At home or in hospital: Home treatment and mental health stigma.

BACKGROUND: Stigmatized attitudes towards people with mental illness may influence treatment choice for oneself and others. AIM: To gauge the attitudes of the UK general public towards treatment at home for mental illness and to assess the extent to which non-acceptability was related to stigmatized attitudes.
METHODS: Two hundred and two (101 female) people living in the UK completed an online (vignette) questionnaire in which we asked demographic details and personal experience of mental illness. To measure stigma, we used an adapted version of the Attitudes to Mental Illness Questionnaire (AMIQ) with vignettes asking about treatment at home and using scales for social distance and poor expectations; participants also filled in the Mental Health Knowledge Schedule (MAKS).
RESULTS: Participants did not evidence overall agreement with treatment at home for mental illness (i.e. >0; range = -16-to-+16, Mean (M) = 0.86, 95% confidence interval (CI) = -0.08, 1.80, p = .073), although they showed significant agreement with treatment at home should they experience mental illness themselves (range = -8-to-+8, M = 1.36, CI = 0.82, 1.89, p < .001). Acceptability for treatment at home differed according to specific mental illness considered (range = -4-to-+4); depression (M = 0.47, CI = 0.13, 0.81, p = .006) and alcohol abuse (M = 1.46, CI = 1.14,1.77, p < .001) were considered suitable for being treated at home but schizophrenia was not (M = -0.78, CI = -1.13,-0.43, p < .001). Multivariate analyses revealed that older age and attitudes indicating comfort with less social distance from people with mental illness were independently associated with treatment at home agreeability.
CONCLUSIONS: Public acceptability of home treatment for mental illness remains ambivalent in the UK, most obviously when considering treatment approaches for individuals other than themselves and for people with schizophrenia. Disagreement with home treatment is particularly evident in younger people and those who prefer less social contact with people with mental illness.

Introduction

It is generally assumed that people do not want to be admitted to a psychiatric hospital, should they become mentally ill, and that they prefer being treated at home. This has been one of the reasons for introducing home treatment teams (Hubbeling & Bertram, 2012; Wheeler et al., 2015). The empirical evidence for this view is predominantly based on comparisons of treatment satisfaction between patients admitted to hospital and patients treated at home, whereby the latter group show greater satisfaction levels with their treatment (Henderson et al., 1999; Hoult et al., 1983; Johnson et al., 2005a). One study suggested that a little over half of carers for patients who had experienced both home treatment and admission to hospital preferred treatment at home in the event of acute mental illness relapse in a relative. Home treatment preferences were linked to carers’ perception of their relative’s illness and relapse behaviours as less severe with lower levels of subjective burden, expectations of poorer behavioural responses (in their relative) to hospitalisation, and higher levels of confidence for coping during their relative’s acute episodes (Fulford & Farhall, 2001). However, the extent to which the general public thinks it is preferable to treat people with mental health problems at home or in hospital is unknown, and the reasoning underpinning any preference is unclear, in particular, whether it is influenced by stigmatized attitudes towards people with mental illness.

Home Treatment Teams (HTTs) were introduced in England by the National Health Service (NHS) Plan of 2000 as an alternative for an acute admission to a psychiatric hospital (Department of Health, 2000), whereby patients receive intensive treatment at home for a couple of weeks. Similar services were developed in other parts of the United Kingdom and in the rest of the developed world (Johnson, 2013). There have been a number of randomized controlled trials (RCTs) investigating efficacy of treatment at home as an alternative to hospitalization. In an early study, Hoult et al. (1983) reported that home treatment achieved a clinically superior outcome to standard care with respect to readmissions within 12-months, although in this trial, patients were not referred back to their community mental health team (CMHT), outpatient clinic or primary care provider after the treatment period with the HTT but continued with the intervention treatment throughout the study period. The outcomes of two more recent RCTs (one in England [Johnson et al., 2005b] and one in Switzerland [Stulz et al., 2020]), whereby patients were referred back to their usual care provider after the HTT intervention, suggested that many people experiencing an acute mental health problem can be successfully treated at home instead of in hospital. Treatment at home, at least in a research setting, can also be cost-effective (Hoult et al., 1983; McCrone et al., 2009). However, it is not clear whether HTTs reduce costs in routine clinical practice (Tulloch et al., 2015) because HTT involvement may last longer than an admission to hospital, and in practice, cost-effectiveness likely depends on specific local circumstances.

Many people with mental health problems never seek help and many others do not get the most appropriate treatment, because they perceive seeking help and receiving treatment as not socially acceptable (Corrigan et al., 2014; Schnyder et al., 2017), indicative of stigmatized attitudes towards mental illness. Stigma is a multifaceted concept (Thornicroft et al., 2007) consisting of attitudes (prejudice), knowledge (ignorance or misinformation) and behaviour (discrimination), and is measured in different ways, including via questionnaires and vignettes (Brakel, 2006). Vignettes have the advantage that a more elaborate stimulus is used, even though it remains hypothetical (Link et al., 2004), while questionnaires are advantageous in so much as they capture more general attitudes.

Luty et al. (2006) conducted a postal survey using vignettes showing highly stigmatized attitudes towards people with drug addiction, more positive attitudes towards people with depression, and in-between positions for alcohol addiction and schizophrenia. Henderson et al. (2012)’s population-based study observed less stigma in people who reported experiencing mental illness themselves or knowing somebody suffering from mental illness. But, importantly, levels of stigma differed according to whether data was collected via face-to-face interviews or web-based questionnaires, the authors suggesting the latter may be preferable in view of the potentially bias introduced by interviewer effects and socially desirable responding in interviews.

Therefore, a web-based study of a sample of the population in the UK seemed appropriate to establish the extent to which people agree with being treated at home for mental health problems and the association between wanting treatment at home and stigma surrounding mental illness. Both vignettes and questionnaires were employed in this study. We anticipated that agreement with home treatment in the general public would vary across different mental health conditions and that less agreement with home treatment would relate to less knowledge about mental illness and higher levels of stigma.

Methods

Sample

Participants were recruited from people living in the UK (18 years or older) via Prolific Academic, a provider of online research participants (Peer et al., 2017). A previous study showed that data quality from Prolific Academic surveys was high (Peer et al., 2017). This sample was stratified to recruit equivalent numbers of men and women and those aged ⩽45 years and >45 years. Because home treatment instead of hospital admission is offered to adults of all ages in the UK, this study aimed for a wider range of participants than Henderson et al. (2012), who studied a sample aged 25–45 years. This was done because their study evaluated the ‘Time to Change’ anti stigma campaign which focused on the most relevant age group. All participants gave their (online) consent with data anonymised at the point of collection. The study was approved by the St George’s Medical School Ethics Committee (reference number 2019.0072).

Measures

Sociodemographic details were collected from each participant. We also asked whether they had experienced mental illness themselves and if they knew somebody who was affected by mental illness (e.g. partner, family, friend, work colleague, neighbour).

Because there was no previous study investigating the link between attitudes towards treatment at home and mental illness stigma – four vignettes from the AMIQ (Luty et al., 2006) based on presenting symptoms (depression, schizophrenia, alcohol addiction, drug addiction) were adapted to include questions about treatment at home. An example vignette was ‘Michael has schizophrenia. He needs an injection every 2 weeks but missed it last week. He is hearing voices from the Devil and thinks that he can cause earthquakes’ (see Supplemental Material for the questions used).

For each vignette, attitudinal items (measuring agreement/perceived likelihood on a 5-point Likert scale, ranging from ‘Strongly Disagree’/’Very Unlikely’ [−2] to ‘Strongly Agree’/’Very Likely’ [+2] with a ‘Don’t Know’ response option scored at the midpoint [0]) concerning whether this person could be treated successfully at home (e.g. ‘Michael can be treated at home for his mental health problems and he does not need to be admitted to a psychiatric hospital’) and whether the research participant would like to be treated at home if he or she were to experience a similar condition (e.g. ‘If I had the same condition as Michael, I can be treated at home’) were included. Preference for home treatment scores were calculated separately for different conditions (by summing the responses to the subject in the relevant vignette and the corresponding item referring to oneself; scale range −4 to +4), and by summing across conditions for items referring to oneself (‘Self’; scale range −8 to +8), across conditions for items concerning the person in the vignette (‘Vignette person’; scale range −8 to +8), and finally, across all attitudinal items concerning treatment at home (‘Total’; scale range −16 to +16). Higher scores indicated greater acceptability of home treatment on all measures.

For each vignette in the adapted version of the AMIQ, there were also three attitudinal items measuring agreement levels with comfort of social distance (neighbour, work colleague, dinner party guest) from people with mental illness (e.g. ‘I would be comfortable having Michael as my neighbour’) and three questions measuring perceived likelihood (expectations) of negative consequences for people with mental illness (with respect to career, trouble with the law and marriage difficulties; e.g. ‘Do you think that this would damage Michael’s career?’), each responded to on a 5-point Likert scale as above. The neighbour question was added to the items used by Luty et al. (2006) because a question about neighbours has been included in other social distance measures (Whatley, 1958). Scores in the four vignettes were combined (summed) to give separate measures of social distance (scale range −24 to +24) and poor expectations (scale range −24 to +24), with higher scores implicating less desire for social distance and less expectations of a poor outcome. Although the distinction between questions assessing social distance and expectations was made on a conceptual basis, it was confirmed by principal component analyses.

In summary, the vignettes combined provided measures of acceptability of home treatment for the person in the vignette (‘Vignette person’; scale range −8 to +8), for the participant him/herself (‘Self’; scale range −8 to +8), and an overall total (‘Total’; scale range −16 to +16), in addition to separate measures of social distance comfort (scale range −24 to +24) and poor expectations (scale range −24 to +24).

The knowledge and behavioural components of stigma were measured using the first six questions of the 12-item MAKS; Mental Health Knowledge Schedule (Evans-Lacko et al., 2010). The MAKS questions assess mental health knowledge (help seeking, recognition, support, employment, treatment and recovery). Agreement levels on each item were measured on a 5-point Likert scale with an additional ‘Don’t Know’ response option. Higher scores on the MAKS indicate greater knowledge of mental illness. The psychometric properties (e.g. test-retest reliability and internal consistency) of the MAKS in research with member of the general public are good (Evans-Lacko et al., 2010).

Statistical analyses

Descriptive data describing the sample and questionnaire scale totals were presented in the form of mean (M) and standard deviation (SD) and n (%) as appropriate. One-sample t-tests were employed to determine if (as a group) participants’ preference for home treatment scores were positive or negative (i.e. significantly differed from a neutral position), and paired-sample t-tests were administered to compare between home treatment preferences for oneself versus those for the person in the vignette, and home treatment preferences and adapted AMIQ scores across different mental illnesses (as described in vignettes).

To test whether preference for being treated at home related to knowledge of mental illness (MAKS) and stigma (social distance and poor expectations), regression models were constructed for overall preference for home treatment (‘Total’), preferences for home treatment for the persons in the vignette (‘Vignette person’) and home treatment preferences for oneself (‘Self’). Socio-demographic factors (age, sex, marital status, employment status, education level) and/or experience of mental illness oneself or in known others were included in regression models only when indicated by significant univariate associations (Pearson’s correlation/Spearman’s rho or independent group t-tests with bias corrected and accelerated bootstrapping using 2000 replications according to data distribution). Statistical analyses were administered using SPSS (IBM, Version 25.0) with a criterion for statistical significance set at p < .05.

Results

Sample characteristics

More than 200 (202) participants were recruited via Prolific Academic in May 2019 and completed online measures. Questionnaire results for one participant were incomplete. Table 1 shows the sociodemographic characteristics of the sample. There were equal numbers of men and women and an average age of just over 40 years (range = 18–77). Almost 60% of participants were working and a little less than two-thirds married or living with their partner. Participants were, on average, highly educated; less than a quarter of participants had no GCSEs or GCSEs only, while just under half were educated at a degree level or more. Sixty participants (30%) stated that they were suffering from a mental illness themselves while only 36 participants (18%) reported that they did not know anybody suffering from a mental illness.

Table 1.

Sociodemographic characteristics of study sample (n = 202). Numbers are frequencies (percentages) unless otherwise stated.

Gender
Male	101 (50%)
Female	101 (50%)
Age (Mean, SD)	41.9 years (15.2)
Employment status
Working full-time	74 (36.6%)
Working part-time	43 (21.3%)
Retired	25 (12.4%)
Studying	28 (13.9%)
Other	32 (15.8%)
Marital status
Married	90 (44.6%)
Living with partner	39 (19.3%)
Single	73 (36.1%)
Highest educational level
No GCSEs	1 (0.5%)
GCSEs	44 (21.8%)
A-level	58 (28.7%)
BA or BSc	67 (33.2%)
MA or MSc	24 (11.9%)
Doctorate	8 (4.0%)
Do you suffer from a mental illness yourself?
Yes	60 (29.7%)
No	142 (70.3%)
Do you know anybody who is suffering from a mental illness?
Yes	166 (82.2%)
No	36 (17.8%)

Note. Age data was not reported by one participant.

Agreement with home treatment and attitudes towards mental illness

Table 2 summarizes participants’ scores on scales measuring agreement with treatment at home for mental illness, attitudes towards mental illness and stigma-related mental health knowledge. Overall, participants evidenced a trend towards agreement with treatment at home (p = .073). Participants showed a trend towards disagreement with the person in the vignette being treated at home (p = .056) but showed significant agreement for being treated at home themselves, if they were to suffer from the same condition (p < .001). Preference to be treated at home oneself and agreement that the person in the vignette could be treated at home were moderately associated (Spearman’s ρ = .62, p < .001), but the difference between the two was significant (t200 = 8.07, p < .001).

Table 2.

Acceptability of treatment at home, knowledge of and attitudes towards mental health/illness, and social desirability answering.

Measure	Mean (95% CI)
Treatment at home
Total (−16, 16)	0.86 (−0.08, 1.80)
Vignette person (−8, 8)	−0.50 (−1.01, 0.01)
Self (−8, 8)	1.36 (0.82, 1.89)
Depression (−4,4)	0.47 (0.13, 0.81)
Schizophrenia (−4,4)	−0.78 (−1.13, −0.43)
Alcohol abuse (−4,4)	1.46 (1.14, 1.77)
Heroin addiction (−4,4)	−0.29 (−0.66, 0.07)
Attitudes towards mental illness questionnaire (AMIQ)
AMIQ depression (−10,10)	2.36 (1.92, 2.80)
AMIQ schizophrenia (−10,10)	−2.74 (−3.27, −2.20)
AMIQ alcohol abuse (−10,10)	−0.99 (−1.49, −0.49)
AMIQ heroin addiction (−10,10)	−3.51 (−4.07, −2.95)
Social distance score (−24,24)	2.96 (1.64, 4.28)
Poor expectations score (−24,24)	−6.56 (−7.48, 5.64)

Note. CI = confidence interval.

Acceptability for treatment at home differed according to the mental illness considered in the vignette (Table 2). Agreement levels across participants suggested that, overall, depression (p = .006) and alcohol abuse (p < .001) were considered suitable for being treated at home (i.e. significantly >0) and schizophrenia (p < .001) was not (i.e. significantly <0). All acceptability of treatment response differences between (pairs of) mental illnesses were significant (p < .019). For all conditions, participants preferred being treated at home significantly more for themselves than for the person in the vignette (Figure 1).

Figure 1.

Participant attitudes towards being treated at home for different mental disorders according to whether subject was the person in the vignette or themselves. Positive scores indicate a positive attitude, zero indicates a neutral attitude, and scores less than zero indicate a negative attitude. Data labels indicate mean values. Error bars represent the standard error of the mean. Asterisks indicate significant pairwise differences; *p < .05, ***p < .001.

Attitudinal scores (AMIQ) for each mental health condition indicated responses that were, overall, more positive for the vignette concerning depression than other illnesses (Table 2, for all pairwise comparisons, p < .001). The vignette concerning heroin addiction elicited the most negative attitudinal responses (for all pairwise comparisons, p < .007).

In general, participants evidenced high levels of mental health knowledge. The mean score on the MAKS, 23.98 (SD = 2.93), was more than a standard deviation higher than the equivalent (M = 19.1, SD = 4.7) in the (online) sample of Henderson et al. (2012).

Relationships of acceptability of home treatment with sociodemographic factors and experience of mental illness oneself or in known others

Age was significantly associated with measures of (overall) preference for treatment at home (Total, ρ = .36, p < .001), reflecting high levels of acceptability in older participants. Acceptability of treatment at home scores indicated that participants who were married/living with partner (Total, M = 1.68, SD = 6.87) were more likely than those who were single to prefer home treatment (Total, M = −0.58, SD = 6.39, t199 = 2.30, p = .023). There were no differences on measures of acceptability of home treatment according to gender, employment status (working versus not working/studying) and education level (no degree versus degree or above), or between those participants who did and did not report suffering from a mental illness and those who did or did not report knowing somebody suffering from a mental illness (for all comparisons, p > .204).

Acceptability of treatment at home and stigma

Overall acceptability of home treatment was significantly correlated with scores on social distance (r = 0.36, p < .001), poor expectations (r = 0.16, p = .026) and MAKS (r = 0.16, p = .024) measures. Linear regression modelling revealed that only age and social distance were significant predictors (Table 3). Specifically, acceptability of treatment scores was associated with an increase of 0.14 points for every year of age and an increase of 0.19 points for every 1-point increase scored on the social distance measure, with equivalent effects between Vignette person and Self measures of home treatment acceptability. The variance explained by each was modest (21%, 19% and 17% for Total, Vignette person and Self acceptability of treatment at home measures, respectively).

Table 3.

Regression analyses of relationships between treatment at home acceptability and sociodemographic factors and knowledge of and attitudes towards mental health/illness.

Variable	Total		Person vignette		Self
	B (CI)	p	B (CI)	p	B (CI)	p
Age (Years)	0.14 (0.08, 0.20)	<.001	0.07 (0.04,0.11)	<.001	0.07 (0.03, 0.10)	<.001
Married/Living together	0.56 (−1.32,2.44)	.558	0.31 (−0.73,1.34)	.561	0.25 (−0.85, 1.36)	.652
Social Distance	0.19 (0.08, 0.30)	<.001	0.10 (0.04, 0.15)	.001	0.09 (0.03, 0.16)	.003
Poor Expectations	0.01 (−0.13,0.15)	.867	0.01 (−0.08, 0.08)	.971	0.01 (−0.07, 0.09)	.803
MAKS	0.22 (−0.09, 0.52)	.168	0.12 (−0.05, 0.29)	.161	0.10 (−0.09, 0.28)	.304

Note. Age data was not reported by one participant; hence the models include 200 participants; Reference category for Married/Living together = Single; Unstandardized beta values (B) and associated 95% confidence intervals (CI) and p values were calculated using linear regression. For ‘Total’, R = 0.48, adjusted R-square = 0.21, p < .001; ‘Vignette person’, R = 0.45, adjusted R-square = .19, p < .001; and ‘Self’, R = .42, adjusted R-square = .18, p < .001; the maximum Variance Inflation Factor for the final models was 1.41 indicating no multicollinearity; MAKS = Mental Health Knowledge Schedule; significant associations are highlighted in bold.

Discussion

This was a web-based study asking members of the public whether they considered treatment at home for mental health conditions acceptable. The primary findings were that, overall, the sample was ambivalent about home treatment for people with mental illness, with less support for home treatment for others experiencing mental illness compared to themselves (if suffering from the same condition) and for mental illnesses such as schizophrenia and drug addiction compared with depression or alcoholism. Acceptability of treatment at home was most closely linked with older age and attitudes reflecting comfort with less social distance from people with mental illness.

In this study, 30% of participants asserted that they were suffering from an illness themselves, slightly higher than is normally reported, 25% (Bebbington & McManus, 2019), and higher than 12% reported in Henderson et al.’s (2012) online study. Bebbington and McManus (2019) have suggested that if ADHD and personality disorder are included, then the prevalence of mental health problems may be one in three, which is comparable to the rate reported here. Participants were recruited via Prolific Academic announcing a study about views of mental illness and it is possible that this attracted more people suffering from a mental health problem.

While participant attitudes towards treatment at home for people suffering from mental illness varied across the sample, overall, participants evidenced no clear agreement with treatment at home. Agreement levels differed considerably across specific mental illnesses, however, with significant acceptability for home treatment for people with depression and alcohol abuse but less favourable attitudes towards home treatment for those with schizophrenia or drug addiction. Given that home treatment teams were introduced in the UK to treat patients with wide-ranging mental illnesses more than a decade ago, and evidence that it is possible to successfully treat people with an exacerbation of schizophrenia symptoms or illness relapse at home (Johnson et al., 2005b), the results suggest that the general public may need more information about the appropriateness of treatment at home for people with mental illness, particularly for patients with schizophrenia.

Irrespective of the specific mental illness, more people thought they could be treated at home themselves if they experienced mental illness than for the person in the corresponding vignette. Eisenberg et al. (2009) found that stigma regarding mental health treatment seeking is more prevalent in responses to items beginning ‘Most people think. . .’ compared to those beginning ‘I think. . ...’. Participants in the present study may have thought that, in general, people with a certain mental health condition should go to hospital but that it would not be necessary for them should they suffer from the same condition.

Multivariate analyses suggested older age and comfort with less social distance with people experiencing mental health difficulties were the most important correlates of preference for treatment at home, both for oneself and for the person in the vignette. The relevance of attitudes concerning social distance to home treatment preferences is intuitive. If, for example, an individual does not want neighbours or work colleagues to know that they are suffering from a mental health problem, they may prefer to not be treated at home, where people in close proximity may become aware. Nevertheless, further research is needed to elucidate the specific mechanisms underlying this relationship. Notably, other recent research suggests that younger people with mental illness report more mental illness stigma (Bhavsar et al., 2019; Nugent et al., 2020). In the present study, however, age was associated with acceptability of home treatment independently of the influence of social distance or other factors related to stigma-related mental illness knowledge. Hence other – currently unknown – factors related to age are likely to play a role as well.

Limitations

This study has some potentially important limitations. The data were cross-sectional, precluding firm conclusions about the causal direction of observed associations. Participants were recruited via an online platform and were highly educated compared to other comparable studies (Henderson et al., 2012), so results may not be representative of the public at large. However, given the observed education levels and that MAKS scores suggested participants may have known more about mental illness than often shown in studies with the general public, it is possible that participants were, on average, more likely to be in favour of being treated at home for a mental crisis than the public at large. The general public may be more ambivalent about the merits of treating people with, for example, a relapse of schizophrenia, at home than participants in this sample.

The vignettes provided limited information about patients and, of course, clinicians gather more information before they make a decision about admission to hospital or approaches directed towards treatment at home. However, the poverty of specific information about vignette cases is also apparent in studies adopting more conventional mental illness stigma questions such as ‘Would you invite somebody with schizophrenia for dinner?’ (Henderson et al., 2012; Luty et al., 2006). In clinical practice, services users and their families sometimes do think that hospital admission is warranted or necessary, and this study suggests that at least some people believe prima facie that patients with mental health problems cannot be treated at home.

Conclusions

Home treatment for people with mental illness is well established in clinical practice in the UK. This study, however, suggests that the general public do not always hold favourable views towards treatment at home for patients with mental illness, particularly when that treatment is intended for another person experiencing mental illness (and not themselves), and that illness is schizophrenia. Younger people and people who are worried about social distance aspects of mental illness stigma are more likely to disagree with home treatment for people with mental illness. Raising public awareness of how people with mental health conditions can be treated successfully at home, in particular patients with schizophrenia, may be helpful in addressing the public’s concerns regarding treatment at home.