Knitting up the ravelled sleeve of care: sleep and psychosis.

Sleep problems are pervasive in people with schizophrenia. There is a strong documented link between insomnia and psychosis symptoms, and longitudinal studies suggest that insomnia predicts new episodes of paranoia. We cannot assume that standard interventions for insomnia will be successful in psychosis and no trial of such interventions in psychosis has been done. On this basis, Daniel Freeman and colleagues’ study in The Lancet Psychiatry is well founded. This pilot feasibility trial assesses the benefi t of cognitive behavioural therapy (CBT) in the context of insomnia in individuals with a schizophrenia spectrum diagnosis and persistent delusions or hallucinations. Participants were successfully recruited, adhered to treatment, and were followed up to a high level. The CBT intervention was well received by participants, leading to reductions in the primary outcome measure of levels of insomnia 12 weeks after treatment (adjusted mean diff erence 6·1, 95% CI 3·0–9·2, eff ect size d=1·9). Despite reductions in insomnia in the large eff ect size range, there was only a weak indication that the intervention would lead to improvements in psychosis symptoms, including paranoia. This trial comes at a time when conventional CBT for psychosis, although recommended by NICE, has come under scrutiny, and for good reason: the eff ect sizes of this fi rst wave of CBT for psychosis are small to moderate in terms of eff ect on psychosis itself. This outcome was evident nearly 10 years ago; we argued that CBT should concentrate mainly on the aff ective dimension of psychosis (which is after all where CBT cut its teeth) and on well theorised mechanisms. The aff ective dimension in psychosis is increasingly understood, as is its link with psychosis onset and persistence. CBT for psychosis has become a very complex intervention and, because the population under study is heterogeneous, it risks losing impact because the eff ect on individual mechanisms and outcomes is diluted. Freeman and colleagues’ study is an excellent example of the new wave of CBT interventions in psychosis focusing on the aff ective dimension and theoretically driven treatment targets, and for which trials are parsimonious and focused on hypothesised mechanisms. The emphasis on feasibility, acceptability, and eff ect sizes in the present study, rather than on p values, is apposite. Notable is the very high rate of completion of the CBT intervention (96%) and the high acceptability, similar to that achieved in equally focused interventions for posttraumatic stress disorder and command hallucinations in psychosis. The Schizophrenia Commission emphasised the need for new interventions, particularly those with high acceptability, which current, predominantly drug-based, treatments tend not to have. This sleep intervention satisfi es this requirement. However, one of the downsides of focusing on single, but nevertheless important, symptoms is that they are rarely present alone. In this case, sleep diffi culty usually accompanies depression and anxiety. The question is raised as to what extent insomnia is embedded in depression; indeed, many patients in the present study were receiving SSRIs and most were severely depressed. This issue is important from a theoretical perspective, because depression and emotional dysregulation have generally been implicated in the ontogeny of psychosis; moreover, depression and suicidal thinking are, over time, virtually ubiquitous in patients with psychosis. The defi nitive trial should examine depression as a mediator of any eff ect on paranoia. Although the eff ect of CBT on insomnia was in the large range, the eff ect on hallucinations and delusions was small at best. Importantly, the eff ect on quality of life and overall fatigue was in the medium range, underscoring the notion that CBT might be most eff ective when focusing on distress and quality of life. Many individuals with persecutory thinking feel under threat and mitigate this in various diff erent ways, and threat monitoring and hypervigilance can lead to (or perhaps in some cases are the same as) insomnia. Although direct intervention in insomnia will still be a valid therapeutic approach, for some patients, it might increase the perceived threat. In this regard, there might be important subgroups in the case of any eff ect of the intervention on psychosis, and the possibility of distinct responder and non-responder subgroups could be examined in the next trial. Finally, the large variation in eff ect sizes by the method and dimension of sleep assessed in the present study was intriguing, with the Insomnia Severity Index having the largest eff ect size and actigraphy (total sleep time) the weakest. What was the cause of this variation and what do the secondary sleep outcomes measure that the En vi sio n/ Co rb is