BackgroundRecovery in mental health is a relatively new concept, but it is becoming more accepted that people can recover from psychosis. Recovery-orientated services are recommended for adult mental health, but with little evidence base to support this.ObjectivesTo facilitate understanding and promotion of recovery in psychosis and bipolar disorder (BD), in a manner that is empowering and acceptable to service users.MethodThere were six linked projects using qualitative and quantitative methodologies: (1) developing and piloting a service user-defined measure of recovery; (2) a Delphi study to determine levels of consensus around the concept of recovery; (3) examination of the psychological factors associated with recovery and how these fluctuate over time; (4) development and evaluation of cognitive–behavioural approaches to guided self-help including a patient preference trial (PPT); (5) development and evaluation of cognitive–behavioural therapy (CBT) for understanding and preventing suicide in psychosis including a randomised controlled trial (RCT); and (6) development and evaluation of a cognitive–behavioural approach to recovery in recent onset BD, including a RCT of recovery-focused cognitive–behavioural therapy (RfCBT). Service user involvement was central to the programme.ResultsMeasurement of service user-defined recovery from psychosis (using the Subjective Experience of Psychosis Scale) and BD (using the Bipolar Recovery Questionnaire) was shown to be feasible and valid. The consensus study revealed a high level of agreement among service users for defining recovery, factors that help or hinder recovery and items which demonstrate recovery. Negative emotions, self-esteem and hope predicted recovery judgements, both cross-sectionally and longitudinally, whereas positive symptoms had an indirect effect. In the PPT, 89 participants entered the study, three were randomised, 57 were retained in the trial until 15-month follow-up (64%). At follow-up there was no overall treatment effect on the primary outcome (Questionnaire about the Process of Recovery total;p = 0.82). In the suicide prevention RCT, 49 were randomised and 35 were retained at 6-month follow-up (71%). There were significant improvements in suicidal ideation [Adult Suicidal Ideation Questionnaire; treatment effect = –12.3, 95% confidence interval (CI) –24.3 to –0.14], Suicide Probability Scale (SPS; treatment effect = –7.0, 95% CI –15.5 to 0) and hopelessness (subscale of the SPS; treatment effect = –3.8, 95% CI –7.3 to –0.5) at follow-up. In the RCT for BD, 67 participants were randomised and 45 were retained at the 12-month follow-up (67%). Recovery score significantly improved in comparison with treatment as usual (TAU) at follow-up (310.87, 95% CI 75.00 to 546.74). At 15-month follow-up, 32 participants had experienced a relapse of either depression or mania (20 TAU vs. 12 RfCBT). The difference in time to recurrence was significant (estimated hazard ratio 0.38, 95% CI 0.18 to 0.78;p < 0.006).ConclusionsThis research programme has improved our understanding of recovery in psychosis and BD. Key findings indicate that measurement of recovery is feasible and valid. It would be feasible to scale up the RCTs to assess effectiveness of our therapeutic approaches in larger full trials, and two of the studies (CBT for suicide prevention in psychosis and recovery in BD) found significant benefits on their primary outcomes despite limited statistical power, suggesting definitive trials are warranted.FundingThe National Institute for Health Research Programme Grants for Applied Research programme.