There is growing recognition that work and employment are important in maintaining and enhancing the health and social functioning of people with mental disorders and in promoting social inclusion, in addition to the economic benefits of employment. ^{Reference Boardman, Grove, Perkins and Shepherd1,Reference Cook and Pickett2} However, people with severe mental illness have lower employment rates than those with other psychiatric difficulties. Studies have reported that 8-20% of people with a psychotic disorder obtain employment in the UK, ^{Reference Kelly, McCreadie, MacEwan and Carey3–5} despite evidence that people with severe mental illness are eager to work and can be successfully employed. ^{Reference Marwaha, Johnson, Bebbington, Stafford, Angermeyer and Brugha6–Reference Hatfield, Huxley and Mohamed10} Unemployment rates are also high for patients with affective disorders, although not as high for those with schizophrenia – studies report rates of 35-60% of unemployment. ^{Reference Marwaha and Johnson11–Reference Cook and Razzano13}

A number of barriers have been identified that may prevent people with severe mental illness from seeking and retaining a job. These range from concern about losing benefits, the belief that they will not be offered employment because of the stigma of mental illness, and worries concerning poor functioning and relapse arising from their mental health problems. ^{Reference Shepherd, Murray and Muijen8,Reference Crowther, Marshall, Bond and Huxley14} The educational achievement of people with severe mental illness is also often low owing to early illness onset, ^{Reference Dion, Tohen, Anthony and Waternaux15} which may also reduce their ability to obtain work.

Previous research has found that in general for people with mental illness, better employment outcomes are connected with younger age. ^{Reference Harrow, Goldberg, Grossman and Meltzer16,Reference Barkow, Maier, Ustun, Gansicke, Wittchen and Heun17} A meta-analysis has also reported that White individuals are significantly more likely than Black and minority ethnic individuals to obtain employment. ^{Reference Barkow, Maier, Ustun, Gansicke, Wittchen and Heun17} However, a study of long-term unemployment in an inner-city setting of adults with severe mental illness by Goldberg et al found that ethnicity was not significantly associated with long-term unemployment. ^{Reference Van Dongen18} There has been little research to establish whether these sociodemographic factors are also associated with employment outcomes in the UK, and we aimed to investigate the sociodemographic and clinical characteristics associated with having had a job in the previous year in people with severe mental illness who were being recruited into a randomised controlled trial of supported employment (further information available from the authors). Our hypotheses were as follows.

1. • Participants who have a diagnosis of affective disorder are more likely than those with a diagnosis of schizophrenia to have had some form of employment in the past year.

2. • Participants who have had a higher education are more likely than those who have not to have had some form of employment in the past year.

3. • Younger participants are more likely than older participants to have had some form of employment in the past year.

4. • Black and minority ethnic participants are less likely to have had a job in the past year compared with White participants.

Method

Participants in this study had been recruited for a randomised controlled trial of a supported employment programme: the Supported Work and Needs study (further information available from the authors). The inclusion criteria consisted of severe mental illness (defined as duration of illness over 2 years, a Global Assessment of Functioning score of 60 or below, ^{Reference Nordt, Muller, Rossler and Lauber19} and psychotic or chronic affective disorder), receipt of out-patient or community psychiatric care from local mental health services, age 18-65 years, ability to read and speak English, and unemployment for at least 3 months prior to enrolment in the study. Each participant gave informed written consent.

Participants were recruited from two boroughs in south London: Central Croydon and South Lambeth. The participants' diagnosis was made by a member of the research team at recruitment on the basis of clinical interviews using the Structured Clinical Assessment in Neuropsychiatry (SCAN), ^{Reference Mueser, Salyers and Mueser20} and information from the participants' case notes. At interview the researcher also collected sociodemographic information and gave the participants a job preferences interview, which asked questions concerning the respondent's employment history in the past year, including how many jobs they had had in the past year and the number of months worked, what type of job they ideally would like, and how many hours per week and what hourly payment they would prefer.

Results

Table 1 sets out the baseline characteristics of the sample and univariate associations with having had a job in the past year. Overall, 72% of participants had a diagnosis of schizophrenia, 15% had bipolar disorder and 13% had depression (28% affective disorder). The mean age at baseline for women was 40 years (s.d. = 9.17) and for men it was 39 years (s.d. = 9.57). More men (n = 147, 67%) than women (n = 72, 33%) were recruited into the study. The ethnicity of the sample comprised White (n = 80, 37%), Black (n = 98, 45%), Asian (n = 28, 13%) and ‘other’ (n = 12, 5%). Owing to the small numbers in the sample, ethnicity was recoded into White or Black and minority ethnic. Only 32 (15%) participants in total had had a job in the previous year.

Table 1 Sociodemographic characteristics of the sample and univariate associations with employment in the past year^a

	Had a job in the past year	Did not have a job in the past year	Association with working	Unadjusted OR	P
Research diagnosis, n (%)
Schizophrenia	18 (60)	129 (75)	χ2 = 2.716	1.32	0.10
Affective disorder	12 (40)	44 (25)
Gender, n (%)
Male	25 (78)	121 (65)	χ2 = 2.005	1.89	0.16
Female	7 (22)	64 (35)
Age, years: mean (s.d.)	34.9 (10.6)	40.18 (9.1)	t = 2.988	0.94	0.003
Birthplace, n (%)
UK	22 (69)	128 (69)	χ2 = 0.003	1.02	0.96
Non-UK	10 (31)	57 (32)
Living situation
Living alone (with/without children)	12 (37)	105 (68)	χ2 = 4.072	2.19	0.04
Living with others	20 (63)	50 (32)
Education level, n (%)
Up to secondary education	18 (58)	125 (68)	χ2 = 1.072	1.50	0.30
Tertiary/further education	13 (42)	60 (32)
Ethnicity, n (%)
White	9 (28)	71 (39)	χ2 = 1.328	1.62	0.25
Black and minority ethnic	23 (72)	112 (61)
Language, n (%)
National language	26 (81)	159 (86)	χ2 = 0.591	1.47	0.44
Other language	6 (19)	25 (14)

In the univariate analysis, living situation was associated with having had a job in the past year, with those living alone less likely to have had a job compared with those living as a couple or living with others. There was also a trend towards people with affective disorder being more likely than those with schizophrenia to have had a job in the past year.

Table 2 summarises the results of the logistic regression model for the dependent variable ‘had a job in the past year’. After adjusting for the other variables in the model, it was found that participants who were older were significantly less likely to have had a job in the past year than their younger counterparts (P<0.05). Participants with a diagnosis of affective disorder were significantly more likely (OR = 3.22, 95% CI 1.24-8.31) to have had a job in the past year than those with a diagnosis of schizophrenia (P = 0.016).

Table 2 Associations with working in the past year (multiple logistic regressions)

	Adjusted OR (95% CI)	P
Age	0.942 (0.901–0.985)	0.009
Male v. female	1.981 (0.730–5.385)	0.180
Other language v. national language	1.883 (0.628–5.650)	0.259
Affective disorder v. schizophrenia	3.215 (1.244–8.309)	0.016
Lives with others v. lives alone	1.607 (0.676–3.819)	0.282
Primary/secondary education v. tertiary/further education	1.477 (0.609–3.581)	0.388
White v. Black and minority ethnic	2.103 (0.810–5.463)	0.127

Discussion

We found that younger age and a diagnosis of affective disorder are independently associated with having had a job in the past year. Older age therefore appears to be a barrier to employment. This is consistent with previous research in which younger age is predictive of better vocational outcomes. ^{Reference Wewiorski and Fabian21} We also found that people with a diagnosis of affective disorder (bipolar disorder or depression) are more likely to have had a job in the past year than those with a diagnosis of schizophrenia. Again, this is consistent with previous research, which has indicated that people diagnosed with schizophrenia have worse employment outcomes than those with other diagnoses, and also that people diagnosed with affective disorders have better vocational outcomes generally than others. ^{Reference Barkow, Maier, Ustun, Gansicke, Wittchen and Heun17} Ethnicity and educational level were not significantly associated with employment in the previous year although there was a trend for White participants to be more likely to have had employment compared with Black and minority ethnic participants. This may reflect racial discrimination in society in general, and the effect of mental illness and ethnicity may need further investigation.