Cannabis Smoking and Schizophrenia


It is well known among clinicians that cannabis smoking interacts with manifest schizophrenia: cannabis smokers experience more relapses (more frequent hospitalisation), their symptoms are more dramatic, and they are less susceptible to the effects of neuroleptic medication.

It is also considered a well-established fact that intensive cannabis smoking can be the precipitating (in the sense of activating) factor when particularly sensitive individuals fall ill for the first time. In addition to this, some studies suggest that the onset of cannabis-precipitated schizophrenia occurs, on average, at a younger age (four years earlier, according to some calculations) than spontaneous onset of schizophrenia.

Can cannabis cause schizophrenia, i.e. provoke schizophrenia in individuals who would not otherwise have developed the illness? There is a growing body of indications in support of the claim that the answer to this question is “yes”. The largest study in this field is the Swedish one of military conscripts. Until recently, no other study had repeated it in order to contribute to corroborating its conclusions in a convincing way. However, given van Os’s (2002) thorough study of 5,000 randomly chosen Dutch people as well as a pair of longitudinal studies (lasting for 21 and 29 years, respectively) from New Zealand, it must now be considered proven that cannabis smoking can provoke a functional (non-toxic) schizophrenia-like psychosis.

Can Cannabis Smoking Cause Schizophrenia?

The issue of how cannabis can cause relapses into schizophrenia and how it can interact negatively in different ways with schizophrenic illness is discussed in the following section. First, though, we will discuss the question of whether cannabis can literally cause schizophrenia, or in other words whether cannabis smoking, viewed from a stress-vulnerability angle, can exert such a heavy strain on the psyche that a schizophrenic illness – which would probably otherwise never have been provoked – breaks out.

Several of the findings from the above-mentioned studies () suggest that there may be such a connection. However, none of the studies meets the methodological requirements necessary for such a conclusion to be tenable. Already in the early 1970s, Breakey et al. () pointed out that there was some form of association between drug abuse (including cannabis abuse) and the onset of schizophrenic illness. He considered that cannabis (and other drugs) precipitated latent schizophrenia, but also believed that there were grounds for suspecting that cannabis could precipitate schizophrenia in cases where the illness would otherwise not have become manifest. The reasons were that the onset of drug-induced schizophrenia occurred on average four years earlier than the onset of other forms of schizophrenia, that the onset was more sudden than in other schizophrenics, and that the patients’ premorbid personality was consistently better than in a comparable group with non-drug-induced schizophrenia. Hereditary aspects were not touched upon in this study. Twenty years later, Eikmeier () arrived at similar results in a larger-scale study which looked only at cannabis-induced schizophrenic psychoses as compared with drug-free schizophrenic psychoses. These studies, it should be stressed, do not prove that cannabis use really does increase the risk of developing schizophrenia; but they do suggest that cannabis use will cause the onset of the illness to occur earlier (which is serious enough in itself).

It is not difficult to understand why there are relatively few studies claiming to help answering this question. On a general note, the complexity of the question requires the use of an advanced scientific method, and probably also a very large study group which is monitored over a long period of time. There is a need not only for interest and skill in the research team, but also for endurance and financial resources.

Until recently, there has really been only a single study which comes close to meeting these requirements: the large-scale follow-up of 45,000 Swedish military conscripts carried out by Andréasson, Allebeck, Engström and Rydberg (). Information recorded at the time of the enrolment of the conscripts (typically at the age of 18 or 19), including on their drug-taking habits, was compared with recorded schizophrenia diagnoses over the subsequent fifteen years. It was found that a person who had claimed, at the age of 18 or 19, to have consumed a great deal of cannabis (on more than 50 occasions) was six times more likely to be diagnosed as schizophrenic during the subsequent fifteen-year period than a person who had claimed at enrolment that he had never used cannabis at all. When account was also taken of other factors which were recorded at the time of enrolment and which may increase the statistical likelihood of being diagnosed with schizophrenia, the cannabis-dependent risk became smaller, but there remained a statistically significant risk increase associated with cannabis use.

On publication, this study justifiably aroused considerable attention; and for the past fifteen years, it has probably been the most frequently cited study in the international research literature on the relationship between cannabis and psychosis. The initial reactions to it ranged from total acceptance () to extensive criticism (). Negrete – the doyen of this field – judged the connection which was claimed to exist between cannabis and the precipitation of latent schizophrenia to be reasonable, on the basis of this study as well as previous ones (). At the same time, though, he was of the opinion that the study suffered from a number of weaknesses.

Andréasson et al. () and, above all, Allebeck et al. () have carried out a couple of supplementary studies in which they have tried, in large part successfully, to eliminate the weak points of the original study, thus adding further strength to their original findings. The latter study involved going through information from the medical records of 112 patients who had been diagnosed with both schizophrenia and cannabis dependence. This information was examined to determine such aspects as the clinical picture at the onset of illness, the temporal relationship between the cannabis abuse and the patient’s falling ill, and the involvement of other drugs, above all amphetamine. In all significant respects, the findings which were made in these studies confirmed the conclusions reached in the original study.

Further support as regards one aspect of that study comes from an examination of 100 randomly chosen medical records of patients diagnosed as having schizophrenia in the period 1973–1977. The researchers found a large degree of consistency across regions and hospitals as well as over time, and also a large degree of conformity with the DSM-IV diagnostic criteria for schizophrenia ().

In the original Swedish study, the first analysis was performed fifteen years after the enrolment of the conscripts. In order to add further strength to their findings and to extend the study, the research team (reinforced by a few British researchers) performed another analysis 27 years after the time of enrolment. By comparing the group of people who had fallen ill within five years of enrolment with those who had fallen ill later, and by re-analysing the data provided at the enrolment interviews – in the light of twelve more years’ worth of research into the importance of various background factors in the development of schizophrenia –, the researchers were able to dismiss a few more of the objections which had been made to the original study ().

Some Recent Studies

A few studies published in the past five years support the Swedish study to some extent. J. van Os et al. () claim to have replicated the Swedish study, which gives their study extra weight. This study doubtless occupies (along with the Swedish one) a particularly prominent place. There is no room here for a detailed description, but suffice it to say that even though the study group is smaller and the period of monitoring is shorter in the Dutch study – both considerably so –, most of the weaknesses hampering the Swedish study have been eliminated.

Within the framework of a national Dutch inventory study of psychiatric morbidity, a random sample of just over 7,000 people (reduced through dropouts and the application of exclusion criteria to 4,100) from 60 different localities was drawn. A baseline assessment and two follow-up assessments, the later one after three years, were made individually by means of questionnaires and, where necessary, in the form of clinical interviews. Those exhibiting psychotic manifestations at follow-up were divided into three groups, of which the “heaviest” one would appear to be rather close to the schizophrenia diagnosis in the Swedish study.

The study by J. van Os et al. shows the following:

a) There is a causal link between cannabis taking and psychotic manifestations.

b) The psychotic manifestations range from psychotic symptoms to mental illness (defined as “need for psychiatric treatment for psychotic illness”).

c) The risk of developing psychosis increases with the amount of cannabis taken.

d) The risk of developing psychosis seems to be greater where the individual has consumed cannabis earlier (before or at the start of the study) than where the individual has done so later (in direct connection with the follow-up after three years). This makes it possible to exclude, in large part, that the psychoses were caused by direct toxic effects.

e) In cannabis-smoking individuals assessed, at the beginning of the study, as having a disposition towards psychosis, there is a synergistic risk increase, with a very high proportion exhibiting manifest psychotic illness at the three-year follow-up.

The studies mentioned in the previous chapter on cannabis psychosis () also support the existence of a causal link between cannabis and psychotic illness.

Arseneault et al. also consider their study as directly supporting the Swedish study. However, unlike that study, which is retrospective in several respects, the findings made by Arseneault et al. in New Zealand are based on a long-term prospective study (data on just over 1,000 children/young people, with the times of assessment including the ages of 11, 15 and 18 as well as the end of the study, when the subjects were 26 years old). Cannabis smoking during the teenage years was found to have a clear impact in that the risk of schizophreniform illness (including an unknown proportion of schizophrenia proper) was four times greater at the age of 26. The younger a person was when first using cannabis, the greater the risk was found to be (which may be because the early start has resulted in the overall consumed amount of cannabis being larger).

Fergusson et al. () showed that cannabis dependence increases the frequency of psychotic symptoms. Degenhardt and Hall () show, in a large study of 6,700 adult Australians, that cannabis dependence doubles the risk of developing psychotic symptoms.

Cannabis and Manifest Schizophrenia

Naturally, the line of argument followed in the preceding section is also supported by the specific interaction between schizophrenia and cannabis smoking which has been observed by many psychiatrists and whose existence has been borne out by several scientific studies.

Treffert’s case descriptions from 1974 of four schizophrenic patients have acquired almost classic status. He let four schizophrenic patients, all of whom were on antipsychotic medication, act as their own controls. They all smoked cannabis occasionally, despite having been warned not to, and regularly experienced deterioration in their condition, sometimes with very serious consequences. It was not difficult for Treffert to demonstrate a direct association between relapses into marijuana smoking and serious deterioration of the schizophrenic condition ().

Most psychiatrists have, over the years, found themselves able to make similar observations: cannabis smoking causes a deterioration regardless of antipsychotic medication.

According to Negrete et al. (), the interaction between cannabis smoking and schizophrenia has the following characteristics: cannabis smokers experience more relapses (more frequent hospitalisation), their (“positive”) symptoms are more dramatic, and they are less susceptible to the effects of neuroleptic medication. The effects seem to be dose-dependent (). There are no significant discrepancies in the way different clinicians or different researchers view the problem. A further question, which however does not fall within the scope of my report, is why schizophrenics are so strongly attracted to cannabis (and amphetamine) despite the risks of deterioration involved.

In recent years, these different aspects of the interaction between cannabis and schizophrenia have also been demonstrated in a series of studies (). Among the authors of reviews, Degerhardt and Hall () may be mentioned.


Selections from the book: “Adverse Health Consequences of Cannabis Use. A Survey of Scientific Studies Published up to and including the Autumn of 2003