Recent research into whether cannabis causes "psychosis", mental illness, or schizophrenia has received a great deal of attention in the media and has triggered calls for the British government to move cannabis from Class C back to its former, more punitive, Class B designation [Erowid Note: Cannabis was reclassified as Class B in January 2009].

This is, of course, not the first time people have said that cannabis causes mental illness, with assertions of a connection going back at least as far as the British Indian Hemp Drugs Commission's investigation into similar charges in 1893. It concluded that "[when the] alleged evil effects of the moderate use of [cannabis are] subjected to careful examination, the grounds on which the allegations are founded, prove to be in the highest degree defective."1

Medical and epidemiological research has come a long way since 1893, or so we'd like to believe, and many papers have been written about what physical and mental harms might be caused by the use of cannabis. In recent years, there has been a flurry of news regarding the analyses of several large longitudinal surveys that show an apparent link between cannabis use and mental illness: "The link between regular cannabis use and later depression and schizophrenia has been significantly strengthened by three new studies..."2

It is easy to dismiss the current Reefer Madness scare as yet another in a long line of such offerings by politically-driven drugabusologists, but there is more to it than hype, even if it is hardly the "drug induced mental health crisis"3 that some allege. The following is an overview of the current state of the debate.

In the last few years there have been dozens of journal articles touching on this issue. A PubMed search for "cannabis psychosis" shows over 60 related articles published since the start of 2004. New additions to the now massive literature include a handful of important papers that present original analyses of survey data, at least three systematic literature reviews (summarizing, comparing, and synthesizing the results of other research), numerous editorial "reviews" of the evidence, and at least one academic book.

The first thing to know about this topic is that it is indisputable that there is a correlation between the repeated use of cannabis and a variety of mental health issues.4,5

Many large scale studies conducted in the last 30 years have found that people who have clinically-meaningful symptoms of schizophrenia or other psychotic disorders tend to use cannabis more often than apparently similar members of the general population. These findings were replicated in two major recent papers by Henquet et al. and Fergusson et al. (see Recent Articles sidebar).6,7

This same data can be read to suggest the opposite, that cannabis users are more likely to have psychotic symptoms than the general population. "Drug abuse" is said to be "comorbid" with schizophrenia and other psychotic disorders, which means that these two problems occur in the same person at a higher than average rate. Comorbidity does not imply any causal linkage, although comorbid conditions may be causally related. Unfortunately, from there, it gets very complicated very quickly.

This bears repeating: correlation does not imply, and should not be assumed to mean, causation. This is especially true in epidemiological surveys, which are based on self-reports. The indirect nature of these studies—compared to controlled experimental designsÑmakes it infeasible to show actual causation with much certainty.

One of the problems I first encountered when trying to understand these issues was figuring out what exactly "psychosis" refers to. The article by Fergusson et al., one of the important recent analyses suggesting a causal relationship between cannabis and psychosis, focuses on "psychotic symptoms" and reports that cannabis "may increase risks of psychosis". But there is no single definition of what constitutes "psychosis". The DSM-IV (the standard Diagnostic and Statistical Manual for psychiatry) defines a variety of psychological problems as "psychotic disorders". Psychotic symptoms include hallucinations, paranoia, and delusions, which are characteristic of schizophrenia, the most common and widely known of the psychotic disorders. However, there is no universal method for diagnosing psychotic disorders, and each of the surveys related to cannabis and psychosis used different diagnostic methods. This makes simple comparisons difficult. When reviewing the current literature, Macleod et al. noted, "Although some measures were similar across studies, no two studies measured either illicit drug exposure or psychosocial outcome in the same way."4

Perhaps most important to the current discussion is that both the Fergusson and Henquet papers identify "psychotic symptoms" from self-reports on surveys. These self-reported "psychotic symptoms" include items such as "having ideas or beliefs that others do not share" and do not reflect actual diagnosed, clinically relevant schizophrenia or psychotic disorders.

Self-report-based survey data are particularly problematic when specific medical conclusions are based on them. Theoretically speaking, survey data, no matter how strong, cannot prove causation. Survey data can only imply causation by isolating and using statistical techniques to control for possible confounds, e.g. factors other than cannabis use that could cause psychotic symptoms in the survey respondents. When the cause of a disease is unknown, as is the case with psychotic disorders, determining which confounds to control for becomes an evolutionary process of educated guesswork.

Survey data can also present additional complications. It appears that people are more willing to report the use of cannabis and alcohol than they are the use of other illegal drugs.8 This would increase the apparent effect of cannabis relative to other psychoactive drugs. Findings could be exaggerated if people with mental problems admit to illegal drug use at different rates than healthier individuals (they may be seeking causes for their condition or just have less to lose). It may be that people willing to admit to cannabis use are also more willing to admit to strange thoughts. It is also a known issue that a certain small portion of the population simply lies on surveys, and the bias this could introduce is not examined in the respective studies.

In their literature review from May 2004, Macleod et al. state that "In other contexts, reporting bias has been shown to be capable of generating strong and substantial associations between measures ..."4 Simply put, it is problematic and difficult to separate correlation from causation using survey data.

Although the studies on cannabis and psychosis attempt to delineate and account for confounds, real world epidemiological research into social and psychological phenomena is complicated. Issues such as family history or childhood traumas play a significant role in schizophrenia and are difficult to separate from other issues such as cannabis use. According to Macleod:

"Discounting confounding is probably the most serious interpretational challenge in observational epidemiology...cannabis use could be a marker, rather than a cause, of a life trajectory more likely to involve adverse outcomes. There are no completely reliable means to identify confounded associations within observational data, and instances where apparently robust observational evidence has later been shown to be seriously misleading are common. The importance of this issue to the epidemiology of drug use might have been underestimated."4

Fergusson et al. agree: "The largest threat to the validity of causal conclusions in this area comes from the possibility of uncontrolled residual confounding."7 For many of the major surveys, apparent correlations were much higher before known confounds were factored in. Unidentified confounds could be equally important.4,5,9

A separate issue is that cannabis use can occasionally cause powerful experiences that trigger bizarre feelings and behaviors. It is well documented that some people, at some doses, can and do react to the effects of cannabis with behavior that appears psychotic, paranoid, and delusional. These temporary effects during acute intoxication should not be confused with real psychotic disorders, as the effects generally resolve when the cannabis wears off.10

One of my thoughts while reading on this topic has been that since, cannabis use rates have reportedly risen dramatically over the last 40 years,8 there should be a matching increase in psychosis rates. This turned out to be harder to research than I had imagined. The diagnostic criteria for schizophrenia and "psychosis" have changed quite a bit over the last 40 years, making it difficult to compare rates over time. Also, reliable epidemiology for these psychological problems is scant.

The most that can be said with certainty is that mental health facilities have not reported clear increases in psychotic disorders. There are a number of estimates of the prevalence of these disorders, most of which range from 0.7-2.0%, though others estimate as high as 5%. The generally accepted standard appears to be around 1%.11,12

There is evidence that the rate of schizophrenia and psychotic disorders has "either been stable or slightly decreased" over the past 40 years.4 Degenhardt et al. (2003) tested the hypothesis that cannabis use caused an increase in cases of schizophrenia that would not have otherwise occurred. They examined the rates of schizophrenia and related outcomes in Australia among populations born between 1940 and 1979 and found no increase in psychotic disorder rates. They concluded that a direct causal connection was unlikely:

"The hypothesis that cannabis causes schizophrenia was not supported by the data on trends in the incidence of this psychosis in Australia. There was no evidence that there has been an increase in incidence over the past 30 years of the magnitude predicted by the hypothesis..."11

A frequently mentioned explanation for the association between cannabis use and schizophrenia is that people with this type of mental illness self-medicate with cannabis to reduce symptoms or make themselves feel better. Recent studies have attempted to establish a direction of causation and control for the Self Medication Hypothesis (SMH) by carefully comparing the timing and order of cannabis use with the timing of reported symptoms. Both Fergusson et al. (2005) and Henquet et al. (2004) address this and report that the SMH did not explain their findings.

There is now substantial evidence that the SMH cannot explain all (or even most) of the correlation between the use of cannabis and schizophrenia and psychotic symptoms. But there is also undoubtedly some part of the association that is properly described as self medication, but this has not yet been quantified.

An alternative framing of self medication is "affect regulation", which is a more general way to describe the reasons someone might want to use a substance to alter how they feel. Instead of focusing solely on the idea of medicating something, the term "affect regulation" also includes social, philosophical, or recreational reasons for use.

Multiple surveys have looked at the reasons people diagnosed with psychotic disorders give for their use of cannabis. In her review of these surveys, Spencer (2004) found a strong correlation between certain reasons for use and how much was used: "[the use of cannabis and other substances] to cope with unpleasant affect and to relieve psychotic symptoms and medication side-effects led to stronger dependence."13 It may be problematic to assume that the dose-response findings in the surveys (where heavier cannabis use is associated with more psychotic symptoms) indicate cannabis causes symptoms to worsen. There is evidence that at least some users increase their cannabis intake in response to worsening symptoms and to curtail the unpleasant side effects of medications.

Related to the Self Medication Hypothesis is the issue of cannabis use during the "prodromal" period. Prodromal refers to symptoms that occur before the full onset of a disease. There is evidence that most psychotic disorders are preceded by an extended prodromal period that often includes a variety of other symptoms.14,15 It is possible that, during this period, some individuals turn to commonly available psychoactives to improve their emotional states. Since the developmental course of psychotic disorders is not clearly understood, the hypothesis that pre-symptomatic individuals prone to psychosis may have a proclivity for seeking cannabis cannot be easily separated from the hypothesis that cannabis use prior to the onset of psychosis causes the later symptoms to occur.

Some studies claim that they can detect those experiencing prodromal effects by their survey answers, but the question of whether survey answers can identify early prodromal individuals has not been adequately explored: many early symptoms may not appear to be psychotic. This makes it very difficult to identify whether cannabis is a cause or an effect of psychotic symptoms.

Another possible explanation for the correlation between cannabis use and psychosis is that the same risk factors that lead to psychotic disorders also lead to heavy cannabis (or other similar psychoactive) use. An oversimplification of this "shared cause" (also called "common vulnerabilities") model is that the genetic background and childhood histories that make people prone to schizophrenia also increase the likelihood of cannabis use. This is similar to, but distinct from, the issue of confounding factors.

There is evidence that the cannabinoid system in the human brain may be involved with schizophrenia and that schizophrenic patients have imbalances in their endogenous cannabinoids.16 It is possible that an underlying condition could both lead to schizophrenia and make it more likely for an individual to use cannabis more heavily because it affects them differently than the average population.

Ferdinand et al. (2005) found that, in a longitudinal study of more than 1,500 respondents, psychotic symptoms and cannabis use appeared to cause each other. "It is remarkable that, in the present study, links between psychotic features and cannabis seemed to run in both directions, from cannabis use towards psychotic symptoms, and vice versa. This might indicate that a common type of vulnerability factor is responsible for the association found."17

Since schizophrenia has a higher average age of onset (21-32 years old)18 than the average age of initiation for cannabis, by default, cannabis would normally be used before the onset of schizophrenic symptoms. According to national surveys in the United States, peak cannabis use occurs between the ages of 18 and 25.8 This overlaps the period during which schizophrenia is likely to first occur, providing an additional complication.

Degenhardt et al. (2003) propose that if cannabis is a precipitating component cause, as the age of initiation of cannabis goes down, so should the age of onset for schizophrenia. Unfortunately the available data does not clarify the matter: "The limited evidence on the average age of onset of schizophrenia makes it difficult to draw any conclusions about this indicator. [There is] some evidence that the age of onset of schizophrenia has decreased ... but clinical samples of first-episode psychosis have not consistently found that cannabis use is associated with an earlier onset of psychosis."11 Other studies have documented a relationship between lower age of first psychotic episode and cannabis use, supporting the precipitating hypothesis, but further research is necessary.19

There are several distinct types of "causes": necessary, sufficient, and component. A necessary cause is one that must be present before the result can occur. A sufficient cause is one that is, by itself, enough to cause the result. A component cause is one that is part of a "constellation" of causes that work together to bring about a result.

All of the recent research into this issue has found that cannabis is neither necessary nor sufficient to cause schizophrenia by itself.5 Rather, it is most likely a component factor when combined with a variety of other potential issues such as genetic pre-disposition or difficult childhood. Cannabis use, along with its attendant lifestyle and subculture, probably worsens symptoms in some of those vulnerable to psychotic disorders. The same is commonly said of the psychedelics such as LSD or psilocybin. As we don't know what causes schizophrenia, determining how important a factor cannabis use is cannot be determined.

Cannabis use could be a precipitating component in several ways. Heavy cannabis use may worsen certain coping mechanisms and weaken familial or social support through social disapproval. Fear of legal problems could cause anxiety and paranoid feelings; actual legal problems could damage self-esteem, damage hopes for life goals, or cause serious life problems. Cannabis intoxication can lower inhibitions and self control, worsen some symptoms to the point of dysfunction, or be part of a lifestyle that involves irregular sleep or other instability. It is easy to imagine that the effects of frequent or high dose cannabis intoxication could exacerbate these underlying instabilities and lead to increased paranoia and delusions.

Hundreds of papers have been published in the last few years looking at the very popular issue of whether cannabis causes psychosis. This issue has received the international attention of media, governments, researchers, and the general public, all looking for answers. Yet we are still sorting through cluttered fields of data. The main anchor point in the noise is that there is a near consensus that cannabis use and psychotic disorders are correlated, but the nature of the relationship is still far from clear.

You can read more about this issue at Cannabis & Psychosis : Additional Notes and References.