Analytical approach:
The study was based on a single randomised controlled trial (RCT). Other evidence was available, but a UK-based RCT was needed as there were differences in the labour markets (in terms of unemployment rates) and welfare systems (in terms of benefits paid) in the non-UK studies. The time horizon was one year. It was expected that at one-year follow-up a significantly greater percentage of those who received the IPS would be in competitive employment compared with those receiving the usual care. The authors did not state the perspective.
Effectiveness data:
The effectiveness data were from a RCT and the randomisation was conducted by an independent unit. Randomisation with minimisation was used and treatment allocation was stratified by age (in 10-year groups) and gender. The baseline patient characteristics were presented and did not differ between the two groups. Both the inclusion criteria and the sample size calculations were reported. A total of 375 patients were referred for treatment and 219 of them were randomised to receive either the intervention or control and followed-up for one year. Drop-outs (loss-to-follow-up, withdrawals, and contact lost) were reported. The main measure of effectiveness was the number of people gaining competitive employment. The other clinical outcomes included psychosocial functioning measured by a number of scales. An intention-to-treat analysis was conducted.
Monetary benefit and utility valuations:
Not relevant.
Measure of benefit:
The primary clinical outcome was the proportion of people gaining competitive employment. Secondary outcomes measured psychosocial functioning.
Cost data:
The cost data included the time costs for employment experts, psychiatrists, doctors, district nurses, community mental health nurses, occupational therapists, and psychologists. They also included the costs incurred during psychiatric in-patient stays, day care or education sessions, and general in-patient stays. The costs of services were calculated using national unit costs and the costs of employment were calculated on the basis of a reported annual cost per worker. The resource use was determined by service data and client recall. Confidence intervals around the differences in costs were estimated, using bootstrapping techniques. The costs were in UK pounds sterling (£) and the price year was 2006 to 2007.
Analysis of uncertainty:
Sensitivity analysis was conducted, with the two assumptions that all those lost to follow-up had worked or had not worked.
