The health and psychological consequences of cannabis use - chapter 7

THIS DOCUMENT HAS BEEN RESCINDED: Chapter 7.6 Does cannabis use cause psychotic disorders?

Page last updated: 1994

Please note the information in this publication may no longer be current but is retained on our website for historical or research purposes.

<< Previous Chapter | Contents | Next Chapter>>

7. The psychological effects of chronic cannabis use

7.6 Does cannabis use cause psychotic disorders?

There is a prima facie case for believing that cannabis use may in certain circumstances be a contributory cause of major psychological disorders such as psychotic disorders, i.e. illnesses in which symptoms of hallucinations, delusions and impaired reality testing are predominant features. First, THC is a psychoactive substance which produces some of the symptoms found in psychotic disorders, namely, euphoria, distorted time perception, cognitive and memory impairments (Brill and Nahas, 1984; Halikas et al, 1971; Thornicroft, 1990). Second, under controlled laboratory conditions with normal volunteers, THC has been shown at high doses to produce psychotic symptoms which include visual and auditory hallucinations, delusional ideas, thought disorder, and symptoms of hypomania (Georgotas and Zeidenberg, 1979; National Academy of Science, 1982). Third, a putative "cannabis psychosis" has been identified by clinical observers in regions of the world with a long history of chronic, heavy cannabis use, e.g. India, Egypt, and the Carribean (Brill and Nahas, 1984; Ghodse, 1986).

Top of page

7.6.1 The nature of the relationship

How might cannabis use causally contribute to the development of psychosis? The following are the major mechanisms that have been suggested by proponents of a relationship between cannabis use and severe psychological disorder (Thornicroft, 1990). Is there a 'cannabis psychosis'?

The first possibility is that acute or chronic cannabis use may produce a "cannabis psychosis". Four possible variants of this hypothesis can be distinguished. The first hypothesis is that the acute use of large doses of cannabis may induce a "toxic" or organic psychosis with prominent symptoms of confusion and hallucination, which remit with abstinence from cannabis. The second hypothesis is that cannabis use may produce an acute functional psychosis, similar in its clinical presentation to paranoid schizophrenia, and lacking the organic features of a toxic psychosis which remits after abstinence from cannabis. A third hypothesis is that chronic cannabis use may produce a chronic psychosis, i.e. a psychotic disorder which persists beyond the period of intoxication. The fourth hypothesis (a variant of the third) is that chronic cannabis use may induce an organic psychosis which only partially remits with abstinence, leaving in its train a residual deficit state with symptoms that are analogous to the negative symptoms of schizophrenia, or a mild chronic brain syndrome. This has also been described as "an amotivational syndrome" which is characterised by withdrawal, lack of interest in others, impaired performance and lack of motivation to perform one's social responsibilities. Does cannabis use precipitate a latent psychosis?

Cannabis use could conceivably precipitate a latent psychosis, i.e. bring forward an episode of schizophrenia or manic depressive psychosis in a vulnerable or predisposed individual. This could occur either as a result of a specific pharmacological effect of THC (or other constituents of cannabis preparations), or as the result of stressful experiences while intoxicated, such as a panic attack or a paranoid reaction to the acute effects of cannabis (Edwards, 1976). Schizophrenia is the disorder about which concern has been most often expressed in the case of cannabis use.

A related hypothesis would be that cannabis use exacerbates the symptoms of a functional psychosis such as schizophrenia or manic depressive psychosis. This could occur if cannabis use precipitated a relapse in the same way that it has been hypothesised to precipitate the onset of a latent psychosis. Alternatively, the pharmacological effects of cannabis might impair the effectiveness of the neuroleptic drugs used to treat major psychoses.

Top of page

7.6.2 Methodological issues

Until recently, our ability to test these hypotheses has been hampered by a lack of sophistication in research design (Mueser et al, 1990; Thornicroft, 1990; Turner and Tsuang, 1990). First, the possible mechanisms for a causal relationship between cannabis use and psychosis have not always been clearly distinguished, and so have not often informed the design of research studies purporting to test them. Second, studies of the relationships between cannabis use and psychological disorder have often been uncontrolled. Only rarely have they compared cannabis use in psychotic patients and controls, or compared the clinical characteristics and course of psychotic patients who have and have not used cannabis. Third, the extent of cannabis and other drug use, and its relationship to the onset of psychotic symptoms, has often been poorly documented. There has been a heavy reliance upon self-reported use, and few attempts have been made to distinguish between use, abuse and dependence (Mueser et al, 1990). Fourth, the diagnosis of a psychotic disorder, or of psychotic symptoms, has only rarely used standardised diagnostic criteria such DSM-III-R or ICD-9. Fifth, many studies have used small samples, reducing the chances of detecting any association between cannabis use and psychotic disorder. As a consequence of these deficiencies, many studies have failed to provide convincing evidence of even an association between cannabis use and psychotic symptoms or psychotic syndromes.

Even when an association between cannabis use and psychosis has been demonstrated, it has proved difficult to distinguish between alternative explanations of it. There has been a readiness to assume that the data supports the hypothesis that cannabis use is a contributory cause of psychosis (whether that is a specific "cannabis psychosis" or a functional psychosis such as schizophrenia). Only recently have other hypotheses been acknowledged, and attempts made to test them (e.g. Dixon et al, 1990, 1991; Turner and Tsuang, 1990).

There are a number of ways in which cannabis use could be associated with psychotic disorders without being a contributory cause of such disorders. One possibility is that the psychosis is a contributory cause of cannabis use, and that cannabis is used to self-medicate depression, anxiety, negative psychotic symptoms, or the side effects of neuroleptic drugs. Another possibility is that drug use among schizophrenic individuals is a consequence of pre-existing personality characteristics which predispose them to use illicit drugs and to develop schizophrenia. A third possibility is that heavy cannabis use may be a marker of the use of amphetamine and cocaine for which there is strong evidence for causing acute paranoid psychoses (Angrist, 1983; Bell, 1973; Connell, 1959).

In the review that follows, the best available clinical and epidemiological studies bearing on these issues is reviewed. Although we have preferred to cite controlled studies, we have not excluded all the early uncontrolled studies which have been most often cited. Attempts will also be made to distinguish the very different non-causal explanations of the apparent association between cannabis use and psychosis.

Top of page

7.6.3 'Cannabis psychoses' Toxic psychosis

Much of the literature on cannabis psychoses consists of case studies (e.g. Carney, Bacelle and Robinson, 1984; Drummond, 1986; Edwards, 1983; Weil, 1970), case series (e.g. Bernardson and Gunne, 1972; Cohen and Johnson, 1988; Kolansky and Moore, 1971; Onyango, 1986) and reviews of such reports (e.g. Tunving, 1985) which often suffer from a circularity in their argument (Thornicroft, 1990). typically a group of patients have been identified as having a toxic "cannabis psychosis" (with little information given on how they came to be so identified) and their behaviour and clinical history are then presented as evidence for the existence of the diagnostic entity they were meant to be testing. The better examples of these reports have attempted to justify their inclusion of cases within this diagnosis, and have attempted to assess the contribution of predisposition and drug use to the development of the psychosis.

Chopra and Smith (1974) have presented one of the largest case series of a toxic "cannabis psychosis". They described the characteristics of 200 East Indian patients who were admitted to a psychiatric hospital in Calcutta between 1963 and 1968 with "psychotic symptoms following the use of cannabis preparations" (p24). Their cases resembled cases of acute organic brain disorder in that their major symptoms included confusion and amnesia. The most common symptoms "were sudden onset of confusion, generally associated with delusions, hallucinations (usually visual) and emotional lability ... amnesia, disorientation, depersonalisation and paranoid symptoms" (p24). Most psychoses were preceded by the ingestion of a large dose of cannabis which produced intoxication and amnesia for the period between ingestion and hospitalisation.

Patients were classified into three groups on the basis of their history of previous psychiatric disorder. The first consisted of a third of patients who had no previous personality problems or psychiatric disorder, whose only constant feature was "recent use of cannabis preparations as the apparent precipitant of the psychotic episode" (p25). They exhibited symptoms of excitement, confusion, disorientation, delusions, visual hallucinations, depersonalisation, emotional instability and delirium. These symptoms were usually of short duration, varying between a few hours and several days, and all these patients returned to their normal state after remission.

The second group consisted of 61 per cent of patients who did not have a prior history of psychosis but had a history of schizoid, sociopathic, and unstable personalities. Their clinical picture was much like that of the first group, and they also had a high probability of remission within a few days of admission. The third group consisted of 10 patients with a prior history of psychosis (most often schizophrenia) who rarely experienced a short remission and usually required continued hospitalisation and treatment.

Top of page

Chopra and Smith argued that their case series provided evidence for the existence of the clinical entity of "cannabis psychosis". Although they conceded that excessive drug use could be a sign of pre-existing psychopathology, they argued that this was an unlikely explanation of their findings, because at least a third of their cases had no prior psychiatric history, the symptoms reported were remarkably uniform regardless of prior psychiatric history, and there was evidence of a dose-time relationship in that those who used the most potent cannabis preparations experienced psychotic reactions after the shortest period of use.

The findings of Chopra and Smith have received some support from case series published in other countries (e.g. Bernardson and Gunne, 1972; Onyango, 1986; Tennant and Groesbeck, 1972). Bernardson and Gunne (1972) reported on 46 cases of putative cannabis psychosis admitted to Swedish psychiatric hospitals between 1966 and 1970. These were primary cannabis users who had no history of psychosis prior to their cannabis use, and who presented with a clinical picture of paranoid delusions, motor restlessness, auditory and visual hallucinations, hypomania, aggression, anxiety and clouded consciousness. Their symptoms usually remitted within five weeks of admission, and those who returned to cannabis use after discharge were most likely to relapse.

Tennant and Groesbeck (1972) report on psychoses they had treated among US servicemen stationed in Germany between 1968 and 1971. During this period, potent hashish was cheap and readily available and widely used, with 46 per cent of servicemen reporting that they had used hashish, and 16 per cent reporting using it three or more times per week. They reported 18 cases of a short-term panic reaction or toxic psychosis developing after a single high dose of hashish, and 85 cases of toxic psychoses developing after the simultaneous consumption of cannabis and other drugs. The toxic psychoses usually resolved within three days on neuroleptic medication.

Onyango (1986) reported one of the few case series which used biochemical measures of recent cannabis use to identify possible cases of toxic cannabis psychosis among young adults who presented to a London psychiatric hospital with psychotic symptoms. He screened the urines of 25 such admissions and found that, although half reported having used cannabis at some time, only four had cannabinoid metabolites in their urines at the time of presentation. In three cases the patients had a prior history of psychosis, their phenomenology was unremarkable, and they did not respond rapidly to treatment. Only one case seemed to fit the picture of a cannabis psychosis. He had no prior history of psychosis, and a history of chronic, heavy cannabis use prior to admission. He presented with hallucinations, delusions, and labile, elated mood which responded rapidly to haloperidol, and he had no further episodes during a two-year follow-up.

All considered, there is a reasonable case for believing that large doses of potent cannabis products can produce a toxic psychotic illness in persons who do not have a personal history of psychotic illness (Edwards, 1976; Negrete, 1983; Thomas, 1993). Such psychoses are characterised by symptoms of confusion and amnesia, paranoid delusions, and auditory and visual hallucinations, and they have a relatively benign course in that they typically remit within a week of abstinence (Chaudry et al, 1991; Thomas, 1993). They seem most likely to occur in populations which use high doses of THC, and probably occur rarely otherwise (Smith, 1968). Given the poor standards of research design and lack of adequate controls in all but a few of these studies, and the failure to use standardised diagnostic criteria, it would be premature to claim that the existence of a toxic "cannabis psychosis" has been established beyond reasonable doubt.

Top of page An acute functional psychosis

Other investigators have argued that heavy cannabis use may produce an acute functional psychosis. That is, it produces an illness which does not reflect an organic state produced by drug intoxication, but rather a psychotic illness that resembles schizophrenia. Thacore and Shukla (1976), for example, reported a case control study comparing cases with a putatively functional cannabis psychosis with controls diagnosed as having paranoid schizophrenia. Their 25 cases of cannabis psychosis had a paranoid psychosis resembling schizophrenia, in which "a clear temporal relationship between the prolonged use of cannabis [longer than five years in all but one case] and the development of psychosis has been observed on more than two occasions" (p384). Their 25 age and sex-matched controls were individuals with paranoid schizophrenia who had no history of cannabis use.

The comparison revealed that the patients with a cannabis psychosis displayed more odd and bizarre behaviour, violence, panic affect, and insight and less evidence of thought disorder. They also responded swiftly to neuroleptic drugs and recovered completely. According to Thacore and Shukla, this functional psychotic disorder could be distinguished from the toxic "cannabis psychosis" reported by Chopra and Smith (1974), because there was no evidence of confusion and amnesia, and the major presenting symptoms were delusions of persecution, and auditory and visual hallucinations occurring in a state of clear consciousness.

Rottanburg et al (1982) provide one of the most convincing research studies in favour of the hypothesis that cannabis can produce an acute functional psychosis. They conducted a case-control study in which psychotic patients with cannabinoids in their urines were compared with psychotic patients who did not have cannabinoids in their urines. Both groups were assessed shortly after admission, and seven days later, by psychiatrists who used a standardised psychiatric interview schedule (PSE) and who were blind as to presence or absence of cannabinoids in the patients' urine.

Every third admission of a Cape coloured man during a period of a year (n=117) were screened for cannabinoids, alcohol and other toxins. Sixty per cent (N=70) had urines that were positive for cannabinoids, and 36 cases had levels which suggested heavy cannabis use prior to admission. Sixteen patients left hospital before the study was completed, leaving a group of 20 cases with psychoses and cannabinoids only in their urines. They were compared with 20 psychotic controls, matched for age and clinical diagnosis, whose urines were negative for cannabinoids and other drugs and toxins.

The results showed that psychotic patients with cannabinoids in their urine had more symptoms of hypomania and agitation, and less auditory hallucinations, flattening of affect, incoherent speech and hysteria than controls. They also showed strong improvements in symptoms by the end of a week, as against no change in the controls despite receiving comparable amounts of anti-psychotic drugs. They concluded that "heavy cannabis intake is associated with a rapidly resolving psychotic illness characterised by marked hypomanic features" (p1366).

Top of page

Imade and Ebie (1991) conducted a retrospective comparison of the symptoms reported by 70 patients with putatively cannabis-induced psychosis, 163 patients with schizophrenia, and 39 patients with mania. No details were provided on how these diagnoses were made, and the ratings of symptoms were made retrospectively from case records by psychiatrists who were not blind as to the patients' diagnoses. A large number of statistical comparisons produced a number of statistically significant differences in individual symptoms between the three patient groups, although they did not differ in symptoms of violence, panic and bizarre behaviour. Imade and Ebie argued that there were no symptoms that were unique to cannabis psychosis, and that there was no consistency of clinical picture that enabled them to distinguish a "cannabis psychosis" from schizophrenia. This negative study is unconvincing. The symptom ratings were made retrospectively from clinical records of unknown quality, and the patients' diagnoses were not made using standardised diagnostic criteria. There was no information on how "cannabis psychosis" was diagnosed, or on the clinical course of the psychoses. The authors also failed to use appropriate statistical methods to test the claim that cannabis psychosis can be distinguished from schizophrenia.

A number of cohort studies have been conducted on the prevalence of psychotic symptoms in chronic cannabis users and controls. Beaubruhn and Knight (1973) conducted a small study comparing the psychiatric history and symptoms of 30 chronic daily Jamaican cannabis users (with a history of at least seven years use) with that of 30 non-cannabis using controls matched on social class, income, age and sex. Both cases and controls were assessed by personal psychiatric interview and personality questionnaires during a six day hospitalisation. There were few statistically significant differences between the two groups, only a higher rate of family history of psychiatric disorder and of hallucinatory experiences in the cannabis users. Only one user and one non-user reported a personal history of psychiatric disorder.

Similar results have been reported by Stefanis et al (1976) in a study of 47 chronic cannabis users in Greece and 40 controls matched for age, family origin, residence at birth and upbringing. They found a higher incidence of personality disorders among their cannabis users, but no statistically significant difference in the rates of psychiatric disorder diagnosed by a personal interview with a psychiatrist. Three cases of schizophrenia were diagnosed in the cannabis using group, but a connection with cannabis use was discounted on the ground that two of the three had a family history of schizophrenia.

The small number of cases and the relative rarity of psychosis makes these studies unconvincing. The authors interpreted their results far too strongly, by inferring that a failure to find a difference in rates of psychiatric disorder in sample sizes of 30 and 47 indicated that there was no difference in prevalence between chronic cannabis users and controls. In Beaubruhn and Knight's study (1973), for example, the failure to detect a difference in the rate of psychosis between 30 cannabis users and 30 controls does not rule out a 17 fold higher rate of psychiatric disorder among cannabis users (as shown by the upper limit of a 95 per cent confidence interval around the odds ratio).

All considered, the case for believing that cannabis use can produce a functional paranoid illness is much less compelling than that for a toxic psychosis (Thomas, 1993; Thornicroft, 1990). The research designs for studies of this diagnosis have more often included control groups, but proponents of this hypothesis have not presented evidence that satisfactorily distinguishes it from other functional psychoses (Thornicroft, 1990).

Top of page

If there is a toxic cannabis psychosis, then a strong case has not been made for distinguishing it from the putatively functional cannabis psychosis. Thacore and Shukla (1976) emphasised the history of chronic heavy cannabis use among their cases of functional cannabis psychoses, and the absence of the confusion and amnesia reported in persons with the toxic psychosis.

The differentiation in terms of chronicity of drug use is unconvincing. Some of the cases of the toxic cannabis psychosis described by Chopra and Smith (1974), for example, had a long history of heavy cannabis use. The hypothesised difference in symptoms is more difficult to evaluate. Because few of the studies used standardised assessments of symptoms, the absence of reports of confusion and amnesia in the functional cases may indicate differences in diagnostic practice. There are also strong similarities between the putatively toxic and functional psychoses, namely, the occurrence of delusions, and auditory and visual hallucinations, and a relatively benign course, typically remitting within a week.

There is some recent support for the distinction between toxic and functional cannabis-induced psychoses. Tsuang et al, (1982) compared the demographic and clinical characteristics, and family histories of four groups of patients: those with drug abuse who had experienced a psychotic illness (DAP), those with diagnoses of drug abuse alone (DA), those with schizophrenia (SC), and those with diagnoses of atypical schizophrenia (AS). They subdivided the patients with drug abuse and psychosis into those with shorter and longer duration of symptoms. They found that the DAP patients were more likely to have abused hallucinogens and cannabis, and less likely to have abused sedative-hypnotics and opiates, than DA patients. The DAP patients also had an earlier onset of illness, and better premorbid personalities than the SC patients.

Comparisons of the DAP patients with short and long duration of illness produced some interesting results. The patients with short duration disorders had better premorbid personalties, fewer psychotic symptoms, and fewer core schizophrenic symptoms, such as poor insight, shallow and inappropriate affect, thought disorder, delusions and Schneiderian "first rank symptoms". They were more likely to have presented with "organic" symptoms such as confusion, disorientation, visual hallucinations, and amnesia than the patients with long duration disorders. By definition, the shorter duration patients had shorter periods of admission; they also had shorter duration of drug treatment, and more were discharged without being referred for further treatment. The prevalence of family histories of schizophrenia among the longer duration DAP patients was similar to that of the SC, while the shorter duration DAP patients had no such family history.

On the basis of their comparisons, Tsuang et al argued that the short duration disorders were drug-induced toxic psychoses, while the longer duration disorders reflected functional psychoses precipitated by drug use in predisposed individuals. If these findings are accepted, the simplest explanation of the allegedly functional "cannabis psychoses" is that they are functional psychoses occurring in heavy cannabis users.

Top of page Chronic psychoses

If cannabis can produce an acute organic psychosis, the possibility must be considered that chronic cannabis use may produce a chronic psychosis in much the same way as chronic alcohol heavy use can produce a chronic organic brain syndrome. As Ghodse (1986) has suggested, it is "theoretically possible in a situation of easy availability of cannabis, that regular, heavy users may suffer repeated, short episodes of psychosis and effectively `maintain' themselves in a chronic, psychotic state" (p477).

Although this is a possibility, there is no good evidence that chronic cannabis use causes a psychotic illness which persists after abstinence from cannabis (Thomas, 1993). This possibility is difficult to study because of the near impossibility of distinguishing a chronic cannabis psychosis from a functional psychosis such as schizophrenia in which there is concurrent cannabis use (Negrete, 1983). Certainly the findings of Tsuang et al (1982) suggest that the strong presumption must be that individuals with a history of drug abuse and a psychotic illness have a functional psychosis which has been precipitated or exacerbated by drug use. Follow-up studies of patients with acute cannabis psychoses, if they could be reliably identified, would be the best way of throwing some light on this issue. A residual state

A number of investigators have described a state among chronic, heavy cannabis users in which the users' focus of interest narrows, they become apathetic, withdrawn, lethargic, and unmotivated, and they have impaired memory, concentration and judgment (Brill and Nahas, 1984; McGlothin and West, 1968). This has been described as an "amotivational state", which some have attributed to an organic syndrome caused by the effects of chronic cannabis intoxication, from which there is incomplete recovery after prolonged abstinence (Tennant and Groesbeck, 1972).

The major clinical evidence in favour of such a hypothesis consists of case series among contemporary chronic cannabis users (e.g. Kolansky and Moore, 1971; Millman and Sbriglio, 1986), and historical reports of the syndrome among chronic, heavy users in countries such as Egypt, Greece, and the Carribean, where there has been a tradition of chronic heavy cannabis use among the lower socioeconomic groups (Brill and Nahas, 1984). These reports are often poorly documented and uncontrolled, and do not permit the effects of chronic drug use to be easily disentangled from those of poverty and low socioeconomic status, or pre-existing personality disorders (Edwards, 1976; Millman and Sbriglio, 1986; Negrete, 1983).

A small number of controlled studies of heavy chronic users in other cultures have largely failed to substantiate the clinical observations (Millman and Sbriglio, 1986), although there are enough reports of regular users complaining of loss of ambition and impaired school and occupational performance (e.g. Hendin et al, 1987), and of ex-users giving this as a reason for stopping (Jones, 1984), to keep the possibility alive. The small number of laboratory studies of long-term heavy use have produced mixed evidence (Edwards, 1976). Georgotas and Zeidenberg (1979), for example, reported that five healthy male marijuana users on a dose regimen of 210mg of THC per day for a month appeared "moderately depressed, apathetic, at times dull and alienated from their environment and with impaired concentration" (p430). Others have failed to observe such effects (e.g. Mendelson et al, 1974). The status of the amotivational syndrome consequently remains uncertain (see pp102-105).

Top of page

7.6.4 Cannabis and schizophrenia Precipitation

The possibility that heavy, chronic cannabis use may precipitate schizophrenia was raised by Tennant and Groesbeck (1972) in their study of the consequences of chronic heavy hashish use among American servicemen in Germany between 1968 and 1971. They reported 112 cases of "persistent schizophrenic reactions following prolonged hashish use" (p134), and they presented evidence that there had been a four fold increase in the incidence of schizophrenia among American servicemen during the period in which hashish use became endemic. As with all ecological evidence, a causal relationship is only one of the possible explanations of the apparently concurrent increase in the prevalence of hashish use and schizophrenia among American servicemen in Germany. The attribution of the increase to hashish use alone was also complicated by fact that many of their cases of schizophrenia also used hallucinogens, amphetamines, and alcohol.

The precipitation hypothesis has received some support from a series of case-control studies of cannabis and other psychoactive drug use among schizophrenic patients (Schneier and Siris, 1987). The usual finding has been that schizophrenic patients have higher rates of use of psychomimetic drugs such as amphetamines, cocaine, and hallucinogens than other patients (Dixon et al, 1990; Schneier and Siris, 1987; Weller et al, 1988) or normal controls (Breakey et al, 1974; Rolfe et al, 1993). The results for cannabis use have been more mixed, with some finding a higher prevalence of use or abuse (e.g. Mathers et al, 1991) and others not having done so (Dixon et al, 1990; Mueser et al, 1990; Schneier and Siris, 1987).

There is also good epidemiological evidence for an association between schizophrenia and drug abuse and dependence in the Epidemiological Catchment Area (ECA) study. In this study (Anthony and Helzer, 1991) there was an increased risk of schizophrenia among men and women with a diagnosis of any form of drug abuse and dependence: the excess risk of schizophrenia was 6.2 for men and 6.4 for women. Although separate estimates were not provided for cannabis abuse and dependence, it seems reasonable to assume that the same sort of relationship applied. Bland, Norman and Orn, (1987) have obtained a similar finding in a population survey of the prevalence of psychiatric disorder in Edmonton Alberta, using the same ECA interview schedule and diagnostic criteria. They found that the odds of receiving a diagnosis of drug abuse and dependence were 11.9 times higher among persons with schizophrenia.

Many researchers have favoured a causal interpretation of the increased prevalence of psychoactive drug use among schizophrenics, that is, they have concluded that cannabis and other drug use precipitates schizophrenic disorders in persons who may not otherwise have experienced them. In support of this hypothesis are the common findings that drug abusing schizophrenic patients have an earlier age of onset of psychotic symptoms (with their drug use typically preceding the onset of symptoms), a better premorbid adjustment, fewer negative symptoms (e.g. withdrawal, anhedonia, lethargy), and a better response to treatment and outcome than schizophrenic patients who do not use drugs (Allebeck et al, 1993; Dixon et al, 1990; Schneier and Siris, 1987).

There are other interpretations of these findings, however. Arndt et al (1992), for example, have suggested that the association between cannabis use and an early onset of schizophrenia in persons with a good premorbid personality and outcome is spurious. According to Arndt et al, schizophrenics with a better premorbid personality were simply more likely to be exposed to illicit drug use among peers than those with a withdrawn and socially inept premorbid personality, and because of this prior exposure to drugs, they were also more likely to use drugs to cope with the symptoms of an emerging psychosis. On this account, cannabis and other illicit drug use is a correlate of a good prognosis in schizophrenia, and pathological drug use is a response to the unrelated emergence of psychotic symptoms.

Top of page

A further possibility is that cannabis and other illicit drug use is a consequence of schizophrenia. That is, such illicit drug use is a form of self-medication to deal with some of the unpleasant symptoms of schizophrenia, such as depression, anxiety, lethargy, and anhedonia, and the side effects of the neuroleptic drugs used to treat it (Dixon et al, 1990). There is some support for this hypothesis in the work of Dixon et al (1990), who surveyed 83 patients with schizophrenia or schizophreniform psychoses about the effects of various illicit drugs on their mood and symptoms. Their patients reported that cannabis reduced anxiety and depression, and increased a sense of calm, at the cost of some increase in suspiciousness, and with mixed effects on hallucinations and energy.

Prospective evidence. The most convincing evidence of an association between cannabis use and the precipitation of schizophrenia has been provided by a prospective study of cannabis use and schizophrenia in Swedish conscripts undertaken by Andreasson et al (1987). These investigators used data from a 15-year prospective study of 50,465 Swedish conscripts to investigate the relationship between self-reported cannabis use at age 18 and the risk of receiving a diagnosis of schizophrenia in the subsequent 15 years, as indicated by inclusion in the Swedish psychiatric case register. Substantial data were collected on the conscripts (such as family circumstances, personal psychiatric disorder and other drug use) and statistical methods were used to examine the effect of these potentially confounding variables on the association between cannabis and schizophrenia.

Their results showed that the relative risk of receiving a diagnosis of schizophrenia was 2.4 times higher [95 per cent confidence interval 1.8, 3.3] for those who had ever tried cannabis compared to those who had not. There was also a dose-response relationship between the risk of a diagnosis of schizophrenia and the number of times that the conscript had tried cannabis by age 18. The crude relative risk of developing schizophrenia was 1.3 times higher [95 per cent confidence interval 0.8, 2.3] for those who had used cannabis one to ten times, 3.0 times higher [95 per cent confidence interval 1.6, 5.5] for those who had used cannabis between one and 50 times, and 6.0 times higher [95 per cent confidence interval 4.0, 8.9] for those who had used cannabis more than fifty times (compared in each case to those who had not used cannabis).

The size of the risk was substantially reduced by statistical adjustment for variables that were independently related to the risk of developing schizophrenia (namely, having a psychiatric diagnosis at conscription, and having parents who had divorced). Nevertheless, the relationship between cannabis use and schizophrenia remained statistically significant and still showed a dose response relationship. The risk of a diagnosis of schizophrenia for those who had smoked cannabis from one to ten times was 1.5 times that of those who had never used, and the relative risk for those who had used 10 or more times was 2.3 times that for those who had never used [95 per cent confidence interval 1.0, 5.3].

Top of page

Andreasson et al (1987) carefully scrutinised the validity of their data on cannabis use and the diagnosis of schizophrenia. They acknowledged that cannabis use was likely to have been under-reported because the information was not confidential, but they argued this was most likely to have under-estimated the relative risk of developing schizophrenia among users and non-users. Self-reported cannabis use at age 18 showed a strong dose-response relationship to the risk of receiving a diagnosis of drug abuse in the subsequent 15 years. Data from a small validity study indicated that 80 per cent of those diagnosed as schizophrenic in the case register met the DSM-III criteria for schizophrenia (which include a minimum duration of six months).

Andreasson et al (1987) and Allebeck (1991) argued for a causal interpretation of the association, conjecturing that cannabis use precipitated schizophrenia in vulnerable individuals. They rejected as implausible the hypothesis that cannabis consumption was a consequence of emerging schizophrenia. The cannabis users who developed schizophrenia had better premorbid personalities, a more abrupt onset, and more positive symptoms than the non-users who developed schizophrenia (Andreasson et al, 1989). Although over half of the heavy cannabis users (58 per cent) had a psychiatric diagnosis at the time of conscription, there was still a dose-response relationship between cannabis use and schizophrenia among those conscripts who did not have such a history. They stressed that cannabis use "only accounts for a minority of all cases" (p1485) since most of the 274 conscripts who developed schizophrenia had not used cannabis, and only 21 of them were heavy cannabis users.

No single study ever settles an issue. Even a prospective study as well designed, and as carefully interpreted as that of Andreasson et al has been criticised (Johnson, Smith and Taylor, 1988; Negrete, 1989). Among these criticisms are the following, which raise a number of alternative explanations to the causal one proposed by Andreasson and his colleagues.

First, there was a large temporal gap between self-reported cannabis use at age 18-20 and the development of schizophrenia over the next 15 years or so (Johnson, Smith and Taylor, 1988; Negrete, 1989). Because the diagnosis was based upon a case register, there was no information on whether the individuals continued to use cannabis up until the time that their schizophrenia was diagnosed. Andreasson et al (1987) anticipated and dealt with this criticism by showing that self-reported cannabis use at age 18 was strongly related to the risk of subsequently attracting a diagnosis of drug abuse. This suggests that cannabis use at age 18 was predictive of continued drug use, and the more so the more frequently it had been used by age 18.

A second possibility is that the excess rate of "schizophrenia" among the heavy cannabis users was due to acute cannabis-induced toxic psychoses which were mistakenly diagnosed as schizophrenia (Johnson et al, 1988; Negrete, 1989). Andreasson et al (1989) attempted to address this criticism by a study of the validity of the schizophrenia diagnoses in 21 conscripts in the case register (8 of whom had used cannabis and 13 of whom had not). This study indicated that 80 per cent of these cases met the DSM-III requirement that the symptoms had been present for at least six months, to exclude transient psychotic symptoms. This sample size (21 case) was small, however, and the confidence interval around a 20 per cent rate of misdiagnosis of schizophrenia is between 3 per cent and 37 per cent. Even if the rate of misdiagnosis was only 20 per cent, this could, if it varied between cannabis and non-cannabis users, be large enough to explain the relationship they observed.

Top of page

A third, more serious concern about the causal interpretation of the relationship between cannabis use and schizophrenia is that the relationship may be a consequence of the use of other illicit psychoactive drugs. Longitudinal studies of illicit drug use indicate that intensity of cannabis use in late adolescence predicts the later use of other illicit drugs. These drugs include amphetamine and cocaine (Johnson, 1988; Kandel and Faust, 1975) which can produce an acute paranoid psychosis (Angrist, 1983; Bell, 1973; Connell, 1959; Gawin and Ellinwood, 1988; Grinspoon and Hedblom, 1975). There is also good evidence that amphetamine was the major illicit drug of abuse in Sweden during the study period (Inghe, 1969; Goldberg, 1968 a, b), which suggests that intervening amphetamine use may have produced the correlation between cannabis use and schizophrenia. Andreasson et al's (1989) study reported that only two of their eight schizophrenic cannabis users had also been abusers of amphetamines prior to the diagnosis of their schizophrenia, but with a sample size as small as this, the true rate (indicated by a 95 per cent confidence interval) could be anywhere between 0 per cent and 55 per cent.

A fourth concern is that Andreasson et al (1987) have not ruled out the possibility that cannabis use at age 18 was a symptom of emerging schizophrenia. Statistical adjustment for a psychiatric diagnosis at conscription did not eliminate the relationship between cannabis use and schizophrenia, but it substantially reduced the size of the relative risk, because over half of the heavy users of cannabis had received a psychiatric diagnosis by age 18. Andreasson et al argued that this hypothesis was implausible because the dose response relationship between cannabis use and the risk of a schizophrenia diagnosis held up among those who did not have a psychiatric history. The persuasiveness of this argument depends upon how credible the screening for psychiatric diagnosis was at the time of conscription, and in particular, how confident we can be that a failure to identify a psychiatric disorder at conscription means that no disorder was present. This is difficult to evaluate.

The fifth and final criticism relates to the validity of self-reported cannabis use at conscription. Andreasson et al (1987) acknowledged that there was likely to be under-reporting of cannabis use because this information was not collected anonymously, but they argued that this was most likely to lead to an under-estimation of the relationship between cannabis use and the risk of schizophrenia. This will only be true, however, if the schizophrenic and non-schizophrenics conscripts were equally likely to under-report. If, however, pre-schizophrenic subjects were more candid about their drug use, the apparent relationship between cannabis use and schizophrenia would be due to response bias (Negrete, 1989). Although a possibility, this seems unlikely in view of the strong dose-response relationship with frequency of cannabis use, and the large size of the unadjusted relative risk of schizophrenia among heavy users.

When all these criticisms are considered, the Andreasson et al (1987) study still provides strong evidence of an association between cannabis use and schizophrenia which is not completely explained by prior psychiatric history. Uncertainty remains about the causal significance of the association because it is unclear to what extent the relationship is a result of drug-induced psychoses being mistaken for schizophrenia, and to what extent it is attributable to amphetamine rather than cannabis use.

Even if the relationship is causal, its public health significance needs to be kept in perspective. Although they did not report calculations of attributable risk, an estimate based upon the relative risk adjusted for psychiatric disorder (Feinstein, 1985) indicates that even if their association is causal, at most 7 per cent of cases of schizophrenia would be attributable to cannabis use. That is, on the prevalence rate of cannabis use reported by Andreasson et al, cannabis use would have explained 7 per cent (at most) of cases of schizophrenia occurring in Sweden during the period of study. Even this small potential contribution to an increased incidence of schizophrenia seems difficult to accept, since there is good independent evidence that the incidence of schizophrenia, and particularly of early onset, acute cases, declined during the 1970s, the period when the prevalence of cannabis use increased among young adults in Western Europe and North America (Der et al, 1990).

Top of page Exacerbation of schizophrenia

There is reason to be concerned about the effects of cannabis on psychotic symptoms among individuals with schizophrenia. Cannabis is psychoactive drug that is probably psychotomimetic in high doses, and its use seems to be relatively common among schizophrenic patients, as indicated above. There is also anecdotal clinical evidence that schizophrenic patients who use cannabis and other drugs experience exacerbations of symptoms (Weil, 1970), and have a worse clinical course, with more frequent psychotic episodes, than those who do not (Knudsen and Vilmar, 1984; Perkins et al, 1986; Turner and Tsuang, 1990).

However, there have been very few controlled studies of the relationship between cannabis use and the clinical outcome of schizophrenia. Negrete et al (1986) conducted a retrospective study using clinical records of symptoms and treatment seeking among 137 schizophrenic patients with a disorder of at least six months duration, and three visits to their psychiatric service during the previous six months. The proportion of cannabis users among their patients was the same as in the Canadian population, but heavy users were over-represented, and the proportion of former users who had stopped using was higher than in the general population. Negrete et al (1986) compared the prevalence of hallucinations, delusions and hospitalisations among the active users (N=25), the past users (n=51), and those who had never used cannabis (N=61). The crude comparison showed higher rates of continuous hallucinations and delusions, and of hospitalisations among active users. This pattern of results persisted after statistical control for differences in age and sex between the three user groups.

Negrete et al argued that cannabis use exacerbated schizophrenic symptoms. They rejected the alternative hypothesis that patients with a poorer prognosis were more likely to use cannabis, because they found that past cannabis users experienced fewer symptoms, and reported a high rate of adverse effects when using (91 per cent). They also discounted the possibility that these were toxic psychoses, because in all cases the minimum duration of symptoms had been six months. They left open the mechanism by which cannabis use exacerbated schizophrenic symptoms, suggesting three possibilities: that cannabis disorganises psychological functioning; that it causes a toxic psychosis that accentuates schizophrenic symptomatology; or that it interferes with the therapeutic action of anti-psychotic medication.

More recently, Cleghorn et al (1991) have provided supportive evidence. They compared the symptom profiles of schizophrenic patients with histories of substance abuse of varying severity (none, moderate, and severe), among whom cannabis was the most heavily used drug. Comparisons with a subset of the patients who were maintained on neuroleptic drugs revealed that the drug abusers had a higher prevalence of hallucinations, delusions and positive symptoms.

These studies provide a slender basis upon which to draw conclusions about the effects of cannabis use on schizophrenic symptoms. One can only agree with the conclusion of Turner and Tsuang (1990) that "the impact of substance abuse on the course and outcome of schizophrenia remains largely undefined" (p93), and that it will remain so until large prospective studies in general population and clinical samples recommended by Turner and Tsuang (1990) have been conducted. Until such research has been undertaken, prudence would demand that schizophrenic patients, and others at risk of schizophrenia by virtue of family history, personality, or marginal social functioning, should be strongly discouraged from using cannabis and other psychoactive drugs, especially the psychostimulants amphetamine and cocaine.

Top of page

7.6.5 Conclusions

There is reasonable evidence that heavy cannabis use, and perhaps acute use in susceptible individuals, can produce an acute psychosis in which confusion, amnesia, delusions, hallucinations, anxiety, agitation and hypomanic symptoms predominate. The evidence for a toxic cannabis psychosis comes from laboratory studies of the effects of THC on normal volunteers and clinical observations of psychotic symptoms in heavy cannabis users, which seem to comprise a toxic psychotic syndrome and which remit rapidly following abstinence from cannabis. There is also an argument by analogy with the fact that heavy chronic amphetamine use has been shown to induce a paranoid psychosis (Angrist, 1983).

There is little support for the hypothesis that cannabis use can cause a chronic psychosis which persists beyond the period of intoxication. Such a possibility is difficult to study because of the likely rarity of such psychoses, and the near impossibility of distinguishing them from individuals with schizophrenia and manic depressive psychoses who also abuse cannabis (Negrete, 1983).

The occurrence of a chronic residual state, or "amotivational syndrome", in chronic heavy cannabis users is not well supported by research evidence. At best, a prima facie case has been made by clinical observations, that withdrawal, lethargy, and apathy occur among a minority of chronic, heavy users. This syndrome has proved difficult to study in the laboratory, difficult to distinguish from the effects of chronic intoxication (Negrete, 1988), and it so far been impossible to rule out confounding effects of pre-existing disease, malnutrition, personality disorder, and lifestyle.

There is strongly suggestive evidence that chronic cannabis use may precipitate a latent psychosis in vulnerable individuals. This is still strongly suggestive rather than established beyond reasonable doubt, because in the best study conducted to date (Andreasson et al, 1987) the use of cannabis was not documented at the time of diagnosis, there was a possibility that cannabis use was confounded by amphetamine use, and there remains a question about the ability of the study to reliably distinguish between schizophrenia and acute cannabis or other drug-induced psychoses.

Even if the relationship between cannabis use and schizophrenia is a causal one, its public health significance should not be overstated. It is most likely to indicate that cannabis use can precipitate schizophrenia in vulnerable individuals, since the estimated attributable risk of cannabis use is small, and the incidence of schizophrenia has declined during the period in which cannabis use has increased among young adults.

The substantial prevalence of cannabis use among young adults in Western societies makes the relationships between cannabis use and psychosis deserving of further research. What are required are case-control studies of people with schizophrenia and normals, and case-control studies of psychotic individuals who do and do not have a documented history of recent heavy cannabis use. Mueser et al (1990) provide detailed suggestions for the types of controls that ought to be incorporated in such studies. If the results of the case control studies warrant it, prospective studies should be done. Longitudinal studies like that undertaken by Andreason et al (1987) would be most desirable, but can probably only be undertaken in exceptional circumstances. Turner and Tsuang (1990) provide detailed suggestions for prospective studies which would clarify the contribution of cannabis and other drug use to the precipitation and exacerbation of schizophrenia and other psychoses.