Randomised controlled trials

Only three randomised controlled trials (RCTs) evaluating CTOs have been completed. This is understandable given the challenges of ensuring informed consent from patients subject to them and randomising patients into non-CTO arms when they are at serious risk of harm to self or others. Each of the RCTs evaluating CTOs has limitations, including the exclusion of certain patients from study entry,^{Reference Burns, Rugkasa, Molodynski, Dawson, Yeeles and Vazquez-Montes2,Reference Swartz, Swanson, Wagner, Burns, Hiday and Borum3} study sample attrition^{Reference Steadman, Gounis, Dennis, Hopper, Roche and Swartz4} and the confounding effect of providing enhanced care as part of the study design.^{Reference Steadman, Gounis, Dennis, Hopper, Roche and Swartz4} A Cochrane review concluded that the quality of evidence for the RCTs was low to moderate and that CTOs did not result in clear differences for the majority of outcomes delivered.^{Reference Kisely, Campbell and O'Reilly5}

Cohort studies

Cohort studies provide an alternative means of assessing the outcomes of CTOs. Studies of this type use a ‘before and after’ CTO design or compare individuals discharged on a CTO following a compulsory admission with a matched control group discharged under voluntary status. Barnett et al reported that ‘before and after’ studies were associated with a large reduction in admission frequency whereas this was not the case for the matched control studies.^{Reference Barnett, Matthews, Lloyd-Evans, Mackay, Pilling and Johnson6} The ‘before and after’ design is vulnerable to confounding influences, including regression to the mean and maturation effects. Although the use of matched controls allows the influence of time and the natural history of the psychiatric illness to be taken into account, we argue that matching controls on the basis of demographic variables fails to take into account intangible factors such as lack of insight and refusal to accept treatment (variables that are key to the decision to commence a CTO).^{Reference Beaglehole, Newton-Howes, Porter and Frampton7} We therefore completed a series of studies evaluating outcomes associated with CTOs using a ‘within-persons’ design, comparing outcomes on and off CTOs (as opposed to using a control group) for a national cohort placed on CTOs over a 10-year period.^{Reference Beaglehole, Newton-Howes, Porter and Frampton7–Reference Beaglehole, Newton-Howes, Porter and Frampton9} Individuals could be on multiple CTOs over this period. This approach allowed key outcomes associated with CTOs to be compared while they are enacted rather than at a distant time period. We believe that establishing the effectiveness of CTOs while they are in place is required prior to measuring any longer-term effects.

Kisely et al's systematic review

Benefits following community treatment orders have an inverse relationship with rates of use: a meta-analysis and meta-regression by Kisely et al in a recent issue of the BJPsych Open is important because it provides clues on which patient groups may benefit from compulsory treatment.^{Reference Kisely10} CTO use in Australasia was examined to evaluate whether factors associated with CTO use or subsequent outcomes vary according to rates of use. Kisely et al confirm the well-established finding that CTO use varies substantially between regions. They also confirm that CTO use is increased for single males not engaged in work or study and Australians with culturally and linguistically diverse (CALD) backgrounds. Their meta-regression reported that CTO recipients in jurisdictions with higher rates of CTO use were more likely to include females and diagnoses other than non-affective psychoses. High-use jurisdictions were less likely to show reductions in readmission rates or bed-days.

We think that the associations with high-use jurisdictions are intriguing. In our work evaluating associations with CTO use, we reported distinct outcome signatures according to diagnosis. All diagnostic groups received more community care and psychiatric medication, including intramuscular antipsychotic medication, during CTOs, but less frequent admissions and reduced bed-nights per year on CTOs only occurred for patients with psychotic disorders.^{Reference Beaglehole, Newton-Howes and Frampton8,Reference Beaglehole, Newton-Howes, Porter and Frampton9} The opposite occurred for other diagnostic groups. In our research, males but not females showed a significant reduction in admissions when on CTOs.^{Reference Beaglehole, Newton-Howes and Frampton8} We also reported that mortality rates were higher during CTOs compared with non-CTO periods.^{Reference Beaglehole, Newton-Howes, Porter and Frampton7} This finding diverged from other studies that compared mortality rates following CTO initiation with matched controls. The divergent finding may relate to the choice of comparator in CTO studies. It is feasible that CTO recipients have lower rates of mortality compared with matched non-CTO recipients and higher rates of mortality during CTOs compared with non-CTO periods. We regard these findings to be informative about the impacts of CTOs while they are in place. However, these were ‘real-world’ studies and further inferences, such as recommending more extensive use of CTOs for psychotic disorders because of the associations reported, should not be drawn from our studies.

Much of the previous research has focused on evaluating whether CTOs are associated with improved outcomes. The review by Kisely et al and our research suggest that reframing the research question is helpful. CTOs have different outcome signatures according to diagnosis and perhaps other demographic variables. Our research supported the use of CTOs to reduce admissions for people with psychotic disorders. For people with other diagnoses, it is possible that increased admissions on CTOs may be regarded favourably (for example, to provide electroconvulsive therapy for depression or containment for mania); however, some would regard this as a negative outcome. We suggest that clinicians integrate likely outcomes associated with CTOs according to diagnosis to inform conversations with patients and family.

We also think that the marked variation in CTO use confirmed by Kisely et al is of interest. Although this finding is well-recognised, the reasons underpinning variation are not well-understood. We have completed an analysis of regional variation in CTO use in New Zealand.^{Reference Lees, Newton-Howes, Frampton and Beaglehole11} This analysis controlled for underlying factors known to be associated with CTO use (age, gender, ethnicity and deprivation) to determine whether these factors explained the extent of variation. We concluded that sociodemographic variance between regions did not explain the wide range of CTO utilisation in New Zealand and that institutional culture and practice differences accounted for the variation in CTO use.