26 May 2010. A prospective, 21-year study in the May Archives of General Psychiatry has entered the fray over whether cannabis use causes psychosis. John McGrath, Queensland Brain Institute, Australia, and colleagues uphold prior reports that cannabis exposure predicts the development of psychosis-related outcomes. Most importantly, in a field plagued by concerns over potentially spurious correlations, they found that the relationship between cannabis exposure and the emergence of psychotic-like symptoms persisted in sibling-pair analyses designed to control for extraneous variables.
Observations that cannabis can elicit transient symptoms of psychosis date back at least to 1845 and have since been confirmed using various methods, Deepak D’Souza of Yale University, New Haven, Connecticut, told SRF. However, psychotic symptoms do not necessarily indicate psychotic disorder. Recent literature reviews (e.g., D’Souza et al., 2009; McLaren et al., 2010) have shown some evidence that cannabis contributes causally to psychotic disorders, but methodological and other concerns keep researchers from drawing firm conclusions. In fact, a Lancet meta-analysis noted that adjusting for possible confounding factors often lowers estimated correlations between cannabis exposure and psychosis-related outcomes, although even after adjustment, it found a doubled risk of a psychotic outcome in the most frequent users compared to non-users (see SRF related news story).
A step forward
Yet the notion that cannabis causes psychosis remains controversial, partly due to the inability to rule out other explanations for the correlations between exposures and outcomes. “McGrath and his colleagues have rightfully focused on residual confounding and present a clever way at getting at some of those issues,” said Dana March of Columbia University, New York. Because siblings share many genetic and environmental influences, sibling-pair analyses increase confidence that any observed correlations reflect exposure differences.
McGrath and colleagues used data from a large pregnancy study that gathered 21-year follow-up data on 3,801 offspring, with additional data from 228 siblings of cohort members. Based on subjects’ retrospective reports, they determined time since first use of cannabis. Via e-mail, McGrath told SRF that this measures early exposure, but not duration of use because some subjects may have stopped using cannabis.
Analyses tested for associations between time since first cannabis use and each of three psychosis-related outcomes, controlling for age, sex, hallucinations at age 14, and parental mental illness. Outcomes included non-affective psychosis, based on the Composite International Diagnostic Interview (CIDI), and psychotic-like symptoms, as reflected in scores on the Peters et al. Delusions Inventory (PDI) and reports of experiencing auditory or visual hallucinations on the CIDI.
In the big cohort, subjects with six or more years since first cannabis use were more likely than those who had never used cannabis to show psychosis-related outcomes at age 21. The former were twice as likely to meet criteria for non-affective psychosis, four times as likely to score in the highest quartile on the PDI, and three times more likely to have reported hallucinations. These findings did not seem to result from subjects being intoxicated with cannabis when they completed the PDI, in that analyses performed without those who had used the drug in the prior month produced similar results.
Nor did high rates of substance use associated with psychosis explain the correlations (see Winklbaur et al., 2006), although subjects who reported hallucinations at age 14 tended to start using cannabis particularly early. Yet even after the researchers controlled for reverse causality, the link between exposure and outcomes persisted. In fact, said McGrath, “I was struck by how stubborn the associations were,” not like the “fragile and inconsistent findings” he expects in schizophrenia epidemiology research.
The cohort included only 10 sibling pairs in which one member had psychosis, limiting the power to study exposure-outcome relationships in this group. However, in the 218 pairs in which neither sibling had psychosis, McGrath and colleagues found a positive association between time since first cannabis use and the difference in PDI scores between siblings. Specifically, siblings who started using cannabis earlier had higher PDI scores, which McGrath said reflect “psychotic symptoms, but below the clinical level.”
The researchers have been studying the continuum of psychotic-like experiences, which McGrath likens to thinking about high blood pressure or depression. March sees value in this approach because looking at the population distribution of symptoms may offer clues regarding the transition to full-blown psychotic disorder.
Pitfalls, priorities, and promise
On the other hand, D’Souza, speaking about the field as a whole, said that researchers often fail to clearly define the term “psychosis” when speaking at meetings or writing journal articles. He stressed the need to distinguish psychotic symptoms from psychotic disorders. Furthermore, he added, most epidemiological studies probing the link between cannabis use and psychosis development have looked only at positive symptoms and ignored the harder-to-study negative symptoms and cognitive deficits.
According to March, epidemiologists often use crude measures of exposure. “One of the things that I would really like to see come out of this domain of research is a better understanding of the exposure itself,” she said. Emphasizing that not all cannabis exposure is alike, both March and D’Souza want to see more attention paid to such issues as exposure duration, frequency of use, and the cannabis strain used, which may alter the mix of cannabinoids and produce different effects. For instance, D’Souza said that some forms have more cannabidiol, which may actually act as an antipsychotic (see SRF related news story).
Regarding the "big" question, both March and D’Souza harbor reservations about whether cannabis causes psychosis. March said, “For my money, I need to understand more about the association to be convinced that it’s actually causal.”
D’Souza agrees, noting that, “It’s a component cause, if anything,” one among many that may interact with other factors, such as genetic risk, to cause a persistent psychotic disorder. However, in the absence of known causes of schizophrenia, he considers studying component causes very important.
As researchers discuss early interventions to keep high-risk individuals from transitioning to psychosis (see SRF related news story), the question of whether preventing cannabis use might help achieve this goal comes to mind. However, March noted that the evidence supporting a causal role for cannabis pales in comparison to what was known about smoking as a cause of lung cancer before the launch of anti-smoking programs. Still, based on this rather thin evidence, researchers have estimated, on the one hand, that cannabis may cause 14 percent of psychotic outcomes in young adults (Moore et al., 2007) and, on the other, that thousands of heavy cannabis users would need to be kept from using cannabis to prevent just one case of schizophrenia (Hickman et al., 2009).
Instead of merely seeing a research field fraught with uncertainty, however, March looks on the bright side: “This is an area of investigation that’s ripe for a lot of creativity and innovation to try to really understand what’s going on.” Studies like the one by McGrath and colleagues may be a start.—Victoria L. Wilcox.
McGrath J, Welham J, Scott J, Varghese D, Degenhardt L, Hayatbakhsh MR, Alati R, Williams GM, Bor W, Najman JM. Association between cannabis use and psychosis-related outcomes using sibling pair analysis in a cohort of young adults. Arch Gen Psychiatry. 2010 May;67(5):440-7. Abstract