Reference Coughlan and CannonCoughlan & Cannon (2017, this issue) discuss a series of recent meta-analyses that suggest a connection between childhood trauma and the experience of psychotic symptoms. They highlight the evidence that childhood sexual abuse appears to be the most consistently associated form of trauma. Although this is interesting for the practising psychiatrist, much of the evidence they present is derived from surveys conducted in the general population rather than in patients with confirmed psychotic illness. The clinical significance of these findings is therefore dependent on the assumption that there is a continuum between psychotic-like experiences and frank psychosis, such that the voice-hearing commonly found in the general population does not differ qualitatively from the distressing auditory verbal hallucinations of schizophrenia. The authors describe a particular version of this model where it is not the psychotic symptoms themselves but their persistence and transition to impairment that characterises psychotic illness (e.g. Reference van Os, Linscott and Myin-Germeysvan Os 2009). While this approach is undoubtedly useful for epidemiological studies, it is unclear how it relates to clinical practice, where the detection and discrimination of psychotic symptoms remains at the core of diagnosing psychotic illness (Reference DavidDavid 2010; Reference Lawrie, Hall and McintoshLawrie 2010). Many studies comparing partially treated psychosis patients and non-psychotic voice-hearers have downplayed any qualitative differences in phenomenology, but there is a large difference in emotional valence and distress between the experiences of psychotic and non-psychotic individuals (e.g. Reference Daalman, Diederen and DerksDaalman 2012) and it seems more likely that the crude instruments used in research fail to adequately capture the phenomenology of true psychotic symptoms (Reference Upthegrove, Broome and CaldwellUpthegrove 2016a), as is suggested by cognitive (e.g. Reference Garrison, Moseley and Alderson-DayGarrison 2017) and neurobiological studies (e.g. Reference Howes, Shotbolt and BloomfieldHowes 2013).

If we consider only those meta-analyses looking at the association between childhood trauma and the development of clinically diagnosed psychotic illness there are actually only two of relevance. Reference Varese, Smeets and DrukkerVarese et al (2012) found a link between childhood trauma and psychosis in case–control and prospective studies. Focusing specifically on schizophrenia, Reference Matheson, Shepherd and PinchbeckMatheson et al (2013) looked at cohort, case–control and cross-sectional studies to show an increased rate of childhood trauma compared with healthy controls. However, this association was not specific and they found significantly greater rates of childhood trauma in dissociative disorders and post-traumatic stress disorder (PTSD) than in schizophrenia and no difference from depression or personality disorders. The rate in schizophrenia was greater than that in anxiety disorders (although there was not a greater rate of sexual abuse).

Looking specifically at the studies of childhood sexual abuse, which Coughlan & Cannon consider to be one of the most potent traumas associated with psychosis, there are few focusing purely on the risk of developing a clinical diagnosis of psychosis. Reference Daalman, Diederen and DerksDaalman et al (2012) in a case–control study found increased rates of childhood sexual abuse in patients with psychosis and auditory hallucinations compared with healthy controls, but not compared with healthy voice-hearers. Similarly, Reference Sheffield, Williams and BlackfordSheffield et al (2013) in a case–control study found increased scores on the Childhood Trauma Questionnaire (CTQ) sexual abuse subscale (but not increased overall trauma scores) in those with psychosis compared with healthy controls. In patients at ‘ultra high risk’ of psychosis there are two studies looking at whether a higher rate of childhood sexual abuse predicts transition to a first episode of psychosis. Reference Thompson, Nelson and Pan YuenThompson et al (2014) found that total score on the CTQ did not predict transition to psychosis but the score on the sexual abuse subscale did. Reference Bechdolf, Thompson and NelsonBechdolf et al (2010) measured a history of trauma (primarily in childhood, given that the average age of their sample was 18 years) and found no statistically significant effect of sexual trauma or all trauma unless they made a questionable adjustment for the number of ultrahigh-risk categories a patient fulfilled. Both these studies used the Comprehensive Assessment of At-Risk Mental States (CAARMS) interview to determine transition to first-episode psychosis, but this may well not be equivalent to a clinical diagnosis of a psychotic illness (Reference Yung, Nelson and ThompsonYung 2010).