Effects of arts on prescription for persons with common mental disorders and/or musculoskeletal pain: A controlled study with 12 months follow-up

Abstract

: Aims: Involvement in arts has shown potential to promote mental health. Thus, arts may be able to complement conventional healthcare to address common mental disorders (CMD). The aim of this study was to evaluate the long-term effect of a 10-week Arts on Prescription program regarding CMD (stress, anxiety, depression), compared to conventional healthcare. The study also aimed to examine whether CMD differed between groups. Methods: A quasi-experimental prospective design with intervention and control group and 6- and 12-month follow-up was used to evaluate an Arts on Prescription program in Sweden, focusing on the effects on stress, anxiety, and depression. Participants were on sick leave due to CMD and/or musculoskeletal pain. Data was collected using questionnaires. Results: The study population consisted of 479 participants (n = 247 intervention group, n = 232 control group). The result indicates a greater effect size (ŋ) in the intervention group compared to the control group for reduction in stress, anxiety, and depression at follow-up after 12 months. The difference in depression was significant. Conclusions: The results indicate AoP could be an adjunct to conventional healthcare interventions to address CMD, especially depression.

Keywords:

• culture activities
• mental health
• participatory arts
• social prescription

1. Introduction

Systematic reviews of existing evidence on arts and health, such as Fancourt and Finn (2019) and Dâmaso (2022), show that participation in arts activities can contribute to health promotion, disease prevention and with the management of ill health conditions. Further, research shows that taking part in arts in health programs promotes mental health and wellbeing on different levels, such as diminishing negative feelings, reducing stress, anxiety and depression, as well as increasing quality of life (Jensen & Bonde, 2018). Studies have also observed improved self-esteem, self-confidence (Chatterjee et al., 2018; Sumner et al., 2021) and social belonging (Redmond et al., 2019). Thus, arts may be able to complement conventional health care to address common mental disorders (CMD). Research shows CMD, including stress, anxiety, and depression, together with musculoskeletal pain, constitute a major and challenging public health problem worldwide (Hallman et al., 2019; Saxena et al., 2013). These health problems are often concurrent (Outcalt et al., 2015). The World Health Organisation (WHO) calls for community engagement (World Health Organization, 2017) and partnership between the health sector and actors outside healthcare to address the challenge regarding mental ill-health (World Health Organization, 2015, 2020). One example of such partnership is social prescribing (Bickerdike et al., 2017), a salutogenic approach aiming to promote wellbeing by referring to a number of non-medical approaches in a community. The theory of salutogenesis, which was developed by Antonovsky (1987), focuses on the origin and prerequisites of health rather than seeking the mechanisms underlying illness (pathogenesis). The salutogenic theory is the base for health promotion, which refers to strengthening a person’s or a population’s health potential to improve and maintain good health (Antonovsky, 1987). One form of salutogenic social prescription is Arts on Prescription (AoP) (Stickley & Hui, 2012).

AoP is an emerging and growing measure for promoting wellbeing. Studies indicate this kind of social prescription can contribute to an increased sense of belonging (Redmond et al., 2019) improved regained hope and meaning (Jensen, 2019; Makin & Gask, 2012) as well as improved self-confidence and empowerment (Stickley & Hui, 2012). These qualities are found to correspond with recovery from mental health problems (Leamy et al., 2011). Furthermore, studies reveal benefits of arts activities for managing musculoskeletal pain. One example is shared reading in a group (Billington et al., 2017) where (indications of) improvements in mood and pain have been identified. Another example is a dance-based program where a significant reduction of the level of pain was found (Murillo-Garcia et al., 2018).

Despite the growing body of arts and health research, there is a need for additional studies to move practice forward. There is a lack of large-scale controlled studies and studies with more long-term outcome as opposed to studies of short-term effects (Billington et al., 2017; Clift et al., 2021; O’Donnell et al., 2022). There is also a demand for studies that take greater account of effect size (Durlak, 2009) beyond the statistical significance of the effects of arts activities on health outcomes.

2. Research approach and methodology

The aim of this study was to evaluate the long-term effect of a 10-week AoP program regarding CMD (stress, anxiety, depression), compared to conventional healthcare. The study also aimed to examine whether CMD differed between groups. Participants were on sick leave due to CMD and/or musculoskeletal pain.

3. Research design

The study had a quasi-experimental prospective design with intervention and control groups, using questionnaires at baseline and follow-up at 6 and 12 months for evaluation. A total of six regions in south and middle Sweden were included in the study, two of which contained participants from both intervention and control groups. Although a randomized controlled design gives higher internal validity compared to a quasi-experimental design, the quasi-experimental design gives high external validity as it involves real-world interventions (Krauss, 2018). The selected design was also judged to be better suited for the current study partly due to the target group’s complexity but also to manage to include sufficient number of participants over a reasonable period of time.

Not all participants answered the questionnaire at both follow-up occasions. A decision was made to also include those who only participated in one follow-up. This decision was taken partly to avoid the risk of type two error, but also because it was not deemed ethically justifiable to exclude those who answered the extensive questionnaire on two occasions, baseline and one follow-up.

3.1. The intervention

The AoP program was community-based, i.e., performed outside of healthcare and coordinated by a person from the municipality (AoP-coordinator). The program in the current study followed an AoP model developed and evaluated, but not yet scientifically studied, for Swedish conditions (Gedeborg-Nilsson, 2015). The program is defined as participating in arts activities for 2.5 h twice a week in closed groups for 10 weeks. The groups were recommended to contain 6–10 participants. The AoP program included sessions with song, dance, drama, and different crafts activities such as painting, felting and pottery. Sessions also included library and museum visits, a theatre play and listening to a concert. In total six different arts professionals led the AoP-sessions. The model suggests that no group leader follows the group to all sessions and that all participants take part in all AoP sessions, but no one is forced to actively participate in activities they do not want to. This model had two aims, to promote personal development by inducing participants outside their comfort zones in a safe environment, and to provide participants with opportunities to experience various arts activities. The AoP program was offered to patients in primary care and outpatient psychiatric care. During 2020–2021, a limited number of AoP groups were carried out due to the COVID-19 pandemic.

3.2. Participants

Inclusion criteria for both the intervention and control groups were that persons be aged 18–66 years and on sick leave due to CMD (depression F31-F39, anxiety F40–42 and stress-related disorders F43-F48) and/or unspecified musculoskeletal pain (M79, R5.2) (World Health Organization, 2004). Exclusion criteria included psychosis and severe depression, suicide risk and known ongoing substance abuse. Participants, both in the intervention group and the control group, were able to participate in conventional treatment from healthcare professionals during the study period.

3.2.1. Recruitment of participants

AoP was one of several rehabilitation efforts that primary care and outpatient psychiatric care staff could offer their patients. If the patient met inclusion criteria and an individual in the healthcare team around the patient, such as a rehabilitation coordinator, counsellor, or doctor, thought that AoP would be an appropriate intervention, the patient was informed of the possibility of participating in AoP. All patients who attended the information meeting before the AoP started were considered prospective research participants. At the information meeting, information was given by the AoP coordinator, about the structure and content of the AoP program and about the study and the possibility of participating. Potential participants in the control group were selected for participation through a stratified selection process. This selection was made either by people who worked centrally with patient data in each region (five regions), or rehabilitation coordinators in the primary care units (one region). The stratification process covered all primary care areas in each region, including both urban and sparsely populated areas. It was performed as follows: persons on sick leave due to CMD or musculoskeletal pain at some point during the intervention period were identified. They were sorted, first by primary care units and then alphabetically by family name, (a-ö or ö-a) or by date of sick leave. The way this was performed differed between regions. Every fifth person was then selected until a maximum of 400 individuals were selected. Once identified, potential participants were asked to participate by letter containing information about the study and the baseline questionnaire.

3.3. Data collection

The questionnaire data collection took place between September 2014 and February 2021. It included background questions regarding age, gender, education, and worry about one’s finances. The included self-assessment instruments were Stress and Crises Inventory − 93 (SCI-93) and the Hospital Anxiety Depression Scale (HADS).

SCI-93 consists of 35 questions. Each item is scored on a Likert-type scale ranging from 0 to 4, generating a total score of 140 (Ericsson et al., 2015). The higher the degree of stress, the higher the value. SCI-93 has been applied in medical context in Sweden and showing good test–retest reliability (Krafft & Nyström, 2002). Cronbach’s α for the SCI-93 instrument in the study population ranged from 0,938- to 0,944.

HADS (Bjelland et al., 2002; Zigmond & Snaith, 1983), consisting of 14 questions divided into two subscales, is designed to measure anxiety and depression respectively. Each item is scored on a Likert-type scale ranging from 0 to 3, generating a total score of 21 of each subscale. The higher the degree of anxiety or depression, the higher the value (Bjelland et al., 2002; Zigmond & Snaith, 1983). HADS has shown satisfactory reliability in several studies and a stable two-factor structure that supports the two subscales (Bjelland et al., 2002). In this study, subscales showed satisfactory internal consistency for both intervention and control groups. Cronbach’s α for HADS in the study population ranged between 0.859 and 0.897 for anxiety, and 0.841 and 0.898 for depression.

3.4. Statistical analyses

Data were analysed using IBM SPSS Statistics version 27.0. If the non-response, at the individual level, was 15% or less, missing values were substituted. The fitted values of a two-way ANOVA with SCI-93 respectively HADS (all persons and all items in each analysis) as response variable, participants as random factor and concerned item in the instrument (SCI-93 or HADS) as fixed factor was used to replace the missing observations. This was done separately for baseline, 6-, and 12 months.

Stress, anxiety, and depression scores were normally distributed within groups (confirmed visually using histograms). Two-tailed chi-square tests and independent two-tailed t-test were performed on demographic characteristics to examine whether there were differences between the intervention and control groups at baseline. Participants included in the analysis had answered the questionnaire at baseline and follow-up 6 and/or 12 months. To see how the participants in the intervention and control group answered the various self-assessment instruments, see Table 1.

Table 1. Number of participants (n) in the intervention and control groups, respectively, who responded to HADS and SCI-93- at baseline and after 6 and/or 12 months

Answered occasions	Intervention group (n)		Control group (n)
	HADS	SCI-93	HADS	SCI-93
Baseline, 6, & 12 months	175	131	192	159
Only baseline & 6 months	35	46	7	10
Only baseline & 12 months	37	39	33	34
Total	247	216	232	203

A 5-way ANOVA was performed to analyse between-group difference, and effect size measured with Partial Eta Square (ŋ) (Fritz et al., 2012). Treatment group, gender, age group and follow-up time, were fixed factors while participants were random factors nested within treatment groups, gender, and age group. All main effects, as well as interaction between treatment groups and follow-up time was included in the model, while other 2-factor interactions between fixed factors were tested but found to be not significant and therefore not used. A 4-way ANOVA was performed to analyse within groups difference and effect size measured with ŋ. Gender, age group and follow-up time, were fixed factors while participants was a random factor nested within age group and gender, all as main effects in the model. To analyse mean differences in CMD between groups at baseline, a 3-way ANOVA was performed. Fixed factors were age group, gender and level of education respectively concern about finances. All main effects, no 2-factor interactions were significant and were therefore removed in model selection. For all analyses, the level of statistical significance was set at p < .05.

An a-priori power analysis was conducted. It estimated that an average difference of 2 units and a standard deviation of 6 units was adequate for independent two-tailed t-test, with a significance level of p < 0.05, and effect size of 0.33. The power analyses indicated that 200 participants in intervention- and control group respectively give approx. 90% power regarding independent two-tailed t-tests. If the study population was 150 participants in each group, approx.80% effect would be obtained.

4. Results

The study population included 479 participants (n = 247 intervention group, n = 232 control group). The distribution within the intervention group and control group regarding gender and age differed significantly with more women and older participants in the intervention group. Gender and age results have been standardized. Characteristics of the study population are presented in Table 2. More, the intervention group had significant higher mean for stress (p = <.001), anxiety (p = 0.05) and depression (p = <.001) at baseline compared to the control group, Table 2.

Table 2. Characteristics of the study population. Presented in descriptive statistic n (%), and difference between intervention- and control group at baseline (p-value)

Study population at baseline	Intervention group n = 247	Control group n = 232	p
Age (y) Mean (SD)	46 (1.7)	44 (12.8)	.06
Age categories n (%)			.003
18–24 y	7 (3)	11 (5)
25–44 y	88 (36)	110 (47)
45–66 y	152 (61)	111 (48)
Education n (%)			.02
Compulsory school	30 (12)	25 (11)
Upper secondary school	137 (56)	116 (50)
University/university college	75 (30)	88 (38)
Missing	5 (2)	3 (1)
Gender n (%)			<.001
Woman	227 (92)	176 (76)
Man	20 (8)	56 (24)
Worried about finances			<.001
Yes (Yes often/Yes rather often)	163 (66)	116 (50)
No (No, seldom/No, never)	76 (31)	112 (48)
Missing	8 (3)	4 (2)
CMC Mean (SD)
Stress	68.4 (25.7)	54.1 (27.9)	<.001
Anxiety	11.8 (4.6)	10,5 (5.0)	.04
Depression	9.4 (4.0)	7.6 (4.6)	<.001

A significant decrease between baseline, and follow-up in mean stress, anxiety and depression was found for both the intervention group and control group, illustrated in Table 3. The effect size ŋ was greater, but not statistically significant, in the intervention group regarding stress and anxiety. However, the change in depression was significantly greater in the intervention group compared to the control group.

Table 3. Mean scores for stress, anxiety and depression in intervention and control group, at baseline and follow-up at 6 and 12 months. Presented in mean, confidence interval (CI), p-value (p) and effect size partial eta squared b (ŋ). Table also present between group measure. Presented in p and effect size ŋ. Results are age and gender standardized

	Within groups difference				Between group difference
	Intervention group			Control group
	Mean (CI)	p	ŋ ^b	Mean (CI)	p	ŋ	p	ŋ
Stress		<.001	.16*		<.001	.09	.15	.005
Baseline	68.5 (76.0 ± 70.0)			54.5 (52.9 ± 56.0)
6 months	61.6 (59.8 ± 63.4)			49.4 (47.7 ± 51.2)
12 months	58.5 (56.8 ± 60.3)			47.7 (46.1 ± 49.3)
Anxiety		<.001	.20*		<.001	.12	.24	.003
Baseline	11.8 (11.5 ± 12.1)			10.5 (10.2 ± 10.9)
6 months	10.0 (9.6 ± 10.3)			9.0 (8.7 ± 9.4)
12 months	9.6 (9.3 ± 10.0)			8.9 (8.6 ± 9.2)
Depression		<.001	.19*		<.001	.09	.03	.008
Baseline	9.4 (9.1 ± 9.8)			7.6 (7.3 ± 7.9)
6 months	7.4 (7.1 ± 7.8)			6.3 (5.9 ± 6.7)
12 months	7.3 (6.9 ± 7.7)			6.3 (5.9 ± 6.6)

^bPartial ŋ=.01 indicates a small effect, partial ŋ=.06 indicates a medium effect and a partial ŋ=.14 indicates a large effect*

4.1. CMD differences between groups

In the intervention group, 65% of the participants answered, they were “often” or “quite often” worried about their finances compared to 50% in the control group, Table 2. When comparing mean score for stress, anxiety, and depression between participants (intervention- and control group together) who answered yes and no respectively to this question, significant differences were found, Table 4. Participants who worried about their finances reported significantly higher scores for stress (p = 0.004), anxiety (p=<0.001) and depression (p=<0.001) compared with those who did not worry. However, educational level had no significant association with stress (p = 0.113), anxiety (p = 0.69) nor depression (p = 0.83),

Table 4. Distribution and difference in mean score for stress, anxiety, and depression at baseline, distributed for level of education and concerns about finances. Results are age and gender standardized

	Compulsory school	Upper secondary school	University/College
	Mean (SE)	Mean (SE)	Mean (SE)	P
Stress	63.4 (4.8)	57.3 (3.6)	50.4 (3.9)	0.07
Anxiety	12.4 (0.8)	11.0 (0.6)	11.0 (0.6)	0.13
Depression	8.3 (0.7)	7.9 (0.5)	7.2 (0.6)	0.16
	Worried about finances “Yes”		Worriedabout finances “No”
	Mean (SE)		Mean (SE)	p
Stress	65.3 (3.3)		44.8 (3.4)	<.001
Anxiety	13.0 (0.5)		9.3 (0.5)	<.001
Depression	9.1 (0.5)		6.2 (0.5)	<.001

5. Discussion

The aim of the present study was to evaluate the long-term influence of a 10-week Arts on Prescription program on stress, anxiety, and depression, compared to conventional care. The study also aimed to examine whether CMD differed in relation to education level and participants’ concerns about their finances. An area where there is a need for more studies to increase knowledge.

Result indicates mean score for stress, anxiety, and depression decreased significantly both in the intervention group and in the control group between baseline and the follow-up after 6 and 12 months. Symptom reduction could be due to a natural recovery and should be evidenced by an equal effect in both groups over time if the intervention had little to no effect. Nonetheless, current results show a greater effect size (ŋ) in the intervention group (consistently large ŋ) compared to the control group (consistently medium ŋ) for stress, anxiety, and depression. This indicates that the capacity for AoP as a complementary intervention could be more effective than conventional healthcare, regarding included CMD. But, when analysing the change between intervention and control group, only mean score decrease in depression was significantly greater in the intervention group than in the control group. It is challenging to measure CMD. The diagnostic groups are complex, and the results are not easily transferable from group results to individual generalizability (van Os et al., 2019). Therefore, the results should be treated with caution. Due to the challenge measuring CMD, mixed methods studies are essential to understand in greater depth how AoP influences this target group. It is, therefore, of great value to reflect on the results of this study with a previously published qualitative study (Bergman et al., 2021) based on focus groups discussions with participants from the same AoP program. Present results verify previous findings of (author et al) where the participants express decreased CMD symptoms, and increased wellbeing. Our result is also in line with previous research that suggest participation in arts activities promotes mental health and wellbeing (Chatterjee et al., 2018; Jensen & Bonde, 2018; Sumner et al., 2021). According to the WHO, healthcare should seek solutions within their own organization and multidisciplinary solutions and partnerships with others in regard to mental ill health (World Health Organization, 2015, 2017). The current study illustrates one such multidisciplinary intervention performed in partnership between regions (healthcare organisation), municipalities, and arts professionals.

Further, our result showed participants who reported worry over finances also reported higher levels of stress, anxiety, and depression. Since the participants’ income was not known, their financial worries might be an expression of higher stress, anxiety and/or depression. On the other hand, it could be a dependence on sick leave benefits, which leads to financial uncertainty. That concern for finances has a negative effect on health has been shown in previous studies. Result in a study by Goldman-Mellor et al. (2010) suggest that negative economic transitions predict increased mental ill health such as depression and suicide. Further, Weich and Lewis (1998) found that financial strain is a powerful independent predictor of both the onset and maintenance of episodes of CMD. To achieve more equal health (Marmot et al., 2008), societal efforts to reduce people’s financial strain are an important factor to be considered and problematized. Another contributing factor may be that a demanding work situation has negatively affected family and work, contributing to ill health (Bratberg et al., 2002).

Furthermore, there was a disproportionate preponderance of women in the study sample, which is a weakness if you want to compare AoP’s effect between genders. However, the skewed distribution in the study population reflected the skewed gender distribution of those participating in AoP and the skewed distribution of sick leave for CMD between men and women (Försäkringskassan [the Swedish Social Insurance Agency], 2021; Freeman, 2022). This difference between women and men participating in AoP raises questions. Amongst the possible reasons for this, it may be men are offered AoP to a lesser extent than women or decline to a greater extent. Studies indicate men seek care to a lesser extent for CMD due to societal norms (Addis & Mahalik, 2003) (Affleck et al., 2018). problematize and discuss this issue and lifts traditional masculine norms and social construction of gender as part of the explanation of why men do not seek help for mental health problems. The gender question is essential to study in the future, highlighted in a recent systematic review of the effects of participatory arts interventions (O’Donnell et al., 2022). Based on their results, O’Donnell et al. (2022) claim that there is a need for more participatory arts interventions that are specifically aimed at men. Incorporating gender-sensitive approaches into participatory arts interventions is critical to improving male engagement.

This study supports previous studies of arts interventions regarding the relationship between participating in arts activities and reducing CMD (Chatterjee et al., 2018; Sumner et al., 2021), but also adds new information. The present study contributes to the growing body of evidence on arts and health with some requested methodological approaches (Belfiore, 2006; Clift et al., 2021) and provides a quantifiable contribution of changes of stress, anxiety, and depression. Despite this contribution, this area of research is still new and developing and more studies are still needed.

6. Conclusions and implications

The result indicates a greater (but not statistically significant) effect size ŋ for AoP on all CMD compared to the control group. For depression, however, the effect of AoP was significantly greater than the control group. This suggests that AoP could be an adjunct to conventional healthcare to address the growing prevalence of public health challenges such as CMD, especially depression. Considering the benefits of arts promoting mental wellbeing and decreasing CMD, it is vital to invest in organizing, financing, and carrying out initiatives to facilitate access to arts activities for persons with CMD as a complement to conventional healthcare.

Ethical approval

The Regional Ethical Board approved the study in Linköping Sweden (Dnr 2017/13–31). Written informed consent was obtained from participants. The participants of this study did not give written consent for their data to be shared publicly, so due to the sensitive nature of the research supporting data is not available.

Acknowledgments

We are grateful to all participants that allocated time to answering the study’s questionnaires. We also want to thank Bo Rolander and Olle Eriksson, Futurum Academy for Health and Care, Region Jönköping County who have provided great support when processing the data. Additionally, Birgitta Ekeberg, Department of Public Health and Healthcare, Region Jönköping County and Jenny Melin, The Coordination Association Älv & Kust who aided with execution of the intervention group’s questionnaires.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by grants from Futurum the Academy for Health Care, Jönköping County Council, Jönköping, Sweden.