The art of being mentally healthy: a study to quantify the relationship between recreational arts engagement and mental well-being in the general population

Arts engagement can be defined as active (e.g. making art) or receptive (e.g. attending concerts) involvement in creative events or activities within a variety of art forms such as the performing arts, visual arts and literature [16]. Arts engagement by members of the general population is high. For example, it is estimated that 78 % of British (16+ years), 86 % of Australian (15+ years), 85 % of New Zealander (15+ years), and 99 % of Canadian adults (15+ years) take part in creative events and activities each year [17‐20].

Clinical studies have found that arts engagement promotes patient recovery, relaxation and reduces patient stress, anxiety and depression [21‐23]. In cancer patients the arts have been used to enhance quality of life after medical treatment [24]. Arts programs for people experiencing mental ill-health have been linked with improved confidence, self-esteem and self-understanding [25, 26]. In the elderly, arts engagement reduces depressed mood, enhances self-worth and promotes positive aging [27, 28]. Arts programs for young people have been linked with improved motivation, self-image, hope for the future and self-esteem [29]. Where population-based studies have been conducted, arts engagement has been associated with perceptions of happiness and stress reduction [30, 31].

The arts may have a unique contribution to make to population health however, evidence-based, dose–response research about (1) the general population, and (2) the art people do as part of their everyday life, for enjoyment, entertainment or as a hobby, rather than therapy, is limited [32‐34]. As pressure on health resources grow, many are calling for holistic solutions that provide members of the general population with knowledge, choice and the capacity to attain higher levels of wellness and self-care [35, 36]. A better understanding of the arts-mental health relationship could contribute to population-based health messages, strategies, policy and practice (e.g. social prescribing or arts-on-prescription programs) [37]. As guided by theories of social epidemiology [38] and the biopsychosocial model of health [39], the first aim of this study was to determine if there was a relationship between arts engagement and mental well-being in the general population. If an association was demonstrated, the second aim was to quantify this relationship.

Between September 2011 and May 2012, telephone interviews were conducted with the Western Australian community (18+ years). Western Australia is representative of the broader Australian population in terms of key health and socio-demographic indicators [40]. Residential telephone numbers were randomly selected from the Australian Electronic White Pages telephone directory. The availability in Australia of a single telephone directory in computer format presents a comprehensive and cost-effective listing of residential “land-line” numbers [2, 41]. So as to be representative of the Western Australian population, (1) the survey research centre was given a sample target for location (30 % country) and sex (50 % female), and (2) once contacted, the adult in the household who would next be celebrating a birthday was invited to participate in the study.

The telephone survey was developed by the research team and guided by an online survey of 280 international experts in the field of the arts or arts-health regarding the definition of arts engagement for population based research [16]. The resulting telephone survey was reviewed by a panel of ten experts with experience in market research, the arts and/or public health and carried out by trained interviewers using a computer assisted telephone interview system [42]. The survey took 15 min to complete and included questions about mental well-being (dependent variable), arts engagement over the previous 12 months (independent variable), and eleven possible confounding or effect modifying variables to the arts-mental well-being relationship (i.e. sports engagement, religious activities, holidays, general health, and demographics).

The dependent variable in this study was subjective mental well-being and measured by asking respondents the 14 items contained in the Warwick-Edinburgh Mental Well-being Scale (WEMWBS), i.e. I’ve been feeling … optimistic, useful, relaxed, interested in others, good about myself, close to others, confident, loved, cheerful; I’ve had energy to spare; I’ve been … dealing with problems well, thinking clearly, able to make up my own mind and interested in new things [43]. WEMWBS measures the mental well-being of the general population. The scale includes hedonic (i.e. happiness, life satisfaction) and eudaimonic (i.e. positive relationships, psychological functioning) items which together measure mental well-being [44]. WEMWBS is designed to assess mental well-being itself and not the determinants of mental well-being (i.e. resilience, problem solving, etc.) [43]. WEMWBS was scored by summing responses (i.e. 1 = none of the time to 5 = all of the time) to each of the 14 items. WEMWBS population scores approximate to a normal distribution, with a minimum possible score of 14 and a maximum score of 70 (population average = 51) [43]. The scale has good face validity, test-retest reliability (Cronbach’s α = 0.83) and internal consistency (Cronbach’s α = 0.89) [43]. WEMWBS has been adapted for cross-cultural use in a number of different countries and has been translated into several languages [45]. Permission to utilise WEMWBS was granted by the University of Warwick.

The independent variable in this study was total hours engaged in the arts in the last 12 months. Quantifying engagement by asking questions about: (1) activities and events over the last 12 months, and (2) measurement in terms of time, are common in the literature [46‐49]. As shown in Fig. 1, arts engagement was measured by asking 14 questions, that is, attendance at arts events (6 questions), participation in the arts (5 questions), learning (1 question), work/volunteering (1 question) and arts related membership (1 question).

For each survey item, respondents were asked if they had engaged in the arts in the previous 12 months (yes/no). If ‘yes’, they were asked to describe the activity or event. Respondents were then asked approximately how many days in the last 12 months they had engaged in each type of activity or event, followed by (on a typical day), how many hours they spent engaging in that activity or event.

To control for the influence of confounding and effect modification, information about eleven possible covariates to the arts-mental health relationship were collected (Table 1). This included demographics (i.e. sex, age group, location, income, education, marital status, children) and a self-assessment of general health. Respondents were also asked if during the last 12 months, (1) they partook in a holiday or break from work for two or more consecutive weeks; (2) if at least once a week for most weeks, they attended a religious service/event at a place of worship (e.g. church, mosque, temple); or (3) if at least once a week for most weeks they engaged in sport (i.e. participation in sports activities, and/or attendance at sports events as a spectator, and/or membership of a sports organisation, society or club).

The analysis strategy involved a descriptive investigation of the data followed by Pearson chi-square tests to explore differences by arts engagement and ANOVAs to explore differences in average WEMWBS scores. Arts ‘attendance’ in the previous 12 months was calculated based on respondents indicating they had attended one or more of the six survey items relating to attendance. Similarly, ‘participation’ in the previous 12 months was calculated based on respondents indicating they had participated in one or more of the five survey items relating to participation. A respondent was considered to be engaged in the arts in the previous 12 months (prevalence) if they had attended an arts event, and/or participated in the arts, and/or took part in arts related learning, and/or worked or volunteered in the arts (on a non-professional basis) and/or had been a member of an arts organization, club or society. ‘Total days engaged in the arts in the previous 12 months’ was calculated by summing together the number of days respondents spent attending, participating, learning, working/volunteering or being a member of an arts organization, club or society. ‘Hours per day engaged in the arts in the previous 12 months’, was calculated by first multiplying hours on a typical day by number of days engaged in each type of arts activity over the last 12 months, this was then summed and the total divided by the sum of days engaged in each type of arts activity. ‘Hours per year engaged in the arts’, was calculated by first multiplying hours on a typical day by number of days engaged in each type of arts activity in the previous 12 months and summing each sub-total together. As the distribution of arts engagement was positively skewed (i.e. 17 % did not engage in the arts at all, median = 23 hours/year and 75^th percentile = 100 hours/year) and the relationship between mental well-being and arts engagement was non-linear, ‘hours per year engaged in the arts’ was grouped into four categories: no art = 0 hours/year, low arts engagement = 0.1 to 22.9 hours/year, medium arts engagement = 23 to 99.9 hours/year and high arts engagement = 100 or more hours/year. This was followed by linear regression analyses to investigate the association between arts engagement and WEMWBS scores. Overall, three models were fitted. The first model estimated the direct (unadjusted) effect of arts engagement; the second model estimated the effect of arts engagement after adjustment for demographics (i.e. age, sex, location, income, education, marital status and children); and the third model adjusted for demographics, general health, engagement in sport, religious activities and holidays from work. With the exception of the effect modification (i.e. interaction) analyses, results were assessed at the 0.05 level of significance. Effect modification was assessed at the 0.01 level of significance to reduce the possibility of a finding due to chance. The data were analysed using SPSS for Windows (Version 21) and SAS for Windows (Version 9.3).

Potential respondents were provided with an explanation of the study and invited to provide consent to participate at the beginning of the telephone interview. The respondents who agreed to take part in the study were assured that their answers were confidential, that they could withdraw from the study at any time and that all questions were voluntary. Respondents were made aware that the information collected would be used for research and publication purposes. Permission to conduct this study was granted by The University of Western Australia Human Research Ethics Committee (RA/4/1/2490).

As shown in Table 2, in the previous 12 months, 78 % of respondents had attended an arts event; 48 % participated in the arts; 11 % took part in arts related learning; 11 % worked or volunteered in the arts and 10 % were a member of an arts society, club or organisation. The highest number of respondents attended a performing arts (63 %) or visual arts event (51 %). On average, in the previous 12 months, respondents spent 16 hrs attending arts events; 63 hrs making or creating art; 5 hrs learning about the arts; 9 hrs working or volunteering in the arts (on a non-professional basis) and 7 hrs as a member of a arts society, club or organisation. The prevalence of arts engagement was 83 %. Respondents spent an average of 101 hours/year engaging in an arts activity or event (SD = 206, median = 23 hrs, minimum = 0 h, maximum = 1572 hrs).

When the data was grouped, 17 % of respondents did not engage in the arts, 33 % had low arts engagement (0.1 to 22.9 h/year), 24 % had medium arts engagement (23 to 99.9 h/year) and 26 % had high arts engagement (100 or more hours/year). As shown in Table 1, females were significantly more likely to be engaged in the arts than males (χ ² = 27.2 df = 1 p < 0.001); as were younger compared to older respondents (χ ² = 12.7 df = 4 p = 0.013). Arts engagement also significantly increased by education (χ ² = 16.7 df = 2 p < 0.001), and was more likely in those who also engaged in sport (χ ² = 17.1 df = 1 p < 0.001), attended religious services/events (χ ² = 7.9 df = 1 p = 0.005) and those who had taken a holiday or break from work for two or more weeks in the previous 12 months (χ ² = 20.8 df = 1 p < 0.001).

The WEMWBS mean score for respondents was 53.4 (median = 54.0, SD = 7.4, minimum = 21.0, maximum = 70.0). As shown in Table 1, respondents who were married or in a defacto relationship had higher average WEMWBS scores than unmarried respondents (F = 8.57 df = 1 p = 0.004); as did older compared to younger respondents (F = 2.91 df = 4 p = 0.021). Average WEMWBS scores were also significantly higher for those with good general health (F = 24.63 df = 4 p < 0.001) and those who had taken a holiday or break from work for two or more weeks in the previous 12 months (F = 20.23 df = 1 p < 0.001).

As shown in Table 3, Model 1 (unadjusted), respondents who engaged in a high level of arts engagement had higher average WEMWBS scores (54.8) than those who did not engage in the arts (52.8) or who had low (52.6) or medium (53.4) levels of engagement. With the exception of having “energy to spare”, for 13 out of the 14 WEMWBS questions, item means were highest for high arts engagement respondents compared to other levels of engagement, especially regarding optimism, interest in other people, thinking clearly, feeling loved, being interest in new things and feeling cheerful.

As shown in Table 3, Model 3, after adjustment for demographics, general health, engagement in sport, religious activities and holidays (Overall Test - 4 groups p = 0.003), it was found that people with a high level of arts engagement (i.e. 100 or more hours/year) had higher WEMWBS scores than those with none, low or medium engagement. In comparison, people with none, low and medium levels of arts engagement were found to have similar WEMWBS scores (Model 3, Overall Test - 3 groups p = 0.358), therefore a trend analysis was not applicable. Overall, the relationship between arts engagement and WEMWBS was nonlinear with evidence of a minimum threshold at 100 or more hours of arts engagement per year (i.e. high engagement vs non/low/medium engagement, Model 3, Overall Test - 2 groups p = 0.0006). A small decrease in significance level was found after adjustment for confounding variables, while effect modification by demographic, sport, religious activites, holidays or general health was not found to be significant. In general, respondents who participated in 100 or more hours of arts engagement per year had WEMWBS scores approximately two points higher than other levels of engagements (i.e. one third of a standard deviation).