IV. COMPARING THE MAGNITUDE OF RISKS
The standard ways of measuring the magnitude of health risks are relative risk and population attributable risk. The relative risk of cannabis use, for example, is the increase in the odds of experiencing an adverse health outcome among those who use cannabis compared to those who do not. The population attributable risk represents the proportion of cases with an adverse outcome which is attributable to cannabis use. Relative risk is of most relevance to individuals attempting to estimate the increase in their risk of experiencing an adverse outcome if they use a drug. Attributable risk is of most relevance to a societal appraisal of the harms of drug use.
The personal and public health importance of the two measures of risk magnitude depends upon the prevalence of drug use and the base rate of the adverse outcome. An exposure with a low relative risk may have a low personal significance but a large public health impact if a large proportion of the population is exposed (e.g. cigarette smoking and heart disease). Conversely, an exposure with a high relative risk may have little public health importance because very few people are exposed to it but major personal health implications for those who are exposed. Consequently, an appraisal of the personal and public health importance of cannabis and other illicit drug use must take account not only of the relative risk of harm but also the prevalence of use and the base rate of the adverse effect.
The Relative Risks of Adverse Health Effects of Cannabis Use
Many of the quantitative risks of cannabis use can only be guessed at in the absence of studies of the dose-response relationship between cannabis use and the various adverse health effects. The following are guesstimates of the risks of cannabis use for the most probable adverse health effects. When in doubt we have adopted the strategy of assuming as a worst case that the relative risks of cannabis use are comparable to the relevant risks of alcohol or tobacco.
Motor Vehicle Accidents
If we assume that driving while intoxicated with cannabis produces a comparable increase in the risk of accidents to that produced by driving while intoxicated with alcohol (say with a blood alcohol level of 0.05% to 0.10%), then a RR in the range of 2 to 4 would be reasonable. The fact that alcohol and cannabis are often used in combination complicates the task of estimating the relative risk of cannabis use alone to motor vehicle accidents.
If we assume that a daily cannabis user who smokes 5 or more joints per day faces a comparable risk of respiratory disease to that of a 20 a day tobacco smoker, then the RR of developing chronic bronchitis would be 5 or greater for those who had ever smoked cannabis, and substantially higher among those who also used tobacco and those had been daily smokers over many years (Holman et al, 1988). The increased risk of respiratory disease is, of course, specific to use of cannabis by smoking.
Respiratory Tract Cancers
If we make the same worst case assumptions about daily cannabis smoking then the relative risks of various cancers of the respiratory tract would be of the order of: 4 for oral cancer, 6 for pharyngeal cancers, 4 for oesophageal cancer, and 7 for lung cancer (Holman et al, 1988). Again these risks could be substantial higher among cannabis smokers who also smoke tobacco, but would be minimal for non-smoking cannabis use.
Low Birthweight Babies
Making a worst case assumption in the absence of good data, a woman who smokes cannabis during pregnancy approximately doubles her chance of giving birth to a low birthweight baby (Holman et al, 1988).
This is one of the few health consequences for which there is quantitative estimate of relative risk. If we use the estimated RR from the study by Andreasson et al (1987) after adjustment for confounding variables, then an adolescent who had smoked cannabis 50 or more times by age 18 would have approximately a 2 to 3 times higher risk of developing schizophrenia than an adolescent who had not been a cannabis smoker.
Since cannabis use is a necessary condition of developing dependence, it is not appropriate to estimate a relative risk. The best way of quantifying the risk of dependence is to estimate the proportion of those who have ever used cannabis, or who have had a history of daily use, who become dependent on the drug. The best estimates of these percentages are based primarily upon US data from the late 1970s and early 1980s. These are that 10% to 20% of those who have ever used, and 33% to 50% of those who have had a history of daily use, will become dependent on cannabis (see Hall et al, 1994). Comparable percentages for tobacco and opiates would be higher than these.
From the perspective of the individual cannabis user, the major health risks of cannabis use are, with one exception, most likely to be experienced by those who smoke the drug daily over a period of years. These are in order of decreasing risk: developing a cannabis dependence syndrome, developing chronic bronchitis, and being involved in a motor vehicle accident if driving while intoxicated. In all these cases, the risk will be increased if cannabis is combined with either alcohol or tobacco or both. The risk most likely to be experienced by the occasional user is an increased risk of a motor vehicle accident if used when driving a car, especially if cannabis is combined with alcohol.
Public Health Significance
Motor Vehicle Accidents
An assessment of the public health significance of motor vehicle accidents caused by cannabis is made difficult by the strong association between cannabis and alcohol use. The epidemiological studies indicate that in its own right, cannabis makes at most a very small contribution to motor vehicle accidents, and so on the whole it may seem be a minor road safety problem by comparison with alcohol. Its major public health significance for road safety may be in amplifying the adverse effects of alcohol in those of drivers who combine alcohol and cannabis intoxication.
The public health significance of respiratory diseases caused by cannabis smoking is probably greater than that for respiratory cancers. This is so for two reasons. First, respiratory cancers require a greater length of exposure to cigarette smoke (15 to 20 years) than is required to develop chronic bronchitis. Second, there are very few cannabis users who use the drug for more than 5 years. The exposure period for chronic bronchitis may be shorter still among those cannabis smokers who also smoke tobacco since there is good evidence that concurrent tobacco and cannabis smoking have additive adverse effects on the respiratory system. The contribution of cannabis to respiratory diseases is more a matter of morbidity than mortality.
Respiratory Tract Cancers
Even if we make the worst case assumption that the risks of cancer are comparable among daily tobacco and cannabis smokers then cannabis smoking will make at most a small contribution to the occurrence of these cancers, at least on the basis of current patterns of use in developed societies. This is because only a minority of those who ever use cannabis become daily users, and a much smaller proportion of these daily users persist in smoking cannabis beyond their middle twenties by comparison with the proportions of tobacco smokers who do so. Among this minority concurrent cannabis and tobacco use may amplify the adverse respiratory effects.
Low Birthweight Babies
Again making a worst case assumption, cannabis smoking during pregnancy may double the risks of a woman giving birth to a low birthweight baby. The public health significance is likely to be much lower than that of tobacco smoking during pregnancy because the prevalence of cannabis use is likely to be much lower. Although foetal exposure to cannabis smoke may be relatively low, the risks of a low birthweight baby will be even higher among those women who also smoke tobacco, as do most of those smoking cannabis during pregnancy.
As argued in detail elsewhere (Hall et al, 1994), there is uncertainty about whether the association observed between cannabis use and schizophrenia is a token of a causal relationship. But even if it is, its public health significance should not be overstated. Schizophrenia affects approximately 1% of the adult population, and on the data of Andreasson et al, cannabis use would account for less than 10% of cases of schizophrenia. Even this low figure seems unlikely, however, since the incidence of schizophrenia has probably declined during the period when cannabis use among adolescents and young adults has increased (Der et al, 1991).
Cannabis dependence is potentially a larger public health problem than any of the other potentially adverse health effects of cannabis. On the ECA estimates, approximately 4% of the adult US population met diagnostic criteria for cannabis abuse or dependence, as against 14% who met diagnostic criteria for alcohol abuse and dependence. This is a nontrivial proportion of the population, although its consequences are somewhat ameliorated because there is probably a high rate of remission of symptoms in the absence of treatment.
Overall, most of these risks are small to moderate in size. In aggregate they are unlikely to produce public health problems comparable in scale to those currently produced by alcohol and tobacco. This is largely because on current patterns of use in developed societies the proportion of the population that uses cannabis heavily over a period of years is much smaller than the proportions that use alcohol or tobacco in a comparable way (Hall, 1995).
We should beware of becoming too complacent about this situation. The comparison based upon existing patterns of use cannot be used to predict what would happen if there was a major change in the prevalence of cannabis use, as some may argue would happen if existing criminal penalties were removed or replaced with civil penalties. But even if there were more users, it is unlikely that the proportion of cannabis users who become very heavy users would ever be as high in industrial societies as it often for stimulants such as tobacco or cocaine, since heavy use of a stimulant fits more easily into the rhythms of daily life in such societies.
In principle, it would seem a simple matter to estimate what the health risks of cannabis use would be if its prevalence was the same as that of alcohol and tobacco. Although conceptually a simple matter, a number of assumptions have to be made. The most questionable assumption is that the public health consequences of an increased prevalence of cannabis use would simply be the product of the current patterns of use multiplied by the ratio of the new to old users. Such a calculation assumes that the risks are the same regardless of the characteristics of the user, or the legal regime under which the drug is used.
The first assumption may be unreasonable. It may be, for example, that cannabis is used by a different population when its prevalence of use is low than when it was high. This phenomenon has been reported with alcohol, for example, with different patterns of alcohol consumption and problems in "dry" and "wet" cultures. If adult use were legalised, it might also be easier to reduce some of these health risks. For example, with greater availability it may be possible to reduce the major respiratory risks of cannabis smoking, either by encouraging cannabis users to ingest rather than to smoke the drug, or by increasing the THC content and reducing the tar content of marijuana, for those who continue to smoke. It would also be easier if cannabis use were legal to give users advice on other ways of reducing their risks of using the drug. Estimating the net effects of such harm reduction efforts is difficult, however, because it would also be likely that decriminalising cannabis for adult use would lead to an increased use by adolescents, and the health effects of this would be difficult to predict.
For these reasons we have not attempted to provide estimates of the health risks of cannabis if its prevalence of use were to approach those of alcohol and tobacco. All that can be said with any confidence is that if the prevalence of cannabis use increased to the levels of cigarette smoking and alcohol use, its public health impact would increase. It is impossible to say by how much with any precision. However, on even the most worst case scenario, it is unlikely that the public health effect of cannabis use would approach those of alcohol or tobacco use. Unlike alcohol, cannabis does not produce cirrhosis for example. Moreover, in developed societies cannabis appears to play little role in injuries caused by violence, as does alcohol, although recently concern has been expressed in some developing countries that cannabis may be used to fortify criminal offenders. Unlike tobacco, all the evidence suggests that the proportion of cannabis smokers who become daily smokers is substantially less than the proportion of tobacco smokers who do so.
Some Direct Comparative Evidence on Consequences:
What Users Report
What Users Report
To a limited extent, epidemiological data are available on the consequences of drug use that users attribute to their drug use. Since this data has not been collated and reviewed, we summarise some of it here. As we shall discuss, the data should be interpreted with caution. And it should be recognised that the range of consequences considered here reaches far beyond the bounds of the clinically significant physical and mental illnesses which are our focus elsewhere in this review.
In a large sample of U.S. men aged 20 - 30, interviewed in 1974 (Table 1), a higher proportion of tobacco smokers rated the effects of their use as bad, and more drinkers of alcohol gave a bad than a good rating. Good ratings outweighed bad for marijuana users. In a survey of Ontario adults in 1994 (Table 1), current users aged 18 to 34 gave a similarly negative weighting to tobacco, but gave alcohol a relatively more favourable weighting. Again, marijuana users were the most likely to give a favourable rating.
In another Ontario survey in 1992, current users were asked whether their use of a drug had a harmful effect on different aspects of their life in the past 12 months. Table 2 shows the results for users of alcohol, tobacco and marijuana, and also for heavier or more frequent users of each drug: drinkers who drank five or more drinks on an occasion at least once a month, marijuana users who smoked at least once a month, and tobacco smokers who smoked at least 11 cigarettes a day. Tobacco smokers were more likely than marijuana or alcohol users to report harm to their physical health, to their finances, and to their friendships or social life, among both lighter and heavier users. The small number of regular marijuana users seemed more likely than heavier drinkers to report harm to their home life or marriage and to their finances. On other comparisons, the proportions reporting harm for each drug were fairly similar.
In the 1991 U.S. National Household Survey on Drug Abuse, large samples of current tobacco, alcohol and marijuana users were asked comparable questions about 11 consequences of use. Marijuana users were a little more likely to report consequences (15.5% reporting any of the 11 consequences) than alcohol users (11.4%) or cigarette smokers (11.2%). If these rates are extrapolated to the whole population, including nonusers, then 1.9% of the population reported consequences of marijuana use, 7.2% reported consequences of drinking, and 3.4% reported consequences of cigarette smoking. Marijuana users reported noticeably higher rates on four items: "became depressed or lost interest in things", "found it difficult to think clearly", "got less work done than usual at school or on the job", and "felt suspicious and mistrustful of people". Responses to the last two items may be particularly influenced by marijuana's illegal status. Drinkers reported higher rates of "arguments and fights with family or friends" and "found it difficult to think clearly". Cigarette smokers reported higher rates of "felt very nervous and anxious" and "had health problems" (USDHHS, 1993, Table 9.2).
Great caution must be used in interpreting the results of such comparisons. In the first place, the base of users is different for each drug. Those using a widely-used drug are likely to differ on salient characteristics from those using a more rarely used drug. Second, the reported consequence may not be seen by the respondents themselves as adverse. For instance, if the purpose of use in intoxication, it may not be seen as a problems that the respondent "found it difficult to think clearly". Third, responses are likely to be influenced by cultural beliefs about causal connections. The high proportion of young adult smokers reported smoking has harmed their health may reflect acceptance of conventional wisdom as much s personal experience. Fourth and most important, the connection between drug use and adverse consequences will be influenced by a variety of factors applying differentially to different drugs. In particular, a drug's illegal status can itself create adverse consequences for the user, not only directly, through arrest, but also indirectly, for instance in the form of "harm to home life" from the adverse reactions by others to the drug use that involves a risk of arrest.
Keeping these caveats in mind, it is clear that a minority of marijuana users do report harm from their smoking, and some would be likely to do so even if cannabis were legalised. In an era where the health consequences of tobacco smoking are well recognised, tobacco smokers seem to be more likely than users of either cannabis or alcohol to regard their use as doing more harm than good in their lives, and the good is seen as outweighing the bad more often by cannabis smokers than by drinkers or tobacco smokers. In the present circumstances of North America, cannabis smokers are least as likely as alcohol drinkers to report adverse consequences of their use. But the higher rate for alcohol of "arguments and fights with family or friends" reminds us of the special potential alcohol consumption has to have harmful effects on others. Given current patterns of use, when rates of consequences are restated on the basis of the whole population, consequences of alcohol and tobacco use are clearly of greater public health significance than consequences of marijuana use.