Journal club psych

Journal Club 20th Feb 2007

The Effectiveness of Supported Employment for People With

Severe Mental Illness: Randomised Controlled Trial’

Dr J Romain

• The Lancet, Vol 370, September 2007, 1146-52. Burns T., Catty J., Becker T., Drake R E., et al

• Introduction

• Method and Results

• Critical Appraisal

• Discussion

Introduction

• People with severe mental health disorders suffer high unemployment rates

• Often seen as a vicious circle

• 40% of people claiming incapacity benefit claim due to mental health issues

• Substantially increased over last decade

How are we getting patients back to work?

• Traditional methods referred as ‘train-and-place’ model

- addresses deficits related to illness

- train in job skills

- once trained, placed in vocational job, often low skilled and poorly paid

Alternative Method

• ‘Place-and-train’ model- direct job placement on basis of patient preference- Often in simple entry level occupations- no period of training required- ‘on the job’ support for both employer and employee from an employment specialist working as part of the mental health team

Individual Placement and Support (IPS)

• The most intensively studied ‘place-and-train’ model

• Very popular and effective in USA, now recommended evidence based practice

- Randomised trials; rates for competitive employment on open job market more than doubled

Europe and USA Differ Greatly in Employment Practices

USA

• Hire and fire culture

EUROPE• Varying amounts of

employment protection

• More generous welfare systems- DLA, IS, housing benefits…

IPS in Europe

• Therefore would IPS be as effective as it had been in the USA?

• Europe itself contains a lot of variations when looking at welfare systems, job markets and unemployment rates.

Study Aim

• To assess the effectiveness of IPS compared with existing good quality rehabilitation and vocational services for people with severe mental illness in terms of open employment outcomes.

• To examine its effectiveness in different European welfare systems and labour markets

Hypothesis

• Patients assigned to IPS more likely to obtain open employment than the control group.

Primary Outcome• Proportion of people entering competitive

employment (working for at least 1 day)Secondary outcomes; number of hours

worked, days employed, job tenure of employed patients, drop-out from service and admissions to hospital.

Design

• Six European Centres- UK, Germany, Italy, Switzerland, Netherlands and Bulgaria.

• Patient inclusion criteria

- severe mental illness

- aged 18 to local retirement age (60-65)

- major role dysfunction for at least 2 years

- living in the community at baseline

- not been in competitive employment for the past year

Design

• Recruited from April 1st 2003- May 30th 2004. Follow-up ended on Nov 30th 2005.

• Patients randomly allocated to IPS or vocational (control) service.

• Randomisation centrally done with MINIM. Researcher recruited patients and submitted to statistician for randomisation.

• Once services assigned, researchers unable to be blind.

IPS Model• Identify patients wanting to work in

competitive market• Worker helps them to develop realistic

goals and seek appropriate employment directly. All IPS workers are novices. Assessed using ‘IPS fidelity scale’.

• No training phase• Network of employers willing to accept

patients• Worker offers open-ended support to both

employee and employer with a caseload of 25 pts

Vocational Model• Best alternative vocational rehabilitation

service available locally.• Based on ‘train-and-place’ model• Assessed patients rehab needs• Provision of structured training programme

concentrating on deficits related to illness.• Training in appropriate work skills such as

reintroducing daily routine, time management, IT skills

• Vocational services questionnaire (VSQ) designed to capture the nature of service offered and its distinctiveness from IPS

Data Collection

• Patients followed for 18 months, interviews at 6, 12 and 18 months

• Data on vocational outcomes, hospital admission and service use by interview

• Inter-rater reliability assessed periodically with videotaped interviews

• Job satisfaction and hours worked done by questionnaire

Statistical Analysis• 95% confidence intervals were calculated for

primary and other binary outcomes.• Continuous vocational outcomes analysed by

presenting bootstrapped estimates for differences in means and their 95% confidence intervals as they were positively skewed

• Prospective meta-analyses used to explore effect of labour market and welfare system factors on the heterogeneity of the effectiveness of IPS and returning to competitive employment. Done to account for variability in the primary employment outcome across the 6 centres.

Statistics cont!

• Variability caused by certain factors;

- local unemployment rate

- percentage change in GDP

- benefit trap

- indirect income redistribution

• Cochran’s Q test used to examine heterogeneity in outcomes between the centres

Results

• 252 pts (81%) completed final follow up interview

• No difference in baseline characteristics of groups

• IPS workers maintained fair levels of fidelity throughout study. No vocational services achieved near fair level

• VSQ confirmed substantial differences in aims and scope of the different models

Results• IPS more effective than vocational

services for every vocational outcome• Primary Outcome

- 85 (55%) of IPS pts worked for at least 1 day compared to 43 (28%) of vocational(difference 26.9%, 16.4 - 37.4)- IPS pts maintained these jobs for longer (214 days vs 108 days)

• IPS pts worked more hours and were employed for more days

Table Summary of ResultsIPS n Vocational n Difference

(95% CI)Worked for at least 1 day

85 (55%) 156 43 (28%) 156 26.9% (16.4 to 37.4)

Number of hours worked

428.8 (706.77)

143 119.1 (311.94) 138 308.7 (189.22 to 434.17)

Number of days employed

130.3 (174.12)

154 30.5 (80.07) 152 99.8 (70.71 to 129.27)

Job Tenure (days)

213.6 (159.42)

83 108.4 (111.95) 39 104.9 (56.03 to 155.04)

Drop-out from service

20 (13%) 156 70 (45%) 156 -32.1% (-41.5 to -22.7)

Admission 28 (20%) 148 42 (31%) 141 -11.2% (-21.5 to -0.90)

% of time spent in hospital

4.6 (13.56)

148 8.9 (20.08) 141 -4.3 (-8.40 to -0.59)