music events,
opera, exhibitions) (14) and participatory arts (e.g. singing, dan-
cing, arts, and crafts) (15). These associations exist independent
of individuals’ sociodemographic characteristics, although it has
also been shown that individuals’ socioeconomic position and
area deprivation explain part of the association between mental
health and engagement rates (14). As a result, the association
may reect differences in individuals’ experience of barriers relat-
ing to psychological and physical capabilities (e.g. lack of con-
dence, lack of sufcient energy, and strength), physical and
social opportunities (e.g. limited access and opportunities), and
motivations (e.g. less interest) to engage in arts activities (14–16).
Furthermore, engagement is lower among groups that are dispro-
portionately more likely to experience poorer mental health
(17, 18), including those from racial/ethnic minority backgrounds,
with fewer educational qualications, of lower socioeconomic sta-
tus, and living in more deprived areas (19–22).
Considered together, the two separate bodies of research imply
that the relationship between arts engagement and mental health
may be bidirectional or reciprocal, and theories applying the lens
of complex adaptive systems science suggest that positive and
negative feedback loops may support the maintenance of this bi-
directionality (10). However, this hypothesis remains to be tested
explicitly. The directionality and magnitude of the link between
arts engagement and mental health are important to explore for
at least three reasons. First, exploring whether the association is
unidirectional or reciprocal could provide important insights
into whether a feedback loop does in fact exist. Previous studies
have been criticized for assuming bidirectionality in the absence
of empirical data. Second, if arts engagement does exert an inu-
ence on mental health, this would have implications for the de-
sign of public health and policy strategies aiming to promote
better mental health through lifestyle and behavioral changes.
Third, if mental health does impact subsequent arts engagement,
this has implications for cultural organizations working to de-
velop strategies to ensure equitable opportunities for arts engage-
ment, suggesting that specic efforts may be required to ensure
accessibility for people with a mental health condition.
Against this backdrop, this study was designed to disentangle
the directionality of the relationship between arts engagement
and mental health using a nonrecursive structural equation mod-
eling strategy with instrumental variables (IVs). We included
three different dimensions of arts engagement (arts participation,
cultural attendance, and heritage visits), which involve various
active ingredients and mechanisms of action (9, 10), exploring
whether and with what level of magnitude a bidirectional rela-
tionship can be seen. Further, given that there is a widely accepted
two-continuum model of mental illness and mental well-being
(such that individuals can experience one but not the other), we
considered both dimensions independently (mental distress and
mental well-being), to enable a comparison of whether a bidirec-
tional relationship is found for both negative and positive aspects
of mental health.
Methods
Data
This study used data from the United Kingdom Household
Longitudinal Study (UKHLS, www.understandingsociety.ac.uk),
which started in 2009 and interviewed around 40,000 households
and 50,000 individuals in the baseline wave. The respondents are
re-interviewed every year, with new sample members added to
compensate for attrition. The UKHLS collects rich longitudinal
data on participants’ demographic characteristics, socioeconomic
positions, leisure activities, participation in the arts, and health. In
our study, data from waves 2 (2010–2012) and 5 (2013–2015) were
analyzed—the only waves in which questions on arts engagement
were included. Of the 54,569 adult participants (aged 16 and
above) in wave 2 (time 1), 42,762 were followed up in wave 5
(time 2). The UKHLS contains a self-completion module, which in-
cludes questions on mental health; and only a representative sub-
sample of respondents was invited to take part in the module. Of
the respondents who participated in both waves, 27,877 took part
in the self-completion module. In total, 26,023 participants pro-
vided complete data across all measures. Missing data (6.7%)
were handled using listwise deletion. Little’s test of missing com-
pletely at random (MCAR) showed that the missing data were not
MCAR (χ
2
distance = 5,353.5, degree of freedom = 758, P < 0.001)
(23). The missingness was largely attributed to responses on
whether or not respondents had been clinically diagnosed with
depression, education level, and living area. We used the longitu-
dinal self-completion weights developed by the UKHLS team to
account for survey and sampling design and attrition (24). This re-
sulted in a nal, weighted analytical sample of 17,927 respondents
(Diagram S1).
Measures
Frequency of engagement with arts, culture, and heritage
Respondents were asked about, overall, how often they engaged in
each type of engagement—namely, arts participation, cultural at-
tendance, and heritage visits—voluntarily in the last 12 months.
For arts participation, 14 activities were provided in the list, in-
cluding dancing, singing, playing an instrument, reading for
pleasure, photography, digital arts, and painting and crafting.
For cultural attendance, 14 events were included, such as art ex-
hibitions, street art displays, musical concerts, opera, and ballet.
For heritage visits, eight historical sites were included, such as vis-
iting a historical park or garden and a monument such as a castle.
A full list of the specic activities covered is presented in Table S1.
For each type of engagement, the respondents were asked to rate
2 | PNAS Nexus, 2024, Vol. 3, No. 11
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
their overall engagement frequency in the listed activities, using a
7-point Likert-type scale ranging from 0 (none in the past 12
months) to 6 (at least once a week).
Mental health
Two measures of mental health were analyzed: (i) mental distress
and (ii) mental well-being. For mental distress, we used the
12-item GHQ mental distress scale, which measures common
symptoms of anxiety and depressive disorder (α = 0.91) (25). All
items were rst summed and then averaged by the number of
items, yielding a scale ranging from 1 to 4; higher scores indicate
more severe mental distress. For mental well-being, we used the
SF-12 Mental Component Summary (MCS) (26), which includes
six items relating to vitality, social functioning, role limitations
caused by emotional problems, and mental health (α = 0.91 at
time 1; α = 0.90 at time 2). We summed up the scores for all items
and averaged the sum by the number of items to create a 1–5
scale, where higher scores indicate better mental well-being.
Covariates
Ten covariates that may confound associations between arts and
cultural engagement and mental health were included (19, 22, 27).
All covariates were measured at time 1 to retain a larger sample
size. These included age (years; top-coded at the 99th percentile
to minimize the inuence of outlier cases) and its quadratic
term, gender (men versus women), ethnicity (White ethnic versus
ethnic minorities), partnership status (without a cohabiting part-
ner/spouse versus married or cohabiting), living with children
aged 15 or under (yes versus no), educational qualication (under-
graduate degree or above versus no degree), employment status
(in employment versus not in employment), gross household in-
come (British pounds; top-coded at the 99th percentile), living
area (rural versus urban), and long-standing mental/physical ill-
ness (yes versus no).
We also identied and used a set of IVs to account for unobserved
endogeneity and potential reverse causality between arts engage-
ment and mental health, which are discussed in the next section.
Statistical analysis
To explore the direction of relationships between arts engage-
ment and mental health and strengthen causal inference, we
used nonrecursive IV models (28). Compared with a recursive
model where all causal effects are unidirectional and error terms
are uncorrelated, a nonrecursive IV model allows us to investigate
possible concurrent (rather than lagged) bidirectional reciprocal
relationships between the arts and mental health, where “feed-
back loops” between two variables can be tested and error terms
are allowed to be correlated. The bidirectional relationships can
be tested by jointly estimating multiple equations with multiple
dependent variables simultaneously and allowing residuals to
be correlated (to obtain joint SEs). Given that nonrecursive models
can often be unidentied, IVs were therefore introduced to aid
model identication (28, 29).
In our models, we estimated reciprocal paths between arts en-
gagement and mental health at time 2 (i.e. the current time point),
respectively, for each type of engagement and each mental health
outcome, while accounting for the error terms for both engage-
ment and mental health and their covariance. The models speci-
ed IVs as predictors for arts engagement and mental health to
account for unobserved endogeneity and potential reverse causal-
ity. Two types of IVs were included: auxiliary IVs (AIVs) and
model-implied IVs (MIIVs). On the one hand, AIVs are a common
type of IV and often include time-lagged variables (30). They can
be particularly helpful in identifying the feedback relationship be-
tween arts engagement and mental health as they are strongly cor-
related with the endogenous predictor being instrumented. MIIVs,
on the other hand, are identied based on existing knowledge to
build a model where particular variables correlate with the pre-
dictor being instrumented and are uncorrelated with the error of
the equation (30). Combining these two types of IVs, therefore,
helped satisfy the four main requirements of IVs: (i) relevance—to
be strongly correlated with the predictor being instrumented; (ii)
exclusion—to only affect the outcome of interest through the pre-
dictor; (iii) exchangeability—to not share common causes with the
outcome; and (iv) effect homogeneity—the effect of IV on the out-
come is constant and homogenous across individuals (31).
Three IVs instrumenting arts engagement were used: (i) arts
engagement measured at time 1, i.e. measures of arts participa-
tion, cultural attendance, and heritage visits at time 1, respective-
ly; (ii) average regional rate of arts engagement (individually for
each type of engagement) at time 1; and (iii) average regional
rate of public library visits at time 1. The region measure distin-
guishes nine broad regions in England, as well as Scotland,
Wales, and Northern Ireland. The latter two variables were iden-
tied as MIIVs and were expected to have an impact on arts en-
gagement. For example, a study used the average regional
cultural attendance rate (alongside book ownership and nancial
hardship on cinema, theater, opera, or other concert attendance),
as IVs instrumenting cultural participation, to estimate the causal
relationship between cultural participation and self-reported
health and depression (32). Studies on libraries suggested that
such spaces can promote arts engagement through enabling li-
brary visitors to consume knowledge and acquire new skills
(such as on arts and crafts or information on cultural performance
or heritage sites) and offering spaces for workshops, seminars,
and programs (such as poetry workshops or book clubs) (33).
Given that MIIVs were measured on a regional level, they were ex-
pected to be strongly associated with and relevant to individual
arts engagement and to be related to mental health only through
their effects on individual arts engagement (Diagram S2a; Fig. S1
for kernel density estimates of arts engagement by IVs).
Three IVs instrumenting mental health were used: (i) mental
health measured at time 1, i.e. measures of mental distress and
mental well-being at time 1, respectively; (ii) whether one was di-
agnosed with clinical depression at time 1; and (iii) whether some-
one they knew passed away in the past year at time 1. Again, the
latter two variables were identied as MIIVs and were expected to
have an impact on mental health. A large volume of psychological
literature has shown that stressful life events, such as the death of
a social contact, are associated with increased major depressive
episodes, panic disorder, anxiety, posttraumatic stress disorder,
generalized anxiety disorder (34), suicidal ideation, and loneliness
(34–37). Although there may be a situation where the person who
passed away might have provided resources and opportunities for
one to engage in the arts and hence affect their engagement, the
immediate shock of knowing someone who passed away is likely
to rst impact their mental health and well-being and later their
behaviors (38). Similarly, clinically diagnosed depression has
also been shown to have a direct impact on symptoms of mental
distress and well-being (39–41); although diagnosed depression
can impact arts engagement, it is usually through the most pre-
sent state of mental distress and well-being (Diagram S2b; Figs.
S2 and S3 for kernel density estimates of mental health by IVs).
To ensure the adequacy of the IVs (42), we used two-stage
least squares regression to perform three statistical tests: (i)
Mak et al. | 3
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
Kleibergen-Paap rk Langurange Multiplier (LM) statistics, (ii)
Kleibergen-Paap rk Wald F statistics, and (iii) the Sargan–Hansen
test. The null hypothesis of the Kleibergen-Paap rk LM is that the
equation is under-identied, testing whether the excluded
instruments are uncorrelated with the predictor being instru-
mented; in other words, whether the instruments are relevant. The
Kleibergen-Paap rk Wald F statistics measure weak instruments,
which occur when their correlation with the predictor being instru-
mented is low. The Sargan–Hansen test examines overidentifying re-
strictions, with the null hypothesis that the IVs are valid instruments
(i.e. IVs are uncorrelated with the error term and that they are cor-
rectly excluded from the estimated equation). These tests showed
that, across all models, we can reject the null hypothesis of tests
that the equation was under-identied (Kleibergen-Paap rk LM P <
0.05) or weakly identied (Wald F statistic ranges from 633.4 to
2,130.3). That means the identied IVs were strongly correlated
with the predictor being instrumented. The Sargan–Hansen test for
all models indicated that the IVs were valid instruments (Sargan–
Hansen P > 0.05; except for the model where mental well-being
was instrumented to estimate its relationship with arts participation
where P = 0.044). Full tables of these tests for all models are pre-
sented in supplementary material(Table S2 for mental health being
instrumented to estimate its relationship with arts engagement; and
Table S3 for arts engagement being instrumented to estimate its re-
lationship with mental health).
Once the IVs were identied, we estimated the reciprocal rela-
tionships between arts engagement and mental health individual-
ly for each type of engagement and measure of mental health,
resulting in a total of six nonrecursive IV models. All variables
were standardized to aid the interpretation of results. The models
included endogenous predictors measured at time 2 to test for
concurrent reciprocity. AIVs, MIIVs, and a set of covariates meas-
ured at time 1 were also included.
One advantage of nonrecursive models is that they allow for as-
sessing the concurrent bidirectional relationships between arts en-
gagement and mental health in the same wave simultaneously. Yet,
it is also possible that previous arts engagement may also inuence
current mental health, and previous mental health may inuence
current arts engagement. To test such lagged bidirectionality, we
ran cross-lagged structural equation models as an additional check
to explore a potential temporal ordering between arts engagement
and mental health. Results from cross-lagged models are likely to
be more conservative, since there was a time gap of ∼3 years be-
tween time 1 and time 2. Respondents’ current arts engagement is
likely to be a more relevant predictor of their current mental health
than their engagement 3 years ago (and vice versa) (43). Despite this,
if results from both nonrecursive and cross-lagged structural equa-
tion models were consistent, this would provide stronger evidence
for a causal relationship between arts engagement and mental
health, which has been the case in our ndings.
For both nonrecursive IV models and cross-lagged structural
equation models, standardized root mean squared residual
(SRMR) and coefcient of determination (CD) were estimated to
evaluate model t. The models showed SRMR <0.05 and a CD be-
tween 0.426 and 0.600, indicating the models were well t (44). All
analyses were performed using Stata version 18.0.
Results
Participant characteristics
The average age of the sample was 47.8 years (SD = 14.3) at time 1,
52.7% were female, 92.5% were of White ethnicity, 24.6% had a
degree or higher qualication, and 59.5% were employed.
Around 76.8% of the sample lived in an urban area and 35.2%
had a long-standing mental/physical illness. More than half of
the sample (55.8%) engaged with arts at least once a week,
18.1% attended cultural events, and 12.3% visited heritage sites
at least once a month. On average, the sample had a mean score
of 1.92 (SD = 0.36, range: 1–4) on the GHQ-12 (measuring mental
distress) and 4.04 (SD = 0.58, range: 1–5) on the SF-12 MCS (meas-
uring mental well-being) at time 1 (Table S4 and Fig. S4 show lev-
els of arts engagement at times 1 and 2).
Concurrent directionality
As shown in Fig. 1, a reciprocal relationship was observed at time 2
between cultural attendance and heritage visits and mental dis-
tress, such that more frequent engagement was associated with
lower mental distress (also see Tables S5a–S5c). Specically, every
one-SD increase in cultural engagement reduced mental distress
by 0.06 SD (95% CI = −0.09 to −0.03; P < 0.001). For heritage visits,
every one-SD increase in visits reduced mental distress by 0.05
SD (95% CI = −0.08 to −0.02; P = 0.004). Simultaneously, higher
mental distress was associated with less frequent cultural attend-
ance and heritage visits, with every one-SD increase in mental
distress reducing cultural attendance by 0.08 SD (95% CI = −0.11
to −0.05; P < 0.001) and heritage visits by 0.06 SD (95% CI = −0.09
to −0.03; P < 0.001). No associations were found between arts
participation and mental distress.
As Fig. 2shows, a reciprocal relationship was also found at time
2 for mental well-being and all types of engagement (also see
Tables S6a–S6c). Specically, every one-SD increase in arts partici-
pation, cultural engagement, and heritage visits increased mental
well-being by 0.04 SD (95% CI = 0.01 to 0.08; P = 0.009), 0.08 SD
(95% CI = 0.06 to 0.11; P < 0.001), and 0.08 SD (95% CI = 0.05 to 0.11;
P < 0.001), respectively. Simultaneously, better mental well-being
was associated with higher levels of arts participation (β = 0.05,
95% CI = 0.02 to 0.08; P = 0.002), cultural attendance (β = 0.11,
95% CI = 0.09 to 0.14; P < 0.001), and heritage visits (β = 0.08, 95%
CI = 0.06 to 0.11, P < 0.001).
Lagged directionality
When considering how arts engagement at time 1 may also inu-
ence mental health at time 2 and vice versa, we found that a
one-SD increase in mental distress at time 1 was associated with
a reduction in cultural attendance by 0.03 SD (95% CI = −0.05 to
−0.02; P < 0.001) and heritage visits by 0.03 SD (95% CI = −0.04
to −0.01, P < 0.001) 3 years later (Fig. 3; Tables S7a–S7c). People
who attended cultural events (β = −0.03, 95% CI = −0.05 to −0.01;
P < 0.001) or heritage sites (β = −0.02, 95% CI = −0.04 to −0.01;
P = 0.004) less often at time 1 experienced greater mental distress
at time 2. Concurring with the results from the nonrecursive mod-
els discussed in the preceding section, no associations were found
between arts participation at time 1 and mental distress at time 2.
For mental well-being, Fig. 4shows that those with better men-
tal well-being at time 1 participated in arts activities (β = 0.03, 95%
CI = 0.01 to 0.04; P = 0.002), attended cultural events (β = 0.06, 95%
CI = 0.05 to 0.08; P < 0.001), and visited heritage sites (β = 0.05, 95%
CI = 0.03 to 0.06; P < 0.001) more frequently at time 2 (also see
Tables S8a–S8c). People who engaged more in these activities
at time 1 were also more likely to report better mental well-being
3 years later (arts participation: β = 0.02, 95% CI = 0.00 to 0.03;
P = 0.010; cultural attendance: β = 0.04, 95% CI = 0.03 to 0.06;
P < 0.001; heritage visits: β = 0.04, 95% CI = 0.02 to 0.05; P < 0.001).
4 | PNAS Nexus, 2024, Vol. 3, No. 11
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
Discussion
This study addressed important theoretical and methodological
questions surrounding the directionality of the relationship be-
tween engagement with the arts, culture and heritage, and men-
tal health, and found that this relationship to be causally
bidirectional. After adjusting for time 1 measures and identifying
confounders, cultural attendance and heritage visits are recipro-
cally associated with mental distress and mental well-being,
while arts participation is only reciprocally associated with men-
tal well-being. The bidirectional causal effects between arts en-
gagement and mental health are relatively modest in scale,
which may be anticipated given arts are one aspect of people’s
broader characteristics and environment. But these effects are
clearly present not only in mental health to arts, but also in arts
to mental health. Cross-lagged models examining lagged direc-
tionality with a 3-year interval supported our main ndings of bi-
directional effects, showing that previous engagement with arts,
culture, and heritage affects current mental health and prior
mental health affects current engagement.
Previous studies suggested that mental health could be both
predictor and result of arts engagement, but mostly in two separ-
ate bodies of literature (3, 14–16), implying reciprocal effects be-
tween arts engagement and mental health. Very few studies
have systematically investigated the potentially reciprocal rela-
tionship between arts engagement and mental health, leaving
the conrmation of bidirectionality and the relative magnitude
of both directions of effect unclear. Using nonrecursive models
to parse out these effects, our analyses show that arts engage-
ment is linked with a lower level of mental distress and a higher
level of mental well-being, and people with better mental health
also engage in participatory arts activities and attend cultural
events and heritage sites more frequently. Importantly, the recip-
rocal effects remain after controlling for demographic back-
ground, socioeconomic position, and health prole. The use of
IVs, including previous levels of arts engagement/mental health
and variables commonly used as IVs (such as average regional en-
gagement rates in arts, culture, and heritage), has further
strengthened the ability of our models to identify concurrent (ra-
ther than lagged) reciprocal effects. Our IV nonrecursive approach
indicates that the bidirectional relationships between arts en-
gagement and mental health are not just a result of related eco-
nomic, cultural, and social capital. Instead, arts engagement can
itself confer mental health benets, and mental health can inu-
ence whether one engages in the arts. This nding is consistent
with a Swedish study on a working population that used a multi-
level structural equation modeling approach and found a lagged
bidirectional association between cultural activity provided at
the workplace and depressive symptoms among employees (45).
Our use of an IV approach to identify concurrent bidirectional re-
lationships across engagement types and multidimensional
measures of mental health further strengthens the evidence for
the bidirectional association (45). Our results invite careful critical
reections on the potential “vicious” and “virtuous” cycles be-
tween arts engagement and mental health.
A vicious cycle occurs when low arts engagement and poor
mental health reinforce each other, in which the situation may
spiral in a downward loop and worsen over time. There are several
theoretical reasons why individuals with poorer mental health
may experience more barriers to engaging in the arts. These bar-
riers are often related to their capabilities, opportunities, and mo-
tivations (46). For instance, one study found that people with
depression and anxiety were more likely to report feeling less psy-
chologically and physically capable of engaging (e.g. limited
awareness of different types of activities available) and report
having fewer social opportunities (16). They were also more likely
to report feeling lower enjoyment and fewer perceived benets
from engaging in the arts (16). These barriers can be persistent,
leaving people of poorer mental health with fewer coping options
to improve their health. The impacts of mental health on arts en-
gagement could also be understood through the expectancy-value
theory (47). This theory suggests that arts engagement decisions
can be inuenced by two sets of beliefs: expectations of success
(how one perceives their abilities in arts and culture) and subject-
ive task value (how one perceives enjoyment, usefulness, import-
ance, and cost of engagement) (47). People with poorer mental
health may hold these beliefs more negatively and may thus
feel less motivated to engage in the arts. Over time, this vicious
cycle can become very difcult to break without extrinsic
intervention.
Fig. 1. Nonrecursive IV structural equation models estimating the bidirectional relationships between engagement with arts, culture and heritage, and
mental distress (modeled separately for each type of engagement). All models controlled for all covariates are listed in Table S4. Three IVs instrumenting
arts engagement were included: arts engagement measured at time 1, average regional rates of arts engagement, and average regional rate of public
library visits. Three IVs instrumenting mental distress were included: mental distress measured at time 1, diagnosed clinical depression, and knowing
someone who passed away in the past year. T1 = time 1, wave 2. T2 = time 2, wave 5. Standardized beta and 95% CI (in parentheses) are presented. Bold
values denote statistical signicance at the P < 0.05 level, with specic P-values presented in the main text.
Mak et al. | 5
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
On the ip side, a virtuous cycle occurs when a person who en-
gages in the arts has better mental health, which provides greater
incentives for them to stay engaged with the activities and then
continues to support their mental health. As proposed in the
Health Belief Model (48), people may be more likely to engage in
an activity if they perceive benets in such engagement.
Applying this lens, individuals who recognize the health benets
of the arts may be more likely to engage in the activities, and
such recognition may be more commonly found in individuals
with better mental health. Furthermore, arts engagement often
involves a “social” active ingredient, in which the social elements
of arts engagement, through either involving social contacts or fa-
cilitating social interactions, could have an additional impact in
improving people’s mental health through developing social cap-
ital and group belonging (3). An increase in social capital and
group belonging may keep people in the loop of engagement and
promote future engagement, as highlighted by the social identity
theory (49).
Consequently, the vicious and virtuous cycles may widen
health inequalities between people with better and poorer mental
health, and between those who engaged and disengaged with the
arts. This suggests that the arts system and public health system
Fig. 2. Nonrecursive IV structural equation models estimating the relationships between engagement with arts, culture and heritage, and mental
well-being (modeled separately for each type of engagement). All models controlled for all covariates are listed in Table S4. Three IVs instrumenting arts
engagement were included: arts engagement measured at time 1, average regional rates of arts engagement, and average regional rate of public library
visits. Three IVs instrumenting mental well-being were included: mental well-being measured at time 1, diagnosed clinical depression, and knowing
someone who passed away in the past year. T1 = time 1, wave 2. T2 = time 2, wave 5. Standardized beta and 95% CI (in parentheses) are presented. Bold
values denote statistical signicance at the P < 0.05 level, with specic P-values presented in the main text.
Fig. 3. Cross-lagged structural equation models estimating the relationships between engagement with arts, culture and heritage, and mental distress
(modeled separately for each type of engagement). All models controlled for all covariates are listed in Table S4 and adjusted for time 1 arts engagement
(in the equation predicting time 2 arts engagement) or mental health (in the equation predicting time 2 mental health). Two IVs instrumenting arts
engagement were included: average regional rates of arts engagement and average regional rate of public library visits. Two IVs instrumenting mental
well-being were included: diagnosed clinical depression and knowing someone who passed away in the past year. T1 = time 1, wave 2. T2 = time 2, wave
5. Standardized beta and 95% CI (in parentheses) are presented. Bold values denote statistical signicance at the P < 0.05 level, with specic P-values
presented in the main text.
6 | PNAS Nexus, 2024, Vol. 3, No. 11
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
may mutually reinforce one another in constituting social and
health stratication, and intervention approaches that increase
people’s engagement in the arts or that mitigate poor mental
health could help prevent any negative feedback that may occur
between the two. While interventions in either arts engagement
or mental health would likely also alter the other, encouraging
arts engagement may be a more feasible approach to breaking
the negative feedback loop.
Another notable nding of our study is that, unlike cultural at-
tendance and heritage visits, arts participation is only reciprocally
associated with mental well-being but not mental distress. One
potential explanation may be that arts participation, which often
involves active and creative engagement, may be more strongly
associated with components of mental well-being (such as vital-
ity, social well-being, and emotional well-being). It is plausible
that such engagement may play a bigger role in improving one’s
positive emotions than reducing negative emotions, such as men-
tal distress, anxiety, and depression. Conversely, people with a
higher level of mental distress may be more attracted to receptive
activities, such as attending cultural events and visiting heritage
sites to boost their mental health, which require less effort. In par-
allel studies that focused on the frequency of using general prac-
titioner (GP) consultations (50) and cause-specic mortality (51),
differences were also observed in the effects of active and creative
engagement versus receptive engagement. This highlights that
different types of engagement involving various key active ingre-
dients (9) and mechanisms of action (10) may have differential
health impacts. It is, therefore, crucial to provide a variety of activ-
ities for people with different mental health needs to access and
participate. Alternatively, given that our denition of arts
participation included activities that could be performed alone,
individuals experiencing mental distress may have felt more mo-
tivational barriers to solo engagement, but may have been more
likely to engage in culture and heritage activities that are fre-
quently done with others or involve passing informal social
contact as part of engaging, thereby providing both the social
pressure to engage in the rst place and social conrmation in
support of that engagement (9). The lack of association between
arts participation and mental distress may also be related to art-
istic abilities that are more relevant to arts engagement (such as
playing a musical instrument, painting, and photography).
Our results help inform the theoretical and practical develop-
ment of interventions and policies across arts and culture, mental
health care, and public health sectors. In mental health-care set-
tings, there has been an increasing emphasis on integrating psy-
chological, social, biomedical, and behavioral approaches to
improving mental health (52) and supporting the health-care sys-
tem and mental health professionals. Offering arts, cultural, and
heritage activities within mental health-care settings can provide
alternatives or complements to pharmacological prescribing, po-
tentially reducing pharmacological prescriptions and the associ-
ated side-effects (53, 54). In the context of public health, it has
been shown that one in ve appointments in general practice in
the United Kingdom are for social reasons, such as loneliness
(55), which are linked to socioeconomic inequalities such as un-
employment, poverty, and discrimination. As socioeconomic pos-
ition, lifestyle, and culture play a substantial role in people’s
mental health (56), mental health conditions may be better man-
aged with the additional inclusion of a creative-based approach
that involves engagement in arts, cultural, and heritage activities
(57). Creative-based approaches are sustainable and scalable, pre-
senting a potential strategy with less-stigma attached than med-
ical approaches to meet the pressure from the growing population
with mental ill health. For instance, a recent economic analysis
found that a social prescribing intervention led to 4.7 fewer GP ap-
pointments and a direct cost saving of £78.4 per participant over a
5-month period (58). In addition, policies expanding access to the
arts, especially for individuals with poor mental health, are im-
portant in breaking any negative feedback loops. Overall, the posi-
tive implications of arts engagement on mental health provide
Fig. 4. Cross-lagged structural equation models estimating the relationships between engagement with arts, culture and heritage and mental well-being
(modeled separately for each type of engagement). All models controlled for all covariates are listed in Table S4 and adjusted for time 1 arts engagement
(in the equation predicting time 2 arts engagement) or mental health (in the equation predicting time 2 mental health). Two IVs instrumenting arts
engagement were included: average regional rates of arts engagement and average regional rate of public library visits. Two IVs instrumenting mental
well-being were included: diagnosed clinical depression and knowing someone who passed away in the past year. T1 = time 1, wave 2. T2 = time 2, wave
5. Standardized beta and 95% CI (in parentheses) are presented. Bold values denote statistical signicance at the P < 0.05 level, with specic P-values
presented in the main text.
Mak et al. | 7
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
further support for social prescribing and other related creative-
based mental health interventions that help improve population
mental health.
Our study has many strength, including demonstrating both
concurrent and lagged reciprocal relationships between arts en-
gagement and mental health, as well as comparing the magnitude
of these relationships. The rich UKHLS data enabled us to use so-
phisticated and robust statistical models (both nonrecursive mod-
els and cross-lagged models) with unique IVs to make stronger
conclusions about whether the relationships are causal. We
were also able to comprehensively test different types of activities
including arts participation, cultural attendance, and heritage vis-
its. However, the study is not without limitations. First, while the
GHQ-12 and SF12 are widely used for screening common mental
disorders and for assessing the impact of health on everyday life,
it remains to be explored whether the reciprocity found in this
study would be replicated if they were to be replaced by other men-
tal health measures, such as depression and anxiety. Second, the
question remains as to whether the strength of the reciprocal rela-
tionships identied in our study is static or evolve across the life
course and how long the relationships last. Although we adjusted
for a range of potential confounders and utilized the IV approach,
unmeasured heterogeneities could still have inuenced our nd-
ings. Third, respondents were asked about their overall frequency
of arts engagement (regardless of whether they had engaged in one
or more than one activity). This means their engagement fre-
quency could be inuenced by the number of activities in which
they were engaged. Relatedly, since measures of arts engagement
were only captured in waves 2 and 5 in UKHLS (2010–2015), our
study was not able to include more recent engagement forms
such as creating videos on social media. Despite this, UKHLS con-
tains a wealth of variables that could be used as covariates and IVs,
enabling us to investigate the bidirectional relationship between
arts engagement and mental health in a representative sample
of the UK population. Building on our baseline study and ndings,
new data collection will enable researchers to examine change and
continuity in the bidirectional relationships between the arts and
mental health.
This study has addressed a key theoretical question regarding
the direction of causality that is often raised in arts and health re-
search. It suggests that interventions that increase people’s en-
gagement in the arts or that mitigate poor mental health may
help prevent or break any negative feedback loops between arts
engagement and mental health. Future research should consider
the potential bidirectional relationships when investigating the
antecedents and/or consequences of arts engagement. The nd-
ings have several implications for practice. For people with poorer
mental health, enhancing routine participation in arts and cul-
tural activities may help to improve their mental health by redu-
cing mental distress and improving mental well-being. For the
health sector, building a strong partnership with the arts and cul-
tural sector may help formulate more personalized, nonpharma-
cological treatments. Finally, if people with poorer mental health
engage less in arts and cultural activities, exploring ways to en-
courage their engagement and removing barriers and obstacles
is key. These could involve co-designing a treatment plan with a
health-care professional or making the arts more accessible by
providing various delivery formats. Given increasing policy atten-
tion to the arts and health link from international bodies like the
World Health Organization, more evidence on the causal and re-
ciprocal associations between the arts and mental health will al-
low us to have a better understanding of the potential mental
health benets of arts engagement and how to effectively
encourage arts engagement to promote public health. While our
analysis establishes new evidence of the bidirectional associa-
tions between the arts and mental health averaged across the
population, future research could build on our study to further
examine sociodemographic heterogeneities in such links to help
develop more targeted interventions.
Supplementary Material
Supplementary material is available at PNAS Nexus online.
Funding
The research of D.F. is supported by the Wellcome Trust (205407/
Z/16/Z). J.K.B. is supported by the National Endowment for the
Arts (1862896-38-C-20), University of Florida, Pabst Steinmetz
Foundation, and Bloomberg Philanthropies.
Author Contributions
H.W.M., Y.H., and D.F. designed the study. H.W.M. analyzed data
and drafted the manuscript. Y.H., F.B., J.K.B., and D.F. assisted
with analytical issues. All authors contributed to the writing,
made critical revisions, and approved the nal manuscript.
Ethics Approval
The University of Essex Ethics Committee has approved all data
collection in the Understanding Society main study and innov-
ation panel waves, including obtaining consent for all data link-
ages except to health records.
Consent to Participate
Respondents aged 16 years or older provided informed consent to
participate.
Consent to Publish
Only anonymous data were analyzed, and thus, consent for pub-
lication is not applicable.
Data Availability
Understanding Society—The UKHLS data can be retrieved via the
UK Data Service: https://beta.ukdataservice.ac.uk/datacatalo
gue/series/series?id=2000053. Data documentation is available
from the Understanding Society website: https://www.understa
ndingsociety.ac.uk/documentation. All codes used for the main
and supplementary analyses are publicly available via the Open
Science Framework: https://osf.io/ar8s7/.
References
1 Dow R, Warran K, Letrondo P, Fancourt D. 2023. The arts in public
health policy: progress and opportunities. Lancet Public Health. 8:
e155–e160.
2 Morse DF, et al. 2022. Global developments in social prescribing.
BMJ Glob Health. 7:e008524.
3 Fancourt D, Finn S. 2019. What is the evidence on the role of the
arts in improving health and well-being? A scoping review.
Copenhagen: WHO Regional Office for Europe. Health Evidence
Network (HEN) Synthesis Report 67.
8 | PNAS Nexus, 2024, Vol. 3, No. 11
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
4 Coulton S, Clift S, Skingley A, Rodriguez J. 2015. Effectiveness and
cost-effectiveness of community singing on mental health-
related quality of life of older people: randomised controlled tri-
al. Br J Psychiatry. 207:250–255.
5 Thomson LJ, Lockyer B, Camic PM, Chatterjee HJ. 2018. Effects of
a museum-based social prescription intervention on quantita-
tive measures of psychological wellbeing in older adults.
Perspect Public Health. 138:28–38.
6 Fancourt D, et al. 2016. Effects of group drumming interventions
on anxiety, depression, social resilience and inflammatory im-
mune response among mental health service users. PLoS One.
11:e0151136.
7 Cruz-Ferreira A, Marmeleira J, Formigo A, Gomes D, Fernandes J.
2015. Creative dance improves physical fitness and life satisfac-
tion in older women. Res Aging. 37:837–855.
8 Martin L, et al. 2018. Creative arts interventions for stress man-
agement and prevention: a systematic review. Behav Sci (Basel).
8:28.
9 Warran K, Burton A, Fancourt D. 2022. What are the active ingre-
dients of “arts in health” activities? Development of the
INgredients iN ArTs in hEalth (INNATE) framework. Wellcome
Open Res. 7:10.
10 Fancourt D, Aughterson H, Finn S, Walker E, Steptoe A. 2021.
How leisure activities affect health: a narrative review and multi-
level theoretical framework of mechanisms of action. Lancet
Psychiatry. 8:329–339.
11 Fancourt D, Garnett C, Spiro N, West R, Müllensiefen D. 2019.
How do artistic creative activities regulate our emotions?
Validation of the Emotion Regulation Strategies for Artistic
Creative Activities Scale (ERS-ACA). PLoS One. 14:e0211362.
12 Vuust P, Heggli OA, Friston KJ, Kringelbach ML. 2022. Music in the
brain. Nat Rev Neurosci. 23:287–305.
13 Fancourt D, Ockelford A, Belai A. 2014. The psychoneuroimmu-
nological effects of music: a systematic review and a new model.
Brain Behav Immun. 36:15–26.
14 Fancourt D, Baxter L. 2020. Differential participation in commu-
nity cultural activities amongst those with poor mental health:
analyses of the UK Taking Part Survey. Soc Sci Med. 261:113221.
15 Fancourt D, Mak HW. 2020. What barriers do people experience
to engaging in the arts? Structural equation modelling of the re-
lationship between individual characteristics and capabilities,
opportunities, and motivations to engage. PLoS One. 15:e0230487.
16 Fancourt D, Baxter L, Lorencatto F. 2020. Barriers and enablers to
engagement in participatory arts activities amongst individuals
with depression and anxiety: quantitative analyses using a be-
haviour change framework. BMC Public Health. 20:272.
17 Djernes JK. 2006. Prevalence and predictors of depression in pop-
ulations of elderly: a review. Acta Psychiatr Scand. 113:372–387.
18 Salvatore MA, Grundy E. 2021. Area deprivation, perceived neigh-
bourhood cohesion and mental health at older ages: a cross
lagged analysis of UK longitudinal data. Health Place. 67:102470.
19 Mak HW, Coulter R, Fancourt D. 2020. Patterns of social inequal-
ity in arts and cultural participation: findings from a
nationally-representative sample of adults living in the UK.
Public Health Panor. 6:55–68.
20 Mak HW, Coulter R, Fancourt D. 2020. Does arts and cultural en-
gagement vary geographically? Evidence from the UK Household
Longitudinal Study. Public Health. 185:119–126.
21 Mak HW, Coulter R, Fancourt D. 2021. Associations between
neighbourhood deprivation and engagement in arts, culture
and heritage: evidence from two nationally-representative sam-
ples. BMC Public Health. 21:1685.
22 Bone JK, et al. 2021. Who engages in the arts in the United States?
A comparison of several types of engagement using data from
The General Social Survey. BMC Public Health. 21:1349.
23 Li C. 2013. Little’s test of missing completely at random. Stata J.
13:795–809.
24 Kaminska O, Lynn P. 2019. Weighting and Sample Representation:
Frequently Asked Questions. Understanding Society: The
Household Longitudinal Study. https://www.understandingsociety.
ac.uk/wp-content/uploads/documentation/user-guides/6614_main_
survey_user_guide_weighting_faqs.pdf.
25 Goldberg DP, et al. 1997. The validity of two versions of the GHQ in
the WHO study of mental illness in general health care. Psychol
Med. 27:191–197.
26 Ware J, Kosinski M, Keller S. 1995. How to score SF-12 physical and
mental health summary scales. 2nd ed. Boston (MA): The Health
Institute, New England Medical Center.
27 Silva M, Loureiro A, Cardoso G. 2016. Social determinants of
mental health: a review of the evidence. Eur J Psychiatry. 30:
259–292.
28 Paxton P, Hipp JR, Marquat-Pyatt S. 2011. Nonrecursive models: en-
dogeneity, reciprocal relationships, and feedback loops. Thousand
Oaks (CA): SAGE.
29 Nagase M, Kano Y. 2017. Identifiability of nonrecursive struc-
tural equation models. Stat Probab Lett. 122:109–117.
30 Bollen KA. 2012. Instrumental variables in sociology and the so-
cial sciences. Annu Rev Sociol. 38:37–72.
31 Lousdal ML. 2018. An introduction to instrumental variable as-
sumptions, validation and estimation. Emerg Themes Epidemiol.
15:1.
32 Weziak-Białowolska D, Białowolski P. 2016. Cultural events—
does attendance improve health? Evidence from a Polish longitu-
dinal study. BMC Public Health. 16:1–9.
33 Wali A, Marcheschi E, Severson R, Longoni M. 2001. More than a
hobby: adult participation in the informal arts. J Arts Manage Law
Soc. 31:212–230.
34 Keyes KM, et al. 2014. The burden of loss: unexpected death of a
loved one and psychiatric disorders across the life course in a na-
tional study. Am J Psychiatry. 171(8):864–871.
35 Stroebe M, Schut H, Stroebe W. 2007. Health outcomes of be-
reavement. Lancet. 370:1960–1973.
36 Herbert D, Young K, Pietrusin
ska M, MacBeth A. 2022. The men-
tal health impact of perinatal loss: a systematic review and
meta-analysis. J Affect Disord. 297:118–129.
37 Kristiansen CB, Kjær JN, Hjorth P, Andersen K, Prina AM. 2019.
The association of time since spousal loss and depression in
widowhood: a systematic review and meta-analysis. Soc
Psychiatry Psychiatr Epidemiol. 54:781–792.
38 Institute of Medicine (US) Committee for the Study of Health
Consequences of the Stress of Bereavement. 1984. Adults’ reac-
tions to bereavement. In: Osterweis M, Solomon F, Green M, edi-
tors. Bereavement: reactions, consequences, and care. Washington
(DC): National Academies Press (US).
39 Jonsson U, et al. 2011. Mental health outcome of long-term and
episodic adolescent depression: 15-year follow-up of a commu-
nity sample. J Affect Disord. 130:395–404.
40 Johnson D, Dupuis G, Piche J, Clayborne Z, Colman I. 2018. Adult
mental health outcomes of adolescent depression: a systematic
review. Depress Anxiety. 35:700–716.
41 McLeod GFH, Horwood LJ, Fergusson DM. 2016. Adolescent de-
pression, adult mental health and psychosocial outcomes at 30
and 35 years. Psychol Med. 46:1401–1412.
42 Baum CF, Schaffer ME, Stillman S. 2023. IVREG2: Stata module for
extended instrumental variables/2SLS and GMM estimation.
Mak et al. | 9
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
Statistical Software Components, Boston College Department of
Economics.
43 Liska AE, Felson RB, Chamlin M, Baccaglini W. 1984. Estimating
attitude-behavior reciprocal effects within a theoretical specifi-
cation. Soc Psychol Q. 47:23.
44 Cangur S, Ercan I. 2015. Comparison of model fit indices used in
structural equation modeling under multivariate normality. J
Mod Appl Stat Methods. 14:152–167.
45 Theorell T, Nyberg A. 2019. Cultural activity at work: reciprocal
associations with depressive symptoms in employees. Int Arch
Occup Environ Health. 92:1131–1137.
46 Michie S, van Stralen MM, West R. 2011. The behaviour change
wheel: a new method for characterising and designing behaviour
change interventions. Implement Sci. 6:42.
47 Eccles JS, Wigfield A. 2020. From expectancy-value theory to sit-
uated expectancy-value theory: a developmental, social cogni-
tive, and sociocultural perspective on motivation. Contemp Educ
Psychol. 61:101859.
48 Janz NK, Becker MH. 1984. The health belief model: a decade lat-
er. Health Educ Q. 11:1–47.
49 Hogg MA. 2016. Social identity theory. In: McKeown S, Haji R,
Ferguson N, editors. Understanding peace and conflict through social
identity theory. Cham: Springer. p. 3–17.
50 Løkken BI, Sund ER, Krokstad S, Bjerkeset O, Rangul V. 2023.
Association between engagement in cultural activities and use
of general practitioner consultations: 7-year follow-up of adults
from the HUNT study, Norway. BMJ Open. 13:e068004.
51 Løkken BI, Merom D, Sund ER, Krokstad S, Rangul V. 2020.
Cultural participation and all-cause mortality, with possible gen-
der differences: an 8-year follow-up in the HUNT Study, Norway.
J Epidemiol Community Health. 74:624–630.
52 Lehman BJ, David DM, Gruber JA. 2017. Rethinking the biopsy-
chosocial model of health: understanding health as a dynamic
system. Soc Personal Psychol Compass. 11:e12328.
53 Polley M, Bertotti M, Kimberlee R, Pilkington K, Refsum C. 2017. A
review of the evidence assessing impact of social prescribing on health-
care demand and cost implications. London: University of
Westminster.
54 Pharmacists and Social Prescribing. Royal Pharmaceutical
Society. 2024. https://www.rpharms.com/recognition/all-our-
campaigns/policy-a-z/social-prescribing.
55 Husk K, Elston J, Gradinger F, Callaghan L, Asthana S. 2019. Social
prescribing: where is the evidence? Br J Gen Pract. 69:6–7.
56 Marmot M, Allen J, Boyce T, Goldblatt P, Morrison J. 2020. Health
equity in England: the marmot review 10 years on. London:
Institute of Health Equity. health.org.uk/publications/reports/
the-marmot-review-10-years-on.
57 Mossabir R, Morris R, Kennedy A, Blickem C, Rogers A. 2015. A
scoping review to understand the effectiveness of linking
schemes from healthcare providers to community resources to
improve the health and well-being of people with long-term con-
ditions. Health Soc Care Community. 23:467–484.
58 Lynch M, Jones CR. 2022. Social prescribing for frequent attend-
ers in primary care: an economic analysis. Front Public Health.
10:3580.
10 | PNAS Nexus, 2024, Vol. 3, No. 11
Downloaded from https://academic.oup.com/pnasnexus/article/doi/10.1093/pnasnexus/pgae465/7824544 by guest on 04 November 2024
