In his article Curtis ^{Reference Curtis1} highlights one of the limitations of the OCTET study, ^{Reference Burns, Racks, Molodynski, Dawson, Yeeles and Vazquez-Montes2} in that patients selected for randomisation may not have been suitable for community treatment order (CTO) placement in the first place. In his conclusions he suggests there may be a small subgroup of patients for whom CTOs are enormously beneficial. Perhaps clinicians need more clarity of the characteristics of the ‘revolving door’ patient better to assess suitability for supervised community treatment.

Most clinicians will have a personal construct of the epidemiological and clinical characteristics of revolving door patients, although this may not be explicitly defined. There is no consistency in the literature as to the definition of revolving door, and previous research in the UK has shown that predictors of readmission are varied and not consistently replicated across studies. Research carried out when the practice of ‘long leash’ Section 17 leave was widespread showed that those placed on extended leave had a history of more frequent compulsory admissions, increased recent dangerousness to others, and decreased adherence to their out-patient follow-up prior to admission. ^{Reference Sensky, Hughes and Hirsch3}

A case-control study was conducted at Leeds Partnership NHS Foundation Trust in 2010, and approved by the local research and development department as a service evaluation. The aim was to compare characteristics of patients placed on CTOs and those discharged from Section 3, to elicit which factors were associated with CTO placement. All patients placed on a CTO between November 2008 and February 2010 were included as cases, and controls were randomly selected from patients who had been detained under Section 3 of the Mental Health Act, but whose Section was rescinded within the same week that the CTO was commenced. A ratio of two controls for each case increased the power of the study. This amounted to 56 cases and 112 controls. Characteristics chosen for analysis were those which previous research had suggested may be of importance and where collection was feasible. The characteristics of the patients placed on CTOs were broadly similar to those recruited into OCTET.

Analysing variables individually, patients on CTOs were significantly more likely (P<0.05) to be single, have a principal diagnosis of schizophrenia, a history of violence, a higher number of previous admissions, a history of criminal conviction and a higher number of convictions within the past year. On logistic regression analysis, patients on CTOs were significantly more likely to have a principal diagnosis of schizophrenia and a higher number of previous admissions.

There remains the outstanding question of who belongs to the elusive group of patients for which CTOs are effective, if indeed this group exists. This study provides insight into the demographic and historical factors that are influencing clinicians' decisions to implement CTOs. There is no proof so far that CTOs are effective in their aims. Perhaps we need to look again at who the truly ‘revolving door’ patients are and take this objective evidence into consideration at the point of deciding whether to initiate supervised community treatment.