Poorer patient views of mental health inpatient treatment predict both further admissions and, for those admitted involuntarily, longer admissions. As advocated in the UK Francis report, we investigated the hypothesis that improving staff training improves patients’ views of ward care.

Cluster randomised trial with stepped wedge design in 16 acute mental health wards randomised (using the ralloc procedure in Stata) by an independent statistician in three waves to staff training. A psychologist trained ward staff on evidence-based group interventions and then supported their introduction to each ward. The main outcome was blind self-report of perceptions of care (VOICE) before or up to 2 years after staff training between November 2008 and January 2013.

In total, 1108 inpatients took part (616 admitted involuntarily under the English Mental Health Act). On average 51.6 staff training sessions were provided per ward. Involuntary patient's perceptions of, and satisfaction with, mental health wards improved after staff training (N582, standardised effect −0·35, 95% CI −0·57 to −0·12, p = 0·002; interaction p value 0·006) but no benefit to those admitted voluntarily (N469, −0.01, 95% CI −0.23 to 0.22, p = 0.955) and no strong evidence of an overall effect (N1058, standardised effect −0.18 s.d., 95% CI −0.38 to 0.01, p = 0.062). The training costs around £10 per patient per week. Resource allocation changed towards patient perceived meaningful contacts by an average of £12 (95% CI −£76 to £98, p = 0.774).

Staff training improved the perceptions of the therapeutic environment in those least likely to want an inpatient admission, those formally detained. This change might enhance future engagement with all mental health services and prevent the more costly admissions.