Associations between occupational and social interaction factors and well-being among people with psychiatric disabilities living in supported housing in Sweden

ABSTRACT

Research indicates that occupation is important for well-being in people with mental illness, but this has not been extensively studied among those with severe psychiatric disabilities. Social contacts may possibly play a more vital role for them. This study aimed to explore how aspects of occupation and social interaction were related with well-being factors in that group, while controlling for the influence of clinical factors. People with psychiatric disabilities living in congregate supported housing (CSH; N = 155) responded to questionnaires addressing occupation, social contacts and well-being aspects, such as subjective health, quality of life, self-mastery, and personal recovery. A comparison group with psychiatric disabilities who lived in an ordinary flat or house and received outreach housing support (N = 111) completed the same instruments. The two groups were compared regarding their pattern of associations between occupation, social contacts, and well-being. Associations between occupation and well-being in the CSH group showed that general satisfaction with everyday occupations in particular was related to all aspects of well-being, whereas activity level and occupational balance were not related to well-being. The relationships were fewer and weaker, in comparison to the group in ordinary housing with outreach support. Indicators of social contacts were basically unrelated to well-being. The study contributes to occupational science by showing that the role of a high activity level for well-being, although important, should not be overemphasized. Future research should focus on narratives to get the voices of people residing in CSH and on exploration of how individually/socially performed occupations are associated with well-being.

Si bien las investigaciones indican que la ocupación es importante para el bienestar de las personas con enfermedades mentales, esto no se ha estudiado ampliamente entre personas que presentan discapacidades psiquiátricas graves. Posiblemente, los contactos sociales pueden desempeñar un papel más vital para ellas. El objetivo del presente estudio fue explorar cómo los aspectos de la ocupación y la interacción social se relacionan con los factores de bienestar de este grupo menos estudiado, controlando al mismo tiempo la influencia proveniente de los factores clínicos. Las personas que padecen discapacidades psiquiátricas y residen en viviendas con apoyo colectivo (CSH; N=155) respondieron cuestionarios que abordaban la ocupación, los contactos sociales y los aspectos del bienestar, como la salud subjetiva, la calidad de vida, el autodominio y la recuperación personal. Asimismo, un grupo de comparación conformado por personas con discapacidades psiquiátricas que vive en un departamento (piso) o casa ordinaria y recibe apoyo para la vivienda (N=111) completó los mismos cuestionarios. Los dos grupos se contrastaron en cuanto a su patrón de asociaciones entre ocupación, contactos sociales y bienestar. Las asociaciones entre ocupación y bienestar en el grupo CSH mostraron que la satisfacción general, sobre todo con las ocupaciones cotidianas, está relacionada con todos los aspectos del bienestar, mientras que el nivel de actividad y el equilibrio ocupacional no se relacionan con este. En comparación con el grupo que reside en viviendas ordinarias con apoyo externo, las relaciones son menores y más débiles. En lo fundamental, los indicadores de contactos sociales no están relacionados con el bienestar. El estudio contribuye a la ciencia ocupacional al mostrar que, aunque el papel ejercido por un alto nivel de actividad es importante para el bienestar, no debe ser sobrevalorado. Las investigaciones futuras deberán centrarse en las narraciones, a fin de conseguir las opiniones de las personas que residen en CSH y analizar cómo las ocupaciones realizadas individual o socialmente se vinculan con el bienestar.

研究表明，生活活动对精神疾病患者的福祉很重要。但在严重精神障碍患者中，这一点尚未得到广泛研究。社会交往可能对他们起着更重要的作用。本研究旨在探讨在控制临床因素影响的同时，生活活动和社会互动的各个方面如何与该群体的福祉因素相关。居住在集体福利房（CSH；N=155）的精神障碍患者对涉及生活活动、社会交往和福祉各方面（如主观健康、生活质量、自我控制和个人康复）的问卷进行了回应。一个精神障碍患者的比较组完成了同样的问卷调查。他们住在普通的公寓或房子里，并获得住房支持。（N=111）。比较两组的生活活动、社会交往和福祉之间的联系模式。CSH组的生活活动与福祉之间的关系表明，对日常生活活动的总体满意度尤其与福祉的各个方面有关，而活动水平及生活活动平衡与福祉无关。与获得支持的普通住房群体相比，这种关系较少且较弱。社会交往指标与福祉基本无关。这项研究对生活活动科学的贡献在于，它表明高活动水平对福祉的作用尽管很重要，但不应过分强调。未来的研究应该集中在叙述上，以获得居住在CSH的人们的声音，并探索个人的/社会的生活活动如何与福祉相关。

RÉSUMÉ

La recherche montre l'importance des occupations pour le bien-être des personnes ayant des troubles mentaux, mais cela n'a pas été étudié en profondeur chez les personnes vivant avec des troubles psychiatriques sévères. Les contacts sociaux peuvent éventuellement jouer un rôle plus essentiel pour ces personnes. Cette étude visait à explorer comment les aspects de l'occupation et de l'interaction sociale étaient liés aux facteurs de bien-être dans ce groupe, tout en contrôlant l'influence des facteurs cliniques. Des personnes présentant des troubles psychiatriques habitant dans des logements supervisés collectifs (LSC; N = 155) ont répondu à des questionnaires au sujet des occupations, des contacts sociaux et d'aspects du bien-être, comme la santé subjective, la qualité de vie, la maîtrise de soi et le rétablissement personnel. Un groupe de comparaison composé de personnes présentant des troubles psychiatriques habitant dans un logement ou une maison ordinaire et recevant un soutien de proximité (N = 111) a répondu aux mêmes questionnaires. Les deux groupes ont été comparés en examinant les associations entre occupations, contacts sociaux et bien-être. Les associations entre occupations et bien-être dans le groupe LSC ont montré que la satisfaction générale à l’égard des occupations quotidiennes était liée à tous les aspects du bien-être, tandis que le nombre d'activités et l’équilibre occupationnel ne l’étaient pas. Les relations étaient moins nombreuses et plus faibles comparativement au groupe habitant un logement ordinaire et bénéficiant d'un soutien de proximité. Les indicateurs de contacts sociaux n'avaient pratiquement aucun lien avec le bien-être. L’étude contribue aux sciences de l'occupation en démontrant que, bien qu'il soit important, le rôle d'un nombre important d'activités pour le bien-être ne devrait pas être survalorisé. Les futures recherches devraient se concentrer sur les récits, afin de faire entendre la voix des personnes résidant en LSC, ainsi que sur l'exploration des liens entre les occupations individuelles/sociales et le bien-être.

KEYWORDS:

• Occupational science
• Participation
• Life satisfaction
• Supported housing
• Social support
• Mental health

MOTS-CLÉS:

• Sciences de l'occupation
• participation
• satisfaction de vivre
• logement supervisé
• soutien social
• santé mentale

Well-being is a desirable state, which is sometimes seen as a personal (WHO, 1946) or humanistic (Medin & Alexandersson, 2000) aspect of health. Ryff (1989) described well-being as constituted by autonomy, personal growth, environmental mastery, purpose in life, positive relations, and self-acceptance, and the relevance of such a definition has been confirmed in more recent research (Ryan & Deci, 2000, 2001; Venkatapuram, 2013).

Research within occupational science and occupational therapy focusing on people with mental illness indicates a close relationship between their subjective perceptions of everyday occupation and well-being. These perceptions include the meaning, satisfaction, and value linked with the occupations, and various aspects of health and well-being, such as self-rated health, quality of life, and self-mastery (Aubin, Hachey, & Mercier, 1999; Bejerholm & Eklund, 2007; Eklund, Hansson, & Bejerholm, 2001; Eklund & Leufstadius, 2007; Nagle, Valiant Cook, & Polatajko, 2002). More recent research has indicated the importance of meaningful occupation for personal recovery from mental illness (Borg & Davidson, 2008; Le Boutillier et al., 2011; Sutton, Hocking, & Smythe, 2012) and personal recovery is increasingly seen as an equivalent to quality of life and well-being (Slade, 2010).

People with mental illness are not a homogeneous group, however, and some are content with a low-active lifestyle (Bejerholm & Eklund, 2004). This may refer particularly to the subgroup of people with mental illness who have a psychiatric disability, which in Sweden has been defined as an enduring (> 2 years) diminished ability to cope with everyday chores (Swedish Government Official Reports, 2006). Recent research compared two groups regarding occupation and well-being factors – people with psychiatric disabilities living in congregate supported housing (CSH) and those living in ordinary housing receiving outreach housing support (Eklund, Argentzell, Bejerholm, Tjornstrand, & Brunt, 2017). Those who resided in CSH had a lower level of psychosocial functioning and a lower activity level compared to the other group, but were more satisfied with their health and quality of life. This indicates that the relationship between occupation and health might not be as straightforward among those with more severe psychiatric disabilities, a low level of psychosocial functioning, and a need for CSH (Eklund et al., 2017). A literature search indicated a knowledge gap regarding the links between occupation, health, and well-being in this group with severe psychiatric disabilities. However, a qualitative study that focused on what brings meaning in life for people leading their lives in the CSH context, reported social interaction as the most frequent theme, followed by engagement in occupations (Eklund, Hermansson, & Håkansson, 2012).

The significance of social interactions for people with mental illness has been shown in other research as well, such as associations between both quantitative and qualitative aspects of the social network and well-being factors (Bengtsson-Tops & Hansson, 2001; Brunt & Hansson, 2002; Eklund & Hansson, 2007). A study in deprived areas, where poor mental health is a common problem (Bond et al., 2012), showed that loneliness and poor mental health were strongly associated (Kearns, Whitley, Tannahill, & Ellaway, 2015).

Although occupations are frequently performed in a social context, and the phenomena are thus often related, attempting to distinguish between them could generate knowledge about their specific relationships to well-being. The relative importance of occupation and social interaction for well-being factors among people with mental illness does not appear to have been researched, which thus adds to the knowledge gap that should be addressed. A greater understanding of this complexity would further the knowledge base of occupational science. Taking for granted that occupation promotes well-being among people with psychiatric disabilities residing in CSH, just because such associations have been demonstrated in other groups with mental illness, may lead to detrimental support measures in the CSH context where social contacts may possibly play a more vital role. These were the rationales for the current study, which had five specific aims:

1. To explore relationships between various aspects of occupation and well-being factors such as subjective health, life satisfaction, self-mastery, and personal recovery among residents in CSH.

2. To explore relationships between social interaction and the aforementioned well-being factors in this group.

3. To explore relationships between clinical and sociodemographic factors and the aforementioned well-being factors in this group.

4. To shed further light on these associations (i – iii) by comparing the strengths of associations found in the CSH context with associations based on a sample with psychiatric disabilities living in an ordinary flat or house receiving outreach housing support.

5. To use multivariate modelling to study relationships between various aspects of occupation, social interaction, clinical factors and socio-demographics on the one hand and the selected well-being factors on the other among residents in CSH.

Methods

This was a secondary analysis of data from a cross-sectional comparative study performed in the supported accommodation context in four regions in southern, western, inland, and eastern Sweden (Eklund et al., 2017). The study was approved by the Regional Ethical Review Board in Lund, Reg. No. 2013/456.

Supported accommodation in Sweden

CSH in Sweden consists of either fully-fitted flats in a housing block, or a bedroom with communal bathroom and eating facilities. The availability of a private kitchen varies between facilities. The staff provide support from office hours up to 24 hours per day for occupations such as shopping, cooking, and doctors’ appointments. CSH facilities generally have a staffed common area and a space where in-house staff can organize occupations. Those who receive outreach housing support in their ordinary flat or house may receive support from once a week up to several times per day, but the most common frequency is 1–2 times per week (Eklund et al., 2017). The type of support is similar in content to that available in CSH but the frequency is considerably lower. The differences in premises and support means that the residents in CSH live closer to and are more likely to meet their neighbors and have more social contacts compared to those living in an ordinary flat or house.

Selection procedure and recruitment

An invitation to participate in the study was addressed to local authorities in municipalities in the four aforementioned regions, strategically selected to represent variation regarding geographical location, socioeconomic background of the population, proportion of migrants, and rural/urban areas. The two larger cities among the approached municipalities were organized in city districts, which entailed that the invitation was sent to authorities at the district level. Nineteen districts, either municipalities or city districts, were invited and 17 agreed to participate. Maximum variation sampling (Patton, 2001) was then applied to select the specific supported accommodation units. Variation was sought on size, but also whether the housing units were well or less well-functioning (in terms of staff’s education level, leadership style, and the unit’s psychosocial climate). These latter aspects were assessed by the manager at the district level who suggested which units to approach to recruit participants living in CSH.

The comparison group living in ordinary housing with outreach support was recruited in the same regions and the same local authorities were approached. Four of the municipalities/city districts declined to collaborate in recruiting participants living in ordinary housing and four other equivalent districts were sought. A further 10 additional districts were invited before four had agreed. The ratio of invited/participating units was thus 19/17 from the CSH context and 27/17 from the comparison context. Re-organizations were ongoing in many of the approached municipalities, which was the main reason for declining at the municipal level and more frequently occurring in the recruitment process for the comparison group.

An information meeting was held at each of the approached CSH units. Prospective participants received both oral and written information including details about the project, what participation would entail and their rights to withdraw participation at any time without giving any reason. The same information was provided to those who had ordinary housing with outreach support. All who agreed to participate, 155 from CSH and 111 in the comparison group, provided their written informed consent.

Data collection

A background questionnaire was devised for the study addressing age, gender, having a friend, seeing a friend, educational level, civil status, having children, perceived physical and psychological problems, and self-reported diagnosis. A set of instruments to address occupation, social contacts, and well-being was used as well, as specified below.

Occupational factors

Four aspects of occupation were addressed in this study, namely occupational engagement, activity level and two estimates of satisfaction with everyday occupations. Occupational engagement was evaluated by the nine-item Profile of Engagement in people with Severe Mental Illness (POES) scale (Bejerholm, Hansson, & Eklund, 2006; Bejerholm & Lundgren-Nilsson, 2015). It uses four response alternatives ranging from less (=1) to more (=4) engaged in occupation. The rating is performed by an occupational therapist, based on a dialogue with the participant about a just completed time-use diary, charting a typical day in the participant’s life. The nine items rated in the instrument concern, e.g., extent of meaningful occupations, rhythm of activity and rest, social environment, variety of occupations, social interplay, and initiating performance. The instrument has shown good inter-rated reliability and internal consistency (Bejerholm et al., 2006) and satisfactory internal and external construct validity (Bejerholm & Lundgren-Nilsson, 2015).

Activity level was estimated with the Satisfaction with Daily Occupations (SDO) instrument (Eklund, Bäckström, & Eakman, 2014). It is composed of 13-14 items, in this study 14, addressing common occupations of everyday life, including work, leisure occupations, home chores and self-care. Each item has two parts. The first part asks whether or not the participant presently performs the occupation or not (yes = 1, no = 0). The number of affirmative responses are summed into an activity level score.

Satisfaction with everyday occupations is an index based on the second part of the SDO items. Whether the participants presently perform the occupation or not, they rate their satisfaction with having/not having that occupation on their present repertoire. The SDO has shown very good internal consistency, adequate construct validity and test-retest stability (Eklund et al., 2014; Eklund & Gunnarsson, 2007; Wästberg, Persson, & Eklund, 2016).

General occupational satisfaction is a one-item estimate of satisfaction with daily occupations, often used to validate the SDO (Eklund et al., 2014) but in the current study used in its own right as an additional aspect of occupation. The wording is “In general, how satisfied are you with your day to day occupations?” A five-point rating scale is used, where a higher rating indicates better satisfaction.

Five items regarding occupational balance have been added to the SDO. Occupational balance may be defined as a personally meaningful pattern of everyday occupations (Christiansen, 1996). These five items ask the respondent whether he or she does too little, too much or just enough of a certain type of occupation. This version of the instrument is termed the SDO-OB (where OB stands for occupational balance). Good construct validity has been shown for each of the five items (Eklund & Argentzell, 2016). Four of the SDO-OB items address specific aspects of everyday occupation (work, leisure, home management, and self-care) and the fifth is a general rating of occupational balance. Only the general rating was used in the current study in order to restrict the number of items. The response scale has five scoring alternatives, from -2 (far too little) to 2 (far too much) where zero denotes just enough of the occupation and a good occupational balance.

Social contacts

Variables to reflect social contacts concerned the housing unit’s possibilities for providing these, together with a few questions from the background questionnaire. The housing unit’s possibilities to provide social contacts is one of four subscales in the Perceived Meaning in Activity – Housing (PMA-H), the others being opportunities for activity, developing as a person, and organization and information (Eklund & Brunt, 2017). The social contacts subscale has 12 items, which consist of statements preceded by a reference to the housing context (e.g., [My housing contributes to that I] have someone I can confide in). The participants respond on a five-point scale from agree (=5) to disagree (=1). The instrument as a whole, as well as all subscales, has shown very good internal consistency and satisfactory construct validity (Eklund & Brunt, 2017). The items from the background questionnaire used to indicate social contacts were: having a friend, seen a friend in the last week, and going to a day center.

Well-being factors

The well-being factors concerned self-rated health, life satisfaction, self-mastery, and recovery. Self-rated health was captured by three questions from SF-36 (Ware & Sherbourne, 1992). The questions concerned general health, health compared to last year, and health compared to others of the same age. The rating scale ranges from 1–5 for each item with a lower value indicating better health. The items were set to form an index in the current study, and internal consistency based on the CSH group was α=0.67.

Life satisfaction was assessed by the first item in the Manchester Short-Assessment (MANSA) of quality of life, which addresses general life satisfaction (Priebe, Huxley, Knight, & Evans, 1999). The Swedish version was used (Björkman & Svensson, 2005). We refrained from using the other items, which target satisfaction with various life domains, since the participants were expected to easily tire. Several researchers have defended the use of a single item for obtaining satisfaction ratings, including quality of life (Bowling, 2005; Cheung & Lucas, 2014).

Self-mastery, defined as the extent to which the participant feels in control of important aspects of his or her life (Pearlin, Menaghan, Lieberman, & Mullan, 1981), was measured with the Swedish version of the Pearlin mastery scale (Eklund, Erlandsson, & Hagell, 2012). Based on seven statements (e.g., I do not have much to say about my life), the participants rate how strongly they agree or disagree on a four-point scale. Two items are reversed and a higher value indicates a higher level of self-mastery. Both the original version and the Swedish version have shown excellent psychometric properties (Eklund, Erlandsson, et al., 2012; Pearlin et al., 1981).

Personal recovery was seen as another facet of well-being (Keetharuth et al., 2018). The Questionnaire on Personal Recovery (QPR) (Law, Neil, Dunn, & Morrison, 2014; Neil et al., 2009), which focuses on intrapersonal and interpersonal factors of relevance for recovery, was used. It consists of statements to which the participant agrees (=5) or disagrees (1) and a higher score indicates a higher level of personal recovery. A Swedish version has shown excellent internal consistency and good construct validity (Argentzell, Hultqvist, Neil, & Eklund, 2017). An abbreviated version with seven items was used for the current study. It was developed for use with participants who have severe psychiatric disabilities and easily tire and has shown good internal consistency (α=0.82) and construct validity (unpublished data).

Clinical variables

Diagnosis was based on data from the background questionnaire. The information on perceived problems and self-reported diagnoses constituted the basis for ICD-10 diagnoses (WHO, 1993), made by a specialized psychiatrist.

The Global assessment of functioning (GAF) scale (Endicott, Spitzer, Fleiss, & Cohen, 1976) is performed by a professional and addresses symptom severity and psychosocial functioning. Each rating is made according to a 100-point scale where a score of 100 denotes a perfect state of mental health and psychosocial functioning and 80 is often used as a cutoff for a healthy state. The GAF is widely used and is regarded as meaningful by a variety of professionals involved in mental health care (Gold, 2014).

Procedure

Research assistants (n = 10), who were occupational therapists with experience of working with the target group and not part of the research team, received training in using the selected instruments. Inter-rater reliability was ensured for the GAF by using videos of fictive cases and calibrating all research assistants against an experienced GAF rater. The research assistants then contacted prospective participants and booked individual meetings in the participants’ homes or a secluded place nearby where the participants could feel safe. Breaks were inserted in relation to the participants’ needs. The research assistants facilitated the participants’ completion of the instruments by explaining the items or helping to complete the forms if the participants requested so, while carefully avoiding influencing the participants’ responses.

Data analysis

The instruments used produced ordinal data, and non-parametric statistical methods were thus employed (Altman, 1993). Spearman correlations were used to calculate associations between variables. The limits for estimating strengths of correlations followed recommendations by Wampold (2001), suggesting the correlations <0.30 are weak, between 0.30 and 0.50 are moderate, and >0.50 are strong. The Fisher r-to-z transformation was applied to compare the sizes of correlation coefficients. Group comparisons were analyzed by the Mann-Whitney U-test or the Kruskal-Wallis test. Logistic regression analyses were performed in order to investigate whether the occupational and social factors could predict the different facets of well-being, while also taking the influence of clinical and socio-demographic factors into account. These regressions were based solely on the CSH group. Each well-being factor was set as the dependent factor in a logistic regression analysis, forward conditional model, where the dependent variable was dichotomized according to the median. All occupational, social, clinical, and sociodemographic factors that showed a statistically significant bivariate association with each well-being variable were entered as independent variables.

The p-value for statistical significance was set at p < 0.05. The software used was the SPSS Statistics 24 core system for all analyses but the Fisher r-to-z transformation, for which an -algorithm retrievable from the Internet was used (Lowry, n.d.).

Results

Description of the participants and their everyday occupations

Details about the participants’ socio-demographic characteristics are displayed in Table 1. A greater number were male and almost all were single. They were mainly born in Sweden and the vast majority reported having a friend. The most common education level was completed nine-year school, while almost 40% had also completed high school. Psychosis was the most common self-reported diagnosis. The level of symptom severity was higher and the level of psychosocial functioning was lower for the CSH group than for the comparison group. This is a logical difference given that the support needs are greater in the former group.

Table 1. Socio-demographic characteristics

Characteristics	CSH (N = 155)	OHS (N = 111)	P-value
Mean age (SD)	48 (12)	46 (11)	ns.
Proportion of women	42%	63%	0.001
Married/cohabiting	1.3%	10.3%	<0.001
Has children	26%	41%	0.007
Has a friend	89%	82%	ns.
Sees a friend	63%	68%	ns.
Born in Sweden	85%	83%	ns.
Education			<0.001
Non-completed 9-year school	6.5%	1.9%
Completed 9-year school	46%	25%
Completed high school	38%	49%
Completed college/university	10%	25%
Years in current accommodation (SD)	6.3 (5.5)	11 (8.3)	<0.001
Attends a day center	47%	54%	ns.
Reports having a psychiatric diagnosis	98%	99%	ns.
Diagnostic category			<0.001
Psychosis	62%	32%
Anxiety/mood disorders	13%	27%
Other (usually Asperger syndrome, attention deficit disorder or unspecified mental disorder)	25%	41%
GAF symptoms (SD)	43.6 (13.9)	53.9 (8.9)	<0.001
GAF functioning (SD)	42.8 (11.1)	51 (10.8)	<0.001

Note. CSH = congregate supported housing; OHS = ordinary housing with outreach support

Table 2 displays the occupations, based on the SDO items, in which the participants were engaged. The most commonly performed occupations in both groups concerned relaxation, physical exercises, managing personal hygiene, planning and performing household chores, and leisure in terms of hobbies on one’s own and taking part in cultural events. The least common occupations concerned paid work or study.

Table 2. Participants’ involvement in occupations targeted in the Satisfaction with Daily Occupations (SDO) instrument, and their satisfaction scores per item/occupation

Item no.	Percentage (mean satisfaction rating/SD) ^1		Content of the SDO item
	CSH	OHS
1	9% (4.6/2.0)	11% (4.2/2.0)	Employed or admitted to school/college
2	10% (4.5/2.1)	14% (4.1/2.0)	Working or enrolled in studies ^2
3	11% (4.6/2.0)	15% (4.6/1.9)	Attending work training ^2
4	41% (5.2/2.0)	37% (4.9/1.9)	Attending a day center ^2
5	23% (4.8/1.9)	34% (4.2/2.0)	Engaged in organized leisure/hobbies at least once a week ^2
6	59% (5.2/1.8)	70% (4.9/1.9)	Performing leisure occupations/hobbies on one’s own at least once a week ^2
7	79% (5.4/1.5)	79% (5.1/1.8)	Taking part in cultural occupations at least once a week ^2
8	83% (5.2/1.8)	72% (4.2/1.8)	Doing household work, such as cleaning and cooking, almost daily ^2
9	20% (5.0/1.8)	37% (4.4/1.9)	Doing repairs and/or gardening ^2
10	73% (6.0/1.4)	75% (4.6/1.8)	Organizing and planning the household work ^2
11	18% (5.3/1.5)	47% (5.0/1.6)	Taking care of a close persons at least once a week ^2
12	87% (5.7/1.4)	87% (5.1/2.0)	Managing own personal hygiene on a daily basis
13	88% (5.2/1.7)	88% (4.5/2.1)	Doing physical exercises at least once a week ^2
14	79% (5.2/1.7)	79% (4.4/2.0)	Doing activities to relax at least once a week ^2

Note. CSH = congregate supported housing; OHS = ordinary housing with outreach support.

¹ The share of patients presently performing each type of occupation. The corresponding mean satisfaction rating, with SD, in parenthesis. Possible score range for the SDO satisfaction items is 1–7

² Concerns the situation the past 2 months.

Associations between occupational factors and well-being

Associations between the occupational factors and health, quality of life, and personal recovery are shown in Table 3. General occupational satisfaction was the occupational factor with the most consistent relationships with all aspects of well-being in both groups. There were fewer statistically significant associations between an occupational factor and a well-being factor in the CSH group than in the comparison group with ordinary housing, and most of them were weak.

Table 3. Associations between occupational and well-being factors in the two groups, and statistically significant differences in correlation coefficients indicated in bold

Occupational factors		Self-rated health^1	Life satisfaction	Self-mastery	Recovery
Occupational engagement	CSH	rs=-0.16 ns.	rs=0.12 ns.	rs=0.22 p = 0.016	rs=0.1 ns.
	OHS	rs=-0.09 ns.	rs=0.27 p = 0.004	rs=0.25 p = 0.008	rs=0.01 ns.
Activity level	CSH	rs=-0.08 ns.	rs=-0 . 11 ns.	rs=0.16 ns.	rs=0 . 07 ns.
	OHS	rs=-0.21 p = 0.032	rs=0 . 33 p = 0 . 001	rs=0.3 p = 0.002	rs=0 . 52 p < 0 . 001
Satisfaction with daily occupations index	CSH	rs=-0.25 p = 0.006	rs=0.4 p < 0.001	rs=0.32 p < 0.001	rs=0 . 23 p = 0 . 014
	OHS	rs=-0.28 p = 0.003	rs=0.32 p = 0.001	rs=0.21 p = 0.026	rs=0 . 52 p < 0 . 001
General occupational satisfaction^2	CSH	rs=0.37 p < 0.001	rs=-0 . 34 p < 0 . 001	rs=-0 . 21 p = 0 . 021	rs=-0.2 p = 0.033
	OHS	rs=0.45 p < 0.001	rs=-0 . 62 p < 0 . 001	rs=-0 . 6 p < 0 . 001	rs=-0.35 p < 0.001
Occupational balance	CSH	rs=-0 . 05 ns.	rs=0 . 08 ns.	rs=0 . 12 ns.	rs=0.06 ns.
	OHS	rs=-0 . 34 p < 0 . 001	rs=0 . 42 p < 0 . 001	rs=0 . 48 p < 0 . 001	rs=0.21 p = 0.027

Note. CSH = congregate supported housing; OHS = ordinary housing with outreach support; p-values for statistically significant differences between the two groups’ correlation coefficients (indicated in bold) range between 0.01 and <0.001.

¹A lower value denotes better health.

²A lower value denotes better satisfaction.

Group differences in strength of occupation–well-being correlations

In all cases where there was a statistically significant difference between the sizes of the correlations (bold figures in Table 3), the stronger association appeared in the comparison group. Such differences pertained particularly to the well-being factors of life satisfaction, self-mastery, and recovery. The occupational factor that showed most group differences, in terms of the strength of associations with well-being factors, was occupational balance. The non-significant associations between occupational balance and all aspects of well-being for those who lived in CSH indicate that they could be under-occupied but still perceive high levels of well-being. The statistically significant associations in the comparison group indicate that this was not the case for that group. Several of the correlations between occupational balance and well-being were moderate to strong in that group, and the sizes of associations between occupation and well-being differed significantly between the two groups. On the other hand, the groups did not differ on correlations between occupational engagement and any of the well-being factors.

Associations between social contacts and well-being

No associations at all were found regarding the variable addressing the possibilities for social contacts in the home environment and any aspect of well-being. This was true for both groups and the same was found for the variable “having a friend”. No differences on well-being factors were found between those who had and those who had not “seen a friend during the past week” in the CSH group. In the comparison group, however, those who had recently seen a friend showed better self-rated health (p = 0.016) and scored higher on life satisfaction (p = 0.027), self-mastery (p = 0.037), and recovery (p = 0.032).

Those who attended a day center in the CSH group rated their health better than those who did not participate in day center services (p = 0.042). In the comparison group, differences between those who attended a day center and those who did not concerned better self-rated health (p = 0.043) and better life satisfaction (p = 0.023) for those who attended. Since all variables addressing social contacts were categorical, it was not possible to show strengths of relationships or perform Fisher r-to-z transformations.

Clinical factors in relation to well-being

To account for the influence of clinical and sociodemographic factors on the relationship between occupation and health, these factors were analyzed for associations with the well-being factors. The clinical factors concerned diagnosis, severity of symptoms, and psychosocial functioning. Diagnosis was not related with any aspect of well-being in the CSH group, whereas differences based on diagnostic group (see Table 1) were found for self-rated health (p = 0.013), life satisfaction (p = 0.007) and self-mastery (p = 0.014) in the comparison group. Regarding all of these aspects, those with a psychosis diagnosis rated their well-being better as compared to those with an anxiety/depressive disorder. Participants with psychosis also rated their life satisfaction and self-mastery better than those who had a diagnosis categorized as “other”.

The only statistically significant association between psychiatric symptoms and a well-being factor found in the CSH group showed that having less severe symptoms was related with a higher level of self-mastery. Level of psychosocial functioning was positively related with self-rated health and self-mastery in that group but not with any other well-being factor. There was also a statistically significant association in the comparison group regarding symptoms and, in their case, less severe symptoms were related with higher ratings of recovery. Focusing on psychosocial functioning in that group, positive relationships were found with life satisfaction, self-mastery, and recovery. These correlations between clinical factors and well-being were mainly weak in both groups (Table 4).

Table 4. Associations from clinical factors and age to well-being factors in the two groups, and differences in correlation coefficients

		Self-rated health^1	Life satisfaction	Self-mastery	Recovery
Severity of psychiatric symptoms	CSH	rs=-0.12 ns.	rs=0.1 ns.	rs=0.31 p = 0.001	rs=0.08 ns.
	OHS	rs=-0.05 ns.	rs=0.13 ns.	rs=0.14 ns.	rs=0.29 p = 0.002
Level of psychosocial functioning	CSH	rs=0.2 p = 0.027	rs=-0.11 ns.	rs=0.29 p = 0.001	rs=0.13 ns.
	OHS	rs=-0.09 ns.	rs=0.21 p = 0.025	rs=0.25 p = 0.008	rs=0.29 p = 0.002
Age	CSH	rs=-0.07 ns.	rs=0.06 ns.	rs=-0.06 ns.	rs=-0 . 05 ns.
	OHS	rs=-0.20 p = 0.034	rs=0.21 p = 0.025	rs=0.25 p = 0.008	rs=0 . 36 p < 0 . 001

Note. CSH = congregate supported housing; OHS = ordinary housing with outreach support; a statistically significant difference (p < 0.001) between the two groups’ correlation coefficients is indicated in bold.

¹A lower value denotes better health.

Group differences in strength of correlations between clinical and well-being factors

Severity of psychiatric symptoms and level of psychosocial functioning were ordinal data, enabling comparisons of correlation coefficients. None of the correlations differed in size between the two groups, as indicated in Table 4 by non-bold coefficients.

Socio-demographic factors in relation to well-being

The investigated socio-demographic variables were age, gender, education, civil status, and having children. Age was not related to any of the well-being aspects in the CSH group, but in the comparison group older age was weakly associated with better self-rated health (r_s=-0.2, 0.034) and self-mastery (r_s=0.24, p = 0.011). With respect to gender, males rated their life satisfaction higher compared to females in the CSH group (p = 0.038). No other gender differences for well-being were found in that group, and no gender differences at all were found in the comparison group. Civil status was not of importance to well-being in the CSH group, but in the comparison group those who lived with a partner rated their life satisfaction higher (p = 0.045). Educational level was not at all related with well-being in any of the groups, nor was having children.

Group differences in strength of correlations between age and well-being factors

All socio-demographic variables except for age were categorical, thus Fisher r-to-z transformation was feasible only for age. The correlation between age and recovery was stronger in the comparison than in the CSH group, as indicated by bold coefficients in Table 4.

Multivariate analyses of associations between occupation and well-being

Occupational, social, clinical, and sociodemographic variables showing statistically significant associations with the well-being variables in the bivariate analyses were entered in logistic regression analyses, presented in Table 5. The independent variables were self-rated health, general occupational satisfaction, satisfaction with daily occupations index, day center attendance, and psychosocial functioning. The satisfaction with daily occupations index became the only significant predictor, explaining 9%. For each increased scale step on satisfaction, the likelihood of belonging to the high group on health increased by 4%. The model predicted 67% of the sample correctly. For life satisfaction, four variables became statistically significant in the bivariate analyses – general occupational satisfaction, activity level, the satisfaction with daily occupations index, and gender. General occupational satisfaction and gender became significant predictors in the regression analysis, together explaining 14% of the variation. The likelihood of belonging to the high group on life satisfaction increased by 73% for each scale step towards better general occupational satisfaction and was more than doubled by male gender compared to females. The model predicted 66% of the sample correctly.

Table 5. Factors predicting well-being factors among people in CSH

Well-being factors, significant predictors and explained variance (%; Nagelkerke R Square)	Wald	P-value	Odds ratio	95% CI
Self-rated health (9%) Satisfaction with daily occupations index	7.82	0.005	1.04	1.01–1.07
Life satisfaction (14%) General occupational satisfaction Gender	7.72 5.82	0.006 0.016	1.73 2.55	1.18–2.55 1.19–5.4
Self-mastery (20%) Symptom severity Satisfaction with daily occupations index	12.59 5.82	<0.001 0.012	1.06 1.04	1.03–1.1 1.01–1.07
Recovery (4%) Symptom severity	3.88	0.049	1.03	1.0–1.05

Note. CSH = congregate supported housing.

Regarding the regression model for self-mastery, five variables were entered – occupational engagement, general occupational satisfaction, the satisfaction with daily occupations index, symptom severity, and psychosocial functioning. Symptom severity and the satisfaction with daily occupations index became significant predictors. The chances of being in the high group increased by 6% for each scale step towards less severe symptoms and by 4% for each step towards better satisfaction with daily occupations. These variables together explained 20% of the variation in self-mastery. The model predicted 69% of the sample correctly.

Three variables showed statistically significant bivariate relationships with recovery and were entered in the regression model – general occupational satisfaction, activity level, and the satisfaction with daily occupations index. The only significant predictor was symptom severity, which explained 4% of the variation in recovery. For each scale step towards less severe symptoms, the chances of belonging to the high group on recovery increased by 3%. The model predicted 58% of the sample correctly. The Hosmer Lemeshow test was non-significant for all models.

Discussion

The major purpose of this study was to investigate relationships between occupation and different aspects of health and well-being in a poorly understood and researched population, namely people with psychiatric disabilities residing in CSH. Several such associations were found. In particular, the general satisfaction with occupations index was related with all aspects of well-being. Activity level and occupational balance were not at all related with well-being in that group. This means that people with a very low activity level, and who considered that they did far too little, could still perceive a relatively high level of well-being, in terms of life satisfaction, personal recovery, etc. It is possible, however, that the lack of relationship between activity level and occupational balance may be due to learned helplessness and a low level of expectation, which is common among people with a history of institutionalization (Chovil, 2005; Hamzaoglu, Ozkan, Ulusoy, & Gokdogan, 2010). Similarly, fear of stigma and rejection from the surrounding society could entail satisfaction with a supportive though restricted environment (Chovil, 2005). Engagement in occupations was also mostly unrelated to the well-being variables, particularly in the CSH group. This may partly be due to the fact that occupational engagement was rated by a research assistant, whereas the well-being factors were self-reports. It is well-recognized that different rater perspectives reflect divergent phenomena (Streiner & Norman, 2008).

Nevertheless, the fact that people in CSH may be pleased with a low-active everyday life is important to consider in the planning of housing support. Being amongst staff and fellow residents, and doing occupations with low demands on themselves, may be a sufficient level of activity to maintain perceived well-being. This finding most probably reinforces the importance of offering a range of quiet to more active occupational choices to housing residents and underscores for occupational science the idiosyncratic nature of occupations for persons with serious mental illnesses. Involvement in quiet occupations, seemingly just sitting and watching others while still feeling engaged in something, can be an important state in the recovery process (Bejerholm & Eklund, 2004; Sutton et al., 2012). A phase characterized by detachment, non-demanding occupations and no pre-planned social expectations may prepare for sparking hope and re-engaging in occupations again (Borg & Davidson, 2008; Doroud, Fossey, & Fortune, 2015; Sutton et al., 2012). The current findings suggest that individuals in the CSH group were in this initial state of recovery and indicate that the two groups studied here may need different approaches to support their recovery process.

An important aspect of the findings was that the associations between occupation and health were generally stronger in the comparison group, as indicated by statistically significant differences between the correlation coefficients (bold figures in Table 3). Associations between occupation and health were thus more straightforward in the group with less severe disabilities. This finding is interesting from an occupational science perspective. The consistent links between aspects of occupation and well-being in the comparison group are in line with other research (Aubin et al., 1999; Bejerholm & Eklund, 2007; Eklund & Leufstadius, 2007). The comparison group seemed more engaged in leisure, doing repairs, and taking care of someone else (cf. Table 2). Such involvement in community life has been found to be closely related to the process of recovery (Doroud et al., 2015). This may at least partly explain the stronger occupation–well-being relationships found in that group. The finding from a previous study that social interaction was more important than occupation for well-being, in terms of meaning in life (Eklund, Hermansson, et al., 2012), did not receive any support in the current study. This calls for further research, particularly since relationships amongst occupation, social interaction, and well-being for persons with psychiatric disability seem to be under-researched, as evidenced by few hits in our systematic literature search.

Focusing on the role of social contacts, only one association was found in the CSH group; participating in a day center was related with better self-rated health. A similar finding was obtained for the comparison group, where life satisfaction was also positively associated with day center participation. One could reason that these relationships were mutual, that the more healthy participants chose to visit day centers, and that their engagement there strengthened their subjective health. In addition to these findings, having seen a friend recently was consistently linked with well-being in the comparison group. Frequently experiencing loneliness has been strongly associated with low mental well-being, poor mental health, and long-term stress, depression, and anxiety (Kearns et al., 2015). The fact that no such association appeared in the CSH group may have to do with their greater access to emotional and practical support.

This contrasting of occupation–well-being relationships in the CSH group compared to the comparison group must be considered from an environmental perspective. A recent systematic review addressed the outcomes of supported housing services in terms of psychosocial and mental health outcomes. The findings showed inconsistent relationships between type of housing and social interaction/functioning and well-being, but a majority of studies saw no change in these factors when people moved to a less restricted housing context. However, a reduction over time, such as diminished social networks and deterioration in quality of life, was observed when people moved to restricted environments such as nursing homes (McPherson, Krotofil, & Killaspy, 2018). Other research has indicated that housing support with a more intensive level of care entails a lower level of autonomy but better quality of life (Killaspy et al., 2016). Thus, although no clear-cut evidence exists regarding the influence of the housing context on well-being and social interaction, it seems reasonable to assume that the relationship differences found between the studied groups were accentuated by dissimilarities in the environment.

The clinical factors were treated as confounders in this study and the statistically significant ones were thus entered into the respective regression analyses performed on the OHS group, together with the significant occupational and social contacts factors. Symptom severity became the most important factor for explaining self-mastery and personal recovery. Symptom severity was the only factor contributing to personal recovery, which was somewhat surprising considering research exploring the factors that lie behind personal recovery, including social contacts and participating in everyday life (Le Boutillier et al., 2011; Leamy, Bird, Le Boutillier, Williams, & Slade, 2011). On the other hand, severity of symptoms explained only 4% of the variation and the current study thus failed to explain the essence of personal recovery.

Gender was of some importance and became the second explaining factor for life satisfaction. That women exhibit worse life satisfaction and quality of life than men is a common finding (Eklund et al., 2001; Love-Koh, Asaria, Cookson, & Griffin, 2015; Röder-Wanner, Oliver, & Priebe, 1997). A negative relationship between depression and quality of life, and that women tend to perceive more depressive symptoms than men, have been used as an explanation for women’s worse quality of life (Röder-Wanner et al., 1997). Nevertheless, women’s needs and quality of life deserve further exploration, since a greater disposition for depression may be related to unrevealed potential intersections of poverty, violence, caregiving, and other structural societal factors.

An occupational satisfaction factor was the main contributor to both self-rated health and life satisfaction and contributed to self-mastery. Actual doing, in terms of activity level or occupational engagement, did not become significant in the regression analyses, nor did any aspect of social contacts. This was unexpected in relation to previous findings of the importance of social contacts for meaning in life (Eklund, Hermansson, et al., 2012) and mental health (Kearns et al., 2015), but in agreement with other research addressing social network and well-being (Hultqvist, Markström, Tjörnstrand, & Eklund, 2018). Methodological aspects, further discussed below, may play a role here.

Methodological concerns

One cannot exclude that the fewer and weaker associations between occupation and well-being in the CSH group were due to a lack of validity in the data collected. Several measures were taken, however, to strengthen the internal validity of the study – using short forms of instruments, inserting breaks to avoid exhaustion, explaining the items when warranted and assisting in completing the questionnaires. Although short forms are recommended for use with people with psychiatric disabilities to avoid strain or discomfort (Mausbach, Moore, Bowie, Cardenas, & Patterson, 2009), use of one-item measures sometimes causes debate. Several renowned researchers have concluded, however, that one item can be as reliable, and more effective, compared to a longer scale (Bowling, 2005; Cheung & Lucas, 2014). We therefore believe the data for the current study can be regarded as trustworthy.

The way social contacts were measured may also be discussed. The study mainly addressed tangible aspects such as seeing friends, possibilities for social contacts in the housing context, and participation in day centers, while the quality of social contacts was not tackled properly. Furthermore, occupational balance was addressed by the SDO-OB, which adopts a time-allocation perspective. Another instrument, reflecting a more personalized and idiosyncratic view of balance, might have resulted in stronger associations.

It should also be acknowledged that the division into occupational, social, and socio-demographic factors was not altogether clear-cut. Socio-demographics such as having children and civil status have a social contact dimension to them, and attending a day center, regarded as an indicator of social contacts, also encompasses an occupational dimension. Nevertheless, these issues have more to do with the framing of variables and would not jeopardize the study’s internal validity. It may also be a general problem in the study of occupational science that the constructs that are important are not mutually exclusive.

Furthermore, a selection bias at that local level is possible. We did not have full insight into how managers at the district level recruited settings. We had an agreement on seeking variation on the selected setting characteristics, however, and nothing indicated the managers had not adhered to that.

Finally, the study is based on many variables and many computations, which infers a risk of mass significance. This would be defendable against the explorative character of the study, however, and the findings may be seen as hypothesis generating for future studies.

Conclusion

The study findings provide important insights for occupational science, informing us that a high activity level is not necessary for well-being, and reinforces previous study findings that well-being is idiosyncratic and clinicians should be wary of providing occupations that they associate with a normal life. Although associations between occupation and well-being were identified in the group residing in CSH, such links were fewer and weaker seen against the comparison group and against previous research. This indicates that people with a lower level of psychosocial functioning tend to have another pattern of associations and that the role of occupation for well-being, although still of relevance, must not be overemphasized in that group. So-called quiet occupations may be sufficient for them to feel that they engage and participate. It is important to avoid placing value on a specific set of occupations associated with a “normal life”. Quiet occupations and detachment can form an important state in a recovery process and benefit re-engagement at a later stage. The findings should be replicated, however, to expand the knowledge base within occupational science. Delving deeper into social contacts is another important target for further research, as well as the influences of learned helplessness and fear of stigma and rejection on the relationship between occupation and well-being. This could inform both occupational science and occupational therapy. How individually performed occupations, versus occupations performed with others, are associated with well-being would be another target for future exploration. Finally, more narrative research is needed to get the voices of people residing in CSH.

Acknowledgement

The study was funded by the Swedish Research Council for Health, Working Life and Welfare under Grant 2014-4488.

Disclosure statement

The authors declare they have no conflict of interest.

ORCID

Mona Eklund http://orcid.org/0000-0002-0136-3079

Carina Tjörnstrand http://orcid.org/0000-0001-6018-0777