Well-being in elderly long-term care residents with chronic mental disorder: a systematic review

ABSTRACT

Objectives: One of the most important objectives of care for older long-term care residents with chronic mental disorders is to facilitate well-being. This review provides an overview of research literature on well-being in this population.

Method: A systematic review was conducted using Pubmed, PsycINFO and PsycARTICLES for all studies up until March 2016. Three reviewers independently assessed the eligibility of the publications and made a selection.

Results: From a total of 720 unique search results, ten studies were deemed eligible. Specialized care, specifically the presence of mental health-workers was associated with increased well-being outcomes. Perceived amount of personal freedom was also related to higher well-being, whereas stigmatization and depression were related to reduced well-being. Size of residence, single or group-accommodation or moving to another locationdid not, however, seem to have an impact on well-being.

Conclusion: Specialized care, aimed at psychiatric disorders and extra attention for depressed residents are useful tools to promote well-being. Additionally, themes like personal freedom and stigmatization should be taken into consideration in the care for older long-term care residents with chronic mental disorder. However, as very little research has been conducted on this topic, conclusions should be interpreted with caution. More research is highly desirable.

KEYWORDS:

• Psychiatric disorders
• elderly
• residential facilities
• long-term care
• well-being

Introduction

Long-term care (LTC) facilities are confronted with an increased number of clients with very complex care demands (Hamers, 2011). This is due to both the ageing population and the trend in Western society to keep older people in their own home for as long as possible (Geller, Guzofski, & Lauterbach, 2008). Older people with chronic mental disorders are among those in need of complex care, since they require a combination of psychiatric and medical support. This group, also referred to as gerontopsychiatric patients (Luzny & Ivanova, 2009; Ponte, Almeida, & Fernandes, 2014) consists of patients of (functional) old age with age-related limitations, and serious psychiatric disorders other than dementia. In this respect they differ from psychogeriatric nursing home residents, for whom dementia is the main reason for nursing home care.

The prevalence of psychiatric disorders is relatively high among LTC residents. In their systematic review Seitz, Purandare, and Conn (2010) found that major depressive disorder occurs worldwide in 5% to 25% (with a median of 10%) of older LTC residents. In another review prevalence of anxiety disorders in nursing homes varied from 3.2% to 20.0% (Creighton, Davison, & Kissane, 2016). Data on other psychiatric disorders among the elderly is scarce and inconclusive, however, according to the National Nursing Home Survey of 2004 there is a 3.6% prevalence of schizophrenia and a 1.5% prevalence of bipolar disorders in USA nursing homes (Seitz et al., 2010).

Attainment and maintenance of well-being or quality of life is one of the most important aims in the care for elderly LTC residents, and thus for the institutionalized gerontopsychiatric population. This is increasingly recognized, both by institutions and by the authorities in Western society, where the biomedical model used to be the predominant model for treatment. Medical care for elderly LTC residents is now more and more focused on enhancing well-being (Brownie & Nancarrow, 2013; Hamers, 2011; Koren, 2010).

If well-being is considered to be one of the main concerns in care, it is important to understand what it is, and how it can be achieved. Well-being is a broad and abstract construct (Diener, Suh, Lucas, & Smith, 1999). In this review, well-being is regarded as a positive judgment or feelings concerning one's life (Dolan, Peasgood, & White, 2008). This definition is deliberately broad and thus allows for literature from different theoretical perspectives.

The level of well-being is often considered to be dependent on the availability of certain determinants or resources (Dodge, Daly, Huyton, & Sanders, 2012; Dolan et al., 2008; Hobfoll, 2002; Ormel, Lindenberg, Steverink, & Verbrugge, 1999). Pinquart and Sörensen (2000) concluded in their meta-analysis on well-being among the elderly, that socioeconomic status (the result of income and employment status) is related to well-being, as is the existence of high-quality social ties. Self-perceived health, functional status (Cummings, 2002) and marital status (Bilgili & Arpacı, 2014) are also found to be related to life satisfaction or well-being.

Gerontopsychiatric LTC residents tend to fall behind when it comes to the aforementioned determinants. In general, their health is poor. Furthermore, Van den Brink, Gerritsen, Voshaar, and Koopmans (2013) have found that these residents are more often unmarried, younger, and have a higher incidence of problem behavior and cognitive impairment than other LTC residents. These characteristics (plus the psychiatric disorder itself) make gerontopsychiatric LTC residents vulnerable to diminished well-being.

An overview of the knowledge on this subject could help care facilities to pay greater attention to well-being in the gerontopsychiatric population, and to highlight the most effective approaches to promote well-being. However, currently the literature on the relationship between different determinants and well-being in this group is scarce, and difficult to find, due to many differences in terminology. Additionally, for the different types of care setting there is tremendous variation in used terms and forms. This systematic review aims to present an overview of all determinants or resources that are found to be related to the level of well-being in gerontopsychiatric LTC residents. In this way it can offer directions for further research, and provide care institutions with a knowledgebase on how to improve well-being in this population.

Method

Search strategy

The systematic literature search was conducted in three databases: PsycINFO, PsycARTICLES and Pubmed. Articles on the well-being or the quality of life of gerontopsychiatric LTC residents, published in the period up to March 2016 were searched. Titles and abstracts were scanned for the following words: (‘well-being’ OR ‘quality of life’ OR wellbeing), (psychiatr* OR schizophren* OR ‘mental health’ OR ‘mental disorder’ OR ‘mental illness’ OR bipolar OR depress* OR psychot* OR psychos* OR anxiety OR autism* OR schizoaffect* OR geropsychiatr* OR gerontopsychiatr* OR ‘double care demanding’), (elderly OR ageing OR aging OR old* OR geriatr* OR aged OR senior), (inpatient* OR ‘nursing home’ OR intramural OR nursery OR ‘residential care’ OR ‘long-term care’ OR facilit* OR ‘elder care home*’ OR ‘residential home*’ OR hospital*). Articles with the following words in title and/or abstract were excluded: cancer, parent*, HIV, MS, HRQOL, dement*, alzheimer*. Exclusion words were based on irrelevant themes that occurred frequently in the search results. This resulted in 1008 hits. Duplicates were removed, which left a total of 720 articles.

Literature selection

Three authors (EvdW, SvH, and WW) independently screened the titles and abstracts of the remaining 720 articles which were subsequently filtered under the following criteria: subjects were diagnosed with psychiatric disorders (excluding primary diagnoses of dementia or mental retardation), subjects were LTC residents, they were aged 40 and older, well-being or quality of life was measured, studies reported original research data and the full text was written in English or Dutch. Disagreement on selection, which occurred in 72 of the 720 cases, was resolved by discussion until consensus was reached. A total of 584, references were rejected, based on the title and abstract. The main reasons for rejection were: measures of quality of life or well-being were not included (N = 62), the research population did not consist of gerontopsychiatric care residents (N = 439), or the article was an editorial, review or in another way not-original research (N = 83). For the remaining 136 articles, 124 full texts were retrieved (12 full texts could not be obtained due to unavailability in the databases, and unavailable or outdated contact information on the authors, or no response after several attempts to contact). The first Author (EvdW) read the 124 available articles in full, rescreened the abstracts from the 12 unfound articles and made a subsequent selection based on the criteria mentioned above. Two co-authors (SvH and WW) read a random sample of 40 full-texts and made an independent selection. After primary disagreement in 6 out of 40 cases, full consensus was reached after discussion. The 128 studies that did not meet one or more of the inclusion criteria were rejected. Reasons for exclusion were as follows: participants did not live in residential setting, they were younger than 40 years of age, did not have a diagnosed psychiatric disorder, or well-being was not, or only partly measured (e.g. only health-related well-being). A total of eight articles were retained, and selected using this procedure.

A second route in the search strategy was taken by checking reference-lists in all included articles and in relevant reviews that came up from the first search. References that seemed to meet the aforementioned criteria were located and screened by the first author. Reference lists of four of the articles were additionally screened by the two co-authors (SvH and WW). Again, a consensus was reached. This second route resulted in sourcing two additional relevant articles. A total of 10 articles were included in the review. See Figure 1 for a flowchart of the selection-process.

Figure 1. Flow chart showing the selection process.

Methodological quality

The methodological quality of the included studies was appraised using two checklists. The first checklist is a guideline by the CBO, a former Dutch institute for health care improvement as published in Collet, De Vugt, Verhey, and Schols (2010), which was used to evaluate the only experimental study in this review (Cooper & Pearce, 1996). All other included studies were not experimental and were thus evaluated using another checklist compiled by Van der Windt, Zeegers, Kemper, Assendelft, and Scholten (2000). Outcomes of the checklists are found in Tables 1 and 2.

Table 1. Methodological quality of observational studies.

Observational studies	Selection criteria	Response rate> 70%	Validity and reliability of determinants	Independence of knowledge about outcome	Validity and reliability of outcomes	Independence of knowledge about determinants?	Adjustment for confounders	Number of participants	Total
Davison et al. (2012)	0.5	?	1	0	1	0	1	0.5	4
Depla, De Graaf, and Heeren (2005)	1	0	1	?	1	?	1	1	5
Depla, De Graaf, Van Weeghel, and Heeren (2005)	1	?	0.5	?	1	?	1	1	4.5
Depla et al. (2006)	1	0	0.5	?	1	?	1	0.5	4
Kallert et al. (2007)	0.5	1	0.5	?	0.5	?	0,5	0.5	3.5
Leiße and Kallert (2000)	0.5	1	0.5	?	0.5	?	0	1	3.5
Nakagawa and Hayashi (2013)	0.5	?	1	?	1	?	0	0.5	3
Smalbrugge et al. (2006)	0.5	0	0.5	?	1	?	0,5	1	3.5
Luzny and Ivanova (2009)	0.5	0	0.5	0	1	0	0	0.5	2.5

Note: 1: Yes; 0,5: Partly; 0: No; ?: Unknown.

Table 2. Methodological quality of experimental studies.

Experimental study	Randomized	Allocation concealed	Comparable baseline characteristics	Blinded providers/participants	Blinded outcome assessors	Attrition rate reported	Intention to treat analysis	Validation of tools	Total
Cooper and Pearce (1996)	Yes	No	Yes	No	No	No	No	Yes	3

Results

A total of 10 studies were selected for this review (Cooper & Pearce, 1996; Davison, McCabe, Knight, & Mellor, 2012; Depla, De Graaf, & Heeren, 2005; Depla, De Graaf, & Heeren, 2006; Depla, De Graaf, Van Weeghel, & Heeren, 2005; Kallert, Leisse, & Winiecki, 2007; Leisse & Kallert, 2000; Luzny & Ivanova, 2009; Nakagawa & Hayashi, 2013; Smalbrugge et al., 2006). Eight of these studies were cross-sectional studies, and two were longitudinal studies (Table 3).

Table 3. Characteristics of included studies.

Study	Design	Participants	Methodological quality	Setting	Instruments	Factors	Outcomes
Davison et al. (2012)	Cross-sectional	Country: Australia N = 50 Diagnosis: MDD Mean age: 83 (SD = 7.2)	4	Four high-level care facilities (nursing homes), 12 low-level care facilities	(parts of) Ryff multidimensional measure of psychological well-being	Diagnosis	Both diagnosed MDD and Self-rated depression are related to three subscales of well-being: autonomy, environmental mastery and purpose in life.
Depla, De Graaf, and Heeren (2005)	Cross-sectional	Country: The Netherlands N = 96 (RH) N = 78 (HC) Diagnosis: mental disorders, other than dementia Mean age: 76 ( = 6)	4	Residential homes operating a supported living programme for at least one year. And Hospital care	PGCMS and MANSA	Type of Setting	On most subscales, patients in psychiatric hospitals showed higher well-being than patients in residential homes.
Depla, De Graaf, Van Weeghel, and Heeran (2005)	Cross-sectional	Country: The Netherlands N = 66 (RH) N = 65 (HC) Diagnosis: Axis-1 disorders other than dementia Mean age: 74 (SD = 5.9)	3.5	Residential homes operating a supported living programme for at least one year. And Hospital care	MANSA	Stigma Social network and Social activities	In all subgroups (psychotic/non-psychotic; Hospital/ residential home) stigma was associated with lower well-being. Larger social network size was associated with higher well-being in the total sample and in the non-psychotic –hospital subgroup. Social activities were not related to well-being. Only for the psychotic residential home sample there was a negative relation.
Depla et al. (2006)	Cross-sectional	Country: The Netherlands N = 73 Diagnosis: psychotic and non-psychotic axis-1 disorders, other than dementia Mean age: 75/76	3.5	Residential homes operating a supported living programme for at least one year.	PGCMS and MANSA	Type of Setting MHW-staff availability Perceived amount of personal freedom	There was no relation between group-living and single living for both subgroups (psychotic and non-psychotic). MHW-staff availability was related to decreased agitation (subscale) but only for psychotic participants Perceived amount of freedom was related to decreased agitation for non-psychotic participants.
Kallert et al. (2007)	Longitudinal	Country: Germany N = 43 Diagnosis: Chronic schizophrenic patients Mean age: unknown	3	Nursing home area of psychiatric hospital	Structured QoL interview based on Berlin QoL Profile. Items on 8 subjects + 1 general life satisfaction item	Type of setting	Over a two-year period patients in nursing homes showed significant deterioration in QoL and life satisfaction. This effect remained after adjusting for age and gender.
Leiße and Kallert (2000)	Cross-sectional	Country: Germany N = 50 Diagnosis: Chronic schizophrenic patients Mean age: 58.5 (SD = 10.4)	2.5	Nursing home area of psychiatric hospital	Structured QoL interview based on Berlin QoL Profile. Items on eight subjects and one general life satisfaction item	Type of setting	No significant results
Nakagawa and Hayashi (2013)	Cross-sectional	Country: Japan N = 66 Diagnosis: schizophrenia Mean age: 68.0 (SD = 8.0)	2.5	Longstay wards of large psychiatric centre	Quality of Life scale, and the Lancashire QoL profile	Observed vs. self-rated QoL. Demographics: Age, age of onset, age first admission, education, duration and frequency of admissions and neuroleptic daily dose Positive and negative symptoms of Schizophrenia (PANSS) Daily life activities (REHAB)	No significant relation between observed and self-rated QoL Observed QoL (or subscales) correlated with age (neg), education (pos), duration of current hospitalization (neg) and age of onset (pos) Also with PANSS subscales: anergia (neg), activation(neg) and lack of judgement/insight. And with alle REHAB subscales (neg) Self-rated QoL correlated with duration of hospitalization (pos), number of admissions (pos) and (PANSS) depression (neg)
Smalbrugge et al. (2006)	Cross-sectional	Country: The Netherlands N = 350 Diagnosis: anxiety and/or depression Mean age: 79.3 (SD = 8.3)	3.5	Nursing homes	PGCMS	Diagnosis Perceived social support	Patients with depression or combined depression and anxiety had significantly lower well-being than patients with pure anxiety, or no disorder at all. Perceived social support was associated with significantly higher well-being. Gender, age, education level, having a partner, density of urbanization, MMSE score and amount of physical illnesses were not significantly related to well-being.
Luzny and Ivanova (2009)	Cross-sectional	Country: Czech Republic N = 297 Diagnosis: several psychiatric conditions Mean age: 73.8 (SD = 6.5) residents aged > 65.	2	Two psychiatric hospitals in Czech republic	WHO-QoL BREF	Diagnosis	No significant differences in WB between gerontpsychiatric- and somatic residents. Qualitative findings: There were complaints about loneliness, lack of social contacts, suffering because of the disease and being hospitalized. Furthermore worries about the future (where to live, health etc.) were mentioned.
Cooper and Pearce (1996)	Longitudinal	Country: Australia N = 54 Diagnosis: Several psychiatric conditions median Age: 72 range: 42–90	3	Living in a PG-hospital, moving to - Supported resident services - Nursing homes in community - Or staying in PG hospital (control group)	Comprehensive quality of life scale (ComQol)	Type of setting Short term effect of relocation	Patients that moved to SRS showed significantly greater improvement on the emotional well-being domain (happiness) than patients who moved to a nursing home. No other significant differences.

Note: RH = residential homes, HC = Hospital care, PGCMS = Philadelphia Geriatric Centre Morale Scale, MANSA = Manchester Short Assessment of QoL, SRS = supported residential services (supported living = in ordinary low-level care facilities, with additional mental health care from the local psychiatric hospital).

Participants

All studies included participants with psychiatric diagnoses. Some studies investigated groups with specific diagnoses, such as major depressive disorder (Davison et al., 2012), depression and/or anxiety (Smalbrugge et al., 2006) or psychotic disorders (Kallert et al., 2007; Leisse & Kallert, 2000; Nakagawa & Hayashi, 2013). Other studies included people with a variety of diagnoses (Cooper & Pearce, 1996; Depla, De Graaf, & Heeren, 2005; Depla et al., 2006; Depla, De Graaf, Van Weeghel, & Heeran, 2005; Luzny & Ivanova, 2009). Mean ages varied from 58.5 years to 83.0 years. In more general care settings, mean ages tended to be higher (71.5–83.0) than in psychiatric care settings (58.5–76.4). In all studies except one, both men and women were included. In six studies women were in the majority, and in three studies relatively more men participated (Cooper & Pearce, 1996; Kallert et al., 2007; Leisse & Kallert, 2000). Nakagawa and Hayashi (2013) solely included female participants.

Settings

Participants lived in a variety of LTC settings. A majority of the studies involved participants living in a general nursing home (Cooper & Pearce, 1996; Davison et al., 2012; Smalbrugge et al., 2006) or in general nursing homes with additional mental health care services, provided by the local psychiatric hospital (Depla, De Graaf, & Heeren, 2005; Depla et al., 2006; Depla, De Graaf, Van Weeghel, & Heeran, 2005). One study involved participants from lower level general care facilities, like assisted living facilities (Davison et al., 2012). In other studies participants from psychiatric nursing homes, or nursing home areas of psychiatric facilities were involved (Kallert et al., 2007; Leisse & Kallert, 2000; Luzny & Ivanova, 2009). Furthermore, some of the studies included participants living in psychiatric hospitals, long-stay-wards of psychiatric centers or other institutions aimed primarily at psychiatric care (Depla, De Graaf, & Heeren, 2005; Depla, De Graaf, Van Weeghel, & Heeran, 2005; Kallert et al., 2007; Leisse & Kallert, 2000; Nakagawa & Hayashi, 2013). In most studies, the participants were recruited from several types of settings. There were some studies that also included participants who lived in other settings, e.g. at home with family, or in a sheltered community residence. These studies were included, but only results relating to the target population (i.e. LTC residents in residential facilities) were investigated in this review.

Measurement instruments

All studies applied one or more measurement instruments to assess quality of life or well-being. A total of eight different instruments were used, three of which were adaptations, based on the same original instrument. First, Davison et al. (2012) used a 42-item version of the Ryff multidimensional measure of psychological well-being. This measure is based on the eudaimonic view on well-being and focuses on the relation with self, others and the surroundings. Then, the Philadelphia Geriatric Centre Morale Scale (PGCMS) was applied by Depla, De Graaf, & Heeren (2005); Depla et al. (2006); Smalbrugge, Pot, Jongenelis, Gundy, Beekman, and Eefsting (2006). This 17-item instrument placing mood more centrally, is constructed for older people living in institutions. The third instrument that was used was the WHOQOL-BREF, the abbreviated 26-item version of the WHOQOL-100 (Luzny & Ivanova, 2009). This generic questionnaire is developed in the context of four domains: physical, psychological, social and environment (Skevington, Lotfy, O'Connell, & Group, 2004). Fourthly, the Comprehensive Quality of Life Scale (ComQol) was applied by Cooper and Pearce (1996), a 35-item measure of subjective quality of life. Nakagawa and Hayashi (2013) used two different instruments for the measurement of well-being. For objective well-being, the 21-item Quality of Life Scale (QLS) (Heinrichs, Hanlon, & Carpenter, 1984) was used, a disease specific, clinician-rated measure for people with schizophrenia. For the measurement of subjective well-being, an adapted version of the Lancashire Quality of Life Profile (LQoLP) was used, based on Lehman's Quality of Life Profile (Lehman, 1983). This instrument is adjusted and also used in two other versions: the Manchester Short Assessment of Quality of Life (MANSA) (Depla, De Graaf, & Heeren, 2005; Depla et al., 2006; Depla, De Graaf, Van Weeghel, & Heeran, 2005) and the Berlin quality of life profile (Kallert et al., 2007; Leisse & Kallert, 2000). The original LQoLP is developed for chronic psychiatric patients. It includes subjective evaluations of satisfaction on nine life domains: living situation, family, social relationships, leisure activities, work/education, finances, personal safety, health and religion. Beside the subjective evaluations, this instrument includes objective life conditions, and a global well-being measure as part of the measurement instrument (Oliver, Huxley, Priebe, & Kaiser, 1997; van Nieuwenhuizen, Schene, Koeter, & Huxley, 2001). In the versions that were used in the included articles some adaptations were made in the domains. The religion and work domains were omitted in most versions, while the MANSA included both sex-life and fellow-residents as domains. Table 3 provides an overview of all measurement instruments in the included articles.

Reported outcomes

Type of residence and characteristics of care

The different aspects of residence and characteristics of care were a factor that received considerable attention. Group- versus single living was studied by comparing well-being scores of participants who spent a major part of the day in a communal living room, with those of participants who spend the day in their own flat, located throughout a care facility (Depla et al., 2006). No differences in well-being were found in either psychotic or non-psychotic participants.

The relation between perceived amount of personal freedom and well-being was dependent on several factors (Depla, De Graaf, & Heeren, 2005). For the non-psychotic subgroup there was a positive relation between perceived amount of personal freedom and well-being for some of the PGCMS subscales, but only the relation with one PGCMS-subscale (aggression) remained significant when adjusted for patient characteristics. There was a positive relation with MANSA, the other well-being instrument, but only when adjusted for housing characteristics. For the psychotic subgroup there was no relation with well-being when adjusted for housing or patient characteristics.

The relationship between duration of hospitalization and well-being resulted in conflicting outcomes among 66 women with schizophrenia. Duration was negatively related to objective well-being, measured with the QLS, and positively related to subjective well-being, measured with the LQLP (Nakagawa & Hayashi, 2013). In the same group, Nakagawa and Hayashi (2013) found that the number of admissions to psychiatric hospitals was positively related to subjective well-being, but not to objective well-being.

No adverse effect was found for moving to a new location (Cooper & Pearce, 1996). Fifty-four residents of a large psychiatric hospital, who either moved to smaller scale nursing homes or to supported residential services (a lower level care institution) and a small control-group of 18 non-movers were investigated. After relocation, the supported residential services-group scored higher on both well-being indexes, compared to the nursing home group and the non-movers. There was no difference in well-being between the nursing home group and the non-movers (Cooper & Pearce, 1996). However, these results should be interpreted with caution, since at baseline the supported residential services-group reported higher material well-being. Also, the method of measuring well-being in the supported residential services (self-rating) differed from the method that was used in the nursing homes (by proxy).

A final aspect with regard to living situation is the difference between general care locations and residences that specifically provide psychiatric care. A higher well-being score was found among residents of a psychiatric hospital, compared to a matched group of residents receiving general care, who lived in residential homes that followed a ‘supported living program’ (Depla, De Graaf, & Heeren, 2005). Depla and her colleagues explored some of the possible explanations for this result. One possible explanation is the difference in availability of mental health workers (MHW-staff). It was found that for participants with a psychotic disorder the availability of MHW-staff was negatively related to agitation, one of the subscales of the PGCMS. No differences were found however, for residents with non-psychotic Axis I disorders (Depla et al., 2006). Another potential explanation for this difference would be stigmatization in the general health care settings. Depla, De Graaf, Van Weeghel, and Heeran (2005) found a negative relation between stigmatization and well-being, a relation that remained significant when controlling for several confounders (i.e. age, gender, cognitive impairment, mastery, ADL assistance needs, behavioural problems, network size and social activities). There was no difference however, in the amount of reported stigmatization between the general care settings and the psychiatric hospital.

Diagnosis

The relationship between type or characteristics of diagnoses and well-being was investigated by Davison et al. (2012), Luzny and Ivanova (2009) and Nakagawa and Hayashi (2013), Smalbrugge et al. (2006).

Both Davison et al. (2012) and Smalbrugge et al. (2006) found a negative relation between depression and well-being. Davison et al. (2012) compared depressed elderly (N = 50) to a matched non-depressive control-group (N = 50). When confounders (duration of residence, health and disability variables) were controlled for, there was still a negative relation between depression and three subscales of well-being: environmental mastery, autonomy and purpose in life. Smalbrugge et al. (2006) compared patients with depression, patients with anxiety, patients with both, and patients without depression or anxiety. These groups were different in size, varying from N = 16 (patients with anxiety) to N = 243 (patients without depression or anxiety). They found that participants with combined depression and anxiety or pure depression experienced lower well-being than participants with pure anxiety or no depression/ anxiety.

No differences were found in well-being among psychiatric and somatic nursing home inhabitants (Luzny & Ivanova, 2009). In this study, participants from two separate institutions with different approaches (holistic vs. conventional) were compared. Confounders were not accounted for. The response-rate was low (26,3% for the somatic nursing home, and 23,4% in the psychiatric hospital).

In a study including chronic schizophrenic women, some symptoms of schizophrenia, measured with the Positive And Negative Symptoms of Schizophrenia (PANSS), correlated negatively with well-being. Anergia was negatively related with objective wellbeing (QLS-total score), and two other PANSS subscales were related to subjective well-being (LQLP): depression (negatively) and paranoid/belligerence (positively) (Nakagawa & Hayashi, 2013).

Social activities and network

Involvement in social activities such as shopping, exercise class or going on visits, was positively related to well-being (Depla, De Graaf, Van Weeghel, & Heeran, 2005). This relationship was however, no longer significant when adjusted for confounders (i.e. gender, age, cognitive impairment, mastery, ADL assistance needs, behavioural problems, social network size and stigmatization). After adjustment, there was even a negative relation between activity and well-being, for the subgroup of psychotic residents in a general care home.

In the same study by Depla, De Graaf, Van Weeghel, and Heeran (2005) network size was studied, which was measured by the number of people with whom respondents maintained regular and meaningful contacts. Both in the crude and in the adjusted regression analysis, a positive relation was found between network size and well-being.

Other factors

Nakagawa and Hayashi (2013) investigated the relationship between various other factors and the level of well-being. Only the strongest correlations are mentioned here. A negative correlation was found between current age and objective well-being measured with the Quality of Life Scale (QLS). This was not the case for subjective well-being, which was measured with the Lancashire Quality of Life Profile (LQLP). The relation between well-being and daily life activities, measured with the Rehabilitation Evaluation Hall and Baker (REHAB) was also studied, using multiple stepwise regression. Only a negative relation between the REHAB subscale ‘community skills’ and objective well-being (QLS) was found (Nakagawa & Hayashi, 2013).

Discussion

The aim of this review is to provide an overview of aspects that are linked to well-being among older, psychiatric long-term care patients. Primarily it is notable that only a few studies have been conducted on well-being in gerontopsychiatry. Considering that well-being is one of the main healthcare outcomes, a mere ten studies on this topic is a disappointingly small number. Roughly one study per subject was readily available and the population tested in these studies was heterogenous in the type of residences, psychiatric disorders and the definitions of well-being that were being tested. This limits the possibilities of general conclusions.

The studies that were included suggest that specialist care aimed at psychiatric disabilities and the availability of mental health workers are positively related to well-being (Depla, De Graaf, & Heeren, 2005; Depla et al., 2006). Depression (Davison et al., 2012; Smalbrugge et al., 2006), and also some symptoms of schizophrenia (Nakagawa & Hayashi, 2013) appear to be negatively related to the level of well-being. Stigmatization perceived by the residents is linked to lower well-being, whereas larger social network size, and (perceived) personal freedom are related to a higher sense of well-being (Depla et al., 2006).

Strengths and limitations

Although a systematic and broad literature search was performed in this review, it always remains possible that articles were overlooked. Also, negative results are not always published, and may therefore be omitted.

Various diagnostic groups were examined. These groups may differ in their characteristics and outcomes, however, due to the small number of studies it is not feasible to discuss diagnostic groups separately. For the measurement of well-being, as many as eight different instruments were used within ten studies, using not entirely similar theoretical constructs of well-being. Also, in some cases when one instrument was used, more methods were applied to measure differences in well-being between two groups within one study (i.e. via interviews or with proxy measurements) (Cooper & Pearce, 1996). Comparability of results on these measures is therefore limited.

When it comes to methodological quality, a general problem in research with this population is the response rate. When data collection is performed through interviews with residents, response rates tend to be low. Due to refusal, cognitive impairment or severity of mental or physical health problems, reported response rates in the included articles vary between 23,4% and 59%. Only Leisse and Kallert (2000) reported they had included all schizophrenic inpatients in a specific nursing home. It remains unclear what approach was used for this high response rate. Due to the relatively low response rate of most studies, the results may have been biased, since it is probable that the non-responders differed in some respects. They may have been more cognitively impaired, have a higher level of anxiety or they may have been more paranoid, or low in socially desirable behaviour.

Another methodological problem is the fact that in four of the included studies the possible influence of confounding factors was overlooked (Cooper & Pearce, 1996; Leisse & Kallert, 2000; Luzny & Ivanova, 2009; Nakagawa & Hayashi, 2013). In the studies by Depla et al., it was shown that confounders such as age, gender and marital status can have a substantial influence on the results. These two methodological issues, the low response rates in the included studies, and the disregard of confounders in some studies, compromise generalizability of the results.

Eight out of ten studies were cross sectional, which means that the direction of the relations that were found remain unclear. This applies strongly to depression for example, where not only the relation may be reciprocal, but in addition it might be argued that depression is part of well-being. Feelings of depression are considered to be an element of well-being in the WHOQOL among others.

Finally, it is clear that not all factors that could potentially influence well-being in the gerontopsychiatric population are investigated. Conclusions in this review are limited to the topics studied so far, which may be an arbitrary selection of the possible noteworthy factors.

A strength in this study is the diversity of countries and continents where the included studies were conducted. Most articles were from Europe: The Netherlands, Germany and Czech Republic, but studies were also performed in Australia and Japan. No studies from the US were found on well-being in gerontopsychiatric LTC residents. A possible explanation is the fact that gerontopsychiatric residents in the US generally live in mainstream nursing homes, among cognitively disabled residents (Grabowski, Aschbrenner, Rome, & Bartels, 2010). They may not be seen as a distinct group and therefore not be investigated separately. Also the stronger emphasis on temporary care in US nursing homes (Fullerton, McGuire, Feng, Mor, & Grabowski, 2009) may be a reason for the absence of studies, since the focus of this study is on long-term care.

Implications and recommendations for care facilities

Due to the small number of studies on well-being for gerontopsychiatry, implications are few, and should be stated with some caution. However, it seems that settings with specialized mental health care meet the needs of gerontopsychiatric LTC residents better than general care settings, and are the preferred setting for this population.

Stigmatization, the feeling of being treated differently, or in a negative way because of a psychiatric disorder is negatively related to well-being. A social environment that is accepting of psychiatric disability is therefore recommended for this population.

Special attention is needed for depressed elderly in LTC settings. Depression appears to be strongly related to well-bhing, stronger than other mental disorders such as anxiety. Treatment of depression should therefore be a priority in care for gerontopsychiatric LTC residents.

Lastly, the perceived amount of personal freedom, as well as the size of social networks appear to be positively related to the level of well-being. Exercise of personal freedom within the boundaries of responsible care should be encouraged by care workers. In addition, assistance in the maintenance or growth of social networks, might promote well-being in the gerontopsychiatric population.

Recommendations for further research

The first recommendation regarding future research on well-being in gerontopsychiatric LTC-residents would be to develop and validate an instrument specifically for the measurement of well-being in this population. This would improve comparability and might prevent drop out, caused by cognitive demands that are too high. Also, to better address the problem of low response-rate, the development of a specialized by proxy instrument is recommended, of which the outcomes are related as closely as possible to a self-rated well-being scale.

More research on the relation between determinants and the level of well-being in gerontopsychiatric LTC residents, is highly desirable. Replication of the existing studies could allow for firmer conclusions to be drawn on the topics described in this study.

Determinants that have already been found to be related to well-being among the elderly or among psychiatric inpatients might also be an interesting focus for further research. Examples of such determinants are functional status (Cummings, 2002), quality of the relationship with health care staff (Custers, Westerhof, Kuin, & Riksen-Walraven, 2010), pain (Jakobsson, Hallberg, & Westergren, 2004), psychiatric diagnosis and severity of the disorder (Picardi et al., 2006), and behavioural disturbance (Banerjee et al., 2006). Special focus should be placed on themes like pain or the relationship with health care staff, since these are subjects that might be influenced by treatment or training, and may therefore be good starting points to explore ways in which well-being can be improved in this population.

Furthermore, well-designed experimental- and longitudinal research is recommended, to investigate the direction of the relationship between different topics and well-being, and the effects of treatment or improvement regarding these topics on well-being.

To conclude, this study shows that although the group of older LTC residents with chronic mental disorders is substantial and still growing, and although well-being is one of the main aims of care for this population, there is very limited evidence based knowledge on this theme. The evidence that is available suggests that an accepting and non-stigmatizing environment with specialist psychiatric care, one that encourages autonomy and provides effective treatment for depression, would result in higher well-being. For good quality of care, more knowledge on well-being and potential related factors is essential.

Disclosure statement

The authors report no conflict of interest.