The OCTET RCT – a reply to Dr Mustafa

Dr Mustafa points out what he considers limitations to the OCTET trial. The trial concluded that since community treatment orders did not confer patient benefits measured over 12 months but significantly restricted personal liberty, we should reconsider their use. No study is perfect, and we outlined limitations to OCTET in the original paper. Dr Mustafa’s letter provides us with the opportunity to address this further and correct misunderstandings. Dr Mustafa suggests that being a clear candidate for a community treatment order (CTO) was an exclusion criterion, limiting external validity. This is not correct. Rather, the criterion was that clinicians had to be willing for patients’ treatment to be randomised, that is, they would need to be in equipoise to recruit to the trial. This is standard RCT procedure and part of the Helsinki declaration. Those who recruited to OCTET accepted that there was no clear scientific evidence and no clinical practice for clinicians to draw on at the time when CTOs were introduced to England and Wales. The clinical characteristics of those recruited to OCTET closely match those reported in other studies and there is no reason to believe they were a particularly narrow or atypical group. Dr Mustafa also suggests that a ‘learning curve effect’, whereby clinicians changed their practice during the trial period as they gained experience of the new regime, was responsible for protocol violations with potential effect on internal validity. Protocol violations were, however, spread across the entire period. While there were more than we had hoped for, this level is not unusual in such pragmatic clinical effectiveness trials. A sensitivity analysis excluding the violations confirmed the ITT analysis: the results are the same. Moreover, qualitative analyses in progress suggest differences in clinical practice even after several years of the CTO regime, so the learning curve hypothesis does not stand. OCTET was an effectiveness trial and not an efficacy trial. An effectiveness trial tests how an intervention works in the real world – one in which patients’ needs change, services are reconfigured and funding to mental health services altered. Unlike efficacy trials, which require optimal conditions and standardisation, effectiveness trials require wide patient and clinician recruitment, which OCTET achieved. Blinding in social psychiatry trials is tricky. Itwould, of course, be impossible to blind clinicians or patients in a trial in which knowledge of the intervention is part of the intervention itself. Moreover, it would be unlawful not to inform each party of legal status. We agree that clinicians’ views impact on use of CTOs, which is why this was designed as an effectiveness trial. The primary outcome (rates of readmission) is derived from clinical notes and not subject to rater bias. As regards the effect of the varying threshold for recall, this is not relevant here as we did not include recall admissions (allowed for periods up to 72 h for assessment or medication) in the readmission rates as we consider them not an outcome, but a part of the CTO process. Our recommendation of reconsidering the use of CTOs in their current form is not informed by OCTET alone, but the other two RCTs and all literature reviews to date. We disagree that we should venture down the hierarchy of evidence: this would not be in the interest of vulnerable patients treated against their will.