Occupational therapy interventions in facilitating return to work in patients with traumatic brain injury: A systematic review.

Nothing was pooled. The review describes 15 studies one at a time, so the figures cannot be combined, and the positive conclusion rests on a handful of them. The randomised trials with return to work as the outcome were negative at their main time point, and the positive work figures come from a non-randomised cohort, a non-randomised comparison and a single case report, none of which can show the intervention caused the change. The evidence is small and uneven. Several studies had fewer than 20 participants and one had a single person, five studies were rated Level 4b, and no study was excluded because of its evidence level. Several studies were not blinded and relied on self-reported outcomes. Most interventions were packages given by whole teams, so what occupational therapy contributed on its own cannot be separated from physiotherapy, neuropsychology, vocational counselling and supported employment. The settings differ from most readers' services: three studies were run in Cape Town, one in Denmark and one in Hong Kong, and the authors note that Hong Kong was said to have insufficient employment services for people with traumatic brain injury. At least one quality rating does not fit the design it was applied to: a randomised trial is described in the results as assessed with the qualitative CASP checklist.

The authors state that they received no financial support for the research, authorship or publication of the article, and that they declared no potential conflicts of interest.

Do not read this review as evidence that occupational therapy vocational or cognitive interventions get people back to work. The largest trial with return to work as its main outcome, 116 people, found no difference between groups at 6 or 12 months; a randomised trial of 40 people found no vocational advantage from virtual reality training; and the positive work figures come from a non-randomised Danish cohort of 32 people and a single case study of one man. The review describes its own evidence as only moderate strength.