Cannabis and psychosis: a further examination using ALSPAC at age 18
Prof. Matt Hickman (DECIPHer)
10.00, Wednesday 10 July. Millennium Lounge, level 5
Background
Although acute cannabis intoxication has been shown to cause transient psychotic experiences (PEs), whether prolonged cannabis use can cause psychotic symptoms is less clear. Systematic reviews have provided some evidence that the relationship is causal, but residual confounding and intoxication effects are hard to completely rule out. Model projections have suggested that the number of cannabis users that need to be treated or prevented to prevent one case of schizophrenia may be considerably high.
Methods
We used data from the Avon Longitudinal Study of Parents and Children (ALSPAC) birth cohort. Substance use at age 16 was assessed via self-report questionnaire. PEs at age 18 were assessed via semi-structured interviews. Confounders (family history, maternal education, IQ, depression, borderline personality traits, strengths and difficulties questionnaire (SDQ), alcohol use at age 16, and other illicit drug use at age 16) were measured variously by questionnaire and interview. Ordered logistic regression analyses were conducted to investigate the associations between cannabis use at 16 and PEs at 18, and tobacco use at 16 and PEs at 18. We also excluded anyone who self reported definite PEs at age 16.
Results
We found cannabis use at age 16 and PEs at age 18 to be associated. Adjustment for pre-birth and childhood confounders did not change the point estimates greatly. Removing those who attribute their PEs only to cannabis intoxication decreased the point estimate, but strong evidence of an association remained. Further adjustment for either tobacco use or other illicit drug use attenuated the relationship substantially. However, both these confounders are highly correlated with cannabis use. We found tobacco use at age 16 and PEs to be associated. Adjustment for pre-birth confounders did not alter the point estimate, but childhood confounders slightly attenuated it. Further adjustment for cannabis use resulted in further attenuation, but not to the same level as for cannabis PE results. The same was also true for other illicit drug use as a confounder.
Conclusions
The relationship between cannabis use and PEs seems to be heavily confounded by other illicit drug use and in particular tobacco use. The relationship between tobacco and PEs appears to be more robust to confounding by cannabis use and other illicit drug use. Other methods are needed to robustly test the independent effects of cannabis, tobacco and other illicit drugs on PEs. Nonetheless, the weakness and non-specificity of the association between cannabis and psychosis undermines hypotheses that the relationship is causal and critically that cannabis prevention is an important target for psychosis and schizophrenia prevention. There are implications also for cannabis drug control which in part was based on concerns over its effect on psychosis.
Although acute cannabis intoxication has been shown to cause transient psychotic experiences (PEs), whether prolonged cannabis use can cause psychotic symptoms is less clear. Systematic reviews have provided some evidence that the relationship is causal, but residual confounding and intoxication effects are hard to completely rule out. Model projections have suggested that the number of cannabis users that need to be treated or prevented to prevent one case of schizophrenia may be considerably high.
Methods
We used data from the Avon Longitudinal Study of Parents and Children (ALSPAC) birth cohort. Substance use at age 16 was assessed via self-report questionnaire. PEs at age 18 were assessed via semi-structured interviews. Confounders (family history, maternal education, IQ, depression, borderline personality traits, strengths and difficulties questionnaire (SDQ), alcohol use at age 16, and other illicit drug use at age 16) were measured variously by questionnaire and interview. Ordered logistic regression analyses were conducted to investigate the associations between cannabis use at 16 and PEs at 18, and tobacco use at 16 and PEs at 18. We also excluded anyone who self reported definite PEs at age 16.
Results
We found cannabis use at age 16 and PEs at age 18 to be associated. Adjustment for pre-birth and childhood confounders did not change the point estimates greatly. Removing those who attribute their PEs only to cannabis intoxication decreased the point estimate, but strong evidence of an association remained. Further adjustment for either tobacco use or other illicit drug use attenuated the relationship substantially. However, both these confounders are highly correlated with cannabis use. We found tobacco use at age 16 and PEs to be associated. Adjustment for pre-birth confounders did not alter the point estimate, but childhood confounders slightly attenuated it. Further adjustment for cannabis use resulted in further attenuation, but not to the same level as for cannabis PE results. The same was also true for other illicit drug use as a confounder.
Conclusions
The relationship between cannabis use and PEs seems to be heavily confounded by other illicit drug use and in particular tobacco use. The relationship between tobacco and PEs appears to be more robust to confounding by cannabis use and other illicit drug use. Other methods are needed to robustly test the independent effects of cannabis, tobacco and other illicit drugs on PEs. Nonetheless, the weakness and non-specificity of the association between cannabis and psychosis undermines hypotheses that the relationship is causal and critically that cannabis prevention is an important target for psychosis and schizophrenia prevention. There are implications also for cannabis drug control which in part was based on concerns over its effect on psychosis.