An accumulation of evidence indicates that early adverse experiences could enhance the risk of developing psychosis (Read et al. Reference Read, van Os, Morrison and Ross2005). In addition, several meta-analyses support the idea that exposure to Δ⁹-tetrahydrocannabinol, the psychoactive component of cannabis, may produce an increased risk of psychosis (e.g. Arseneault et al. Reference Arseneault, Cannon, Witton and Murray2004; Fergusson et al. Reference Fergusson, Poulton, Smith and Boden2006; Large et al. Reference Large, Sharma, Compton, Slade and Nielssen2011). However, for numerous reasons, children and adolescents with poor mental health may be more likely than others to experience trauma or to go on to use drugs (Hawkins et al. Reference Hawkins, Catalano and Miller1992). Thus, initial mental health must be taken into consideration when evaluating if environmental risk factors such as childhood adversity or cannabis use contribute to psychosis.

Surprisingly, this is often not the case (MacLeod & Hickman, Reference MacLeod and Hickman2010). A recent article published in Psychological Medicine neglected to consider initial psychotic symptoms or mental health more generally when demonstrating a large association between early experiences of non-consensual sex and psychosis in adulthood, particularly amongst cannabis users (Houston et al. Reference Houston, Murphy, Shevlin and Adamson2011). It is difficult to ascertain from the extant literature how omitting a measure of initial mental health may affect the findings reported by Houston et al. (Reference Houston, Murphy, Shevlin and Adamson2011). I therefore utilized secondary data to identify if adjusting for the presence of hallucinations during childhood might diminish associations between non-consensual sex, cannabis use and experiencing hallucinations in adulthood.

Of 3649 participants with usable data from the 1970 British Birth Cohort sample (Elliot & Shepherd, Reference Elliott and Shepherd2006), 58.9% of males and 40.2% of participants reported having tried cannabis by age 29 years. Few males had experienced non-consensual sex (0.1%) whereas 1.6% of the female sample indicated they had been forced to have sex by age 16 years. In logistic regression analyses, which adjusted for the participant's socioeconomic background, females who had experienced non-consensual sex by age 16 years were at an elevated risk of visual and auditory hallucinations at age 29 years [χ²=3.8, p=0.05, odds ratio (OR) 8.51, 95% confidence interval (CI) 0.99–73.28]. No such link was found for males who had experienced non-consensual sex, nor was there any evidence for either males or females that cannabis use interacted with non-consensual sex to produce hallucinations in adulthood. As expected, females who experienced hallucinations by age 16 years were at high risk of experiencing non-consensual sex in the same period (χ²=7.29, p=0.007, OR 28.81, 95% CI 2.52–330.38). This finding opens up the possibility that the association between early non-consensual sex and later mental health problems may be confounded by initial psychotic symptoms. In support of this idea, adjusting for initial psychotic experiences eliminated any link between non-consensual sex in females and later visual or auditory hallucinations (χ²=0.29, p=0.59, OR 2.43, 95% CI 0.09–62.88).

The empirical illustration above, although limited to hallucinations, does indicate that the findings of Houston et al. (Reference Houston, Murphy, Shevlin and Adamson2011) should be interpreted with caution. Further research is needed to test if non-consensual sex in childhood combines with cannabis use to place people at enhanced risk of later psychotic experiences over and above the presence of initial psychotic symptoms or poor mental health. In the past, methodologically more robust studies have taken the approach of adjusting for initial symptoms of psychosis or excluding participants with a diagnosis of psychosis at baseline (e.g. see Moore et al. Reference Moore, Zammit, Lingford-Hughes, Barnes, Jones, Burke and Lewis2007). Whilst this step is useful, subclinical psychotic experiences and poor mental health should also be adjusted for. This is because subtle individual differences in mental health amongst those without psychosis at baseline could be the primary determinant of later psychosis.

The study by Houston et al. (Reference Houston, Murphy, Shevlin and Adamson2011) relied on retrospective accounts of environmental exposures and due to the cross-sectional nature of the data collected could not assess the subjective experiences of participants at baseline. However, an attempt could have been made to elicit retrospective information relating to psychotic symptoms (e.g. age of onset, age of first diagnosis if applicable). This information would have helped the researchers to gauge the likelihood that sexual abuse and cannabis use may combine to produce an independent risk of psychosis. Future research would contribute key policy-relevant insights by clarifying the temporal interrelationships between childhood trauma, cannabis use and early mental health and the role of each factor in the development of psychosis (Nelson & Mann, Reference Nelson and Mann2011).