Cohen, Peter (1997), Monitoring cannabis use: A case study. Paper presented at the Invitational Conference on Monitoring Illicit Drugs and Health in the European Union, Amsterdam, 22 May 1997. An earlier version of this paper was published in: Trimbos Instituut (1997), Invitational conference on monitoring illicit drugs and health. Final report. Utrecht: Trimbos Instituut. pp. 83-94.
© Copyright 1997 Peter Cohen. All rights reserved.


Monitoring cannabis use

A case study

Peter Cohen


If we are to measure cannabis use, its developments and consequences, and attribute meaningful explanations to these data, quite a lot of careful and difficult research is needed.

In general, we should distinguish three main groups that need to be investgated.

  • First the general population. We can ask good and large enough samples of this population questions about cannabis use. These general household surveys are beginning to develop and may exhibit a slow increase of quality and uniformity in the EU member countries. They are used to reach a general understanding of drug use and its development in the general population.
  • The second group we should focus on to find out about cannabis use are the users themselves. Finding a good cross-section of users is difficult, so that very few studies this type are conducted. Most research, if any, is limited to captive groups or highly skewed selections, like those receiving treatment. However, since much of the theory about the consequences of cannabis use deals of course with cannabis users, we should direct more of our trials towards getting information about a cross section of these users. In fact, if we do not do so, we omit a central data source.
  • The third group to adress is the cannabis users we find in the drug treatment system. Here we may find very interesting data about what may have driven them into treatment, and whether interventions are useful or not . In combination with area two we might even find out which factors are important for turning to treatment system and whether these factors change over time.

Once we have studied cannabis use in these three entities (the population as a whole, the group of cannabis users within the population, and the users in the treatment system) we are able to draw a clear picture of what cannabis use is about.But what is it we should look at?

First we should look at the proportion of our populations above the age of 12 that has some type of experience with the use of cannabis. Lifetime experience is important, i.e. that is what proportion of our population has ever tried cannabis. But since many people are willing to try cannabis once or twice and then stop, this gives superficial information. We also want to know what proportion of those who have once tried cannabis continue to do so. Continuation rates and discontinuation rates can easily be established by conducting household surveys and asking the right questions.

Another important type of knowledge is the dynamics of these measures. Does the proportion with repeated experience with cannabis grow over time in the population? We know this if we conduct household surveys at regular intervals. The same is true for changes in discontinuation .

Another important measure we can find in household surveys is what other drugs the cannabis users in the population are trying or using regularly. Then our household surveys can tell us something about the length of cannabis use careers. When does cannabis use start? How long do cannabis use careers last? In other words, once cannabis consumption starts, does it last a lifetime or is it contained in certain age cohorts, meaning that over a certain age, people usually stop consuming cannabis?

We see that household surveys can supply essential information, that we need to have if we want to give our mostly ideological discussions on the quality and effects of drug policy some empirical basis. And if we want to seriously compare different drug control policies, we can not do without this information at all.

But there are some important data we can not derive from household surveys. For instance, if cannabis consumption continues over time, does it continue in a particular way (for instance always rising quantities of use) or is consumption very different over time per career period? This and deeper insight into what cannabis consumption means to a user, what functions, and what consequences it has within a certain social setting, can only be discovered if we direct more research attention to the group of users itself. We are doing this type of research in the Netherlands, and we have selected experienced users in three different cities for our research efforts.

Conducting this type research, one has to avoid a few pitfalls. The first one seems to be the quality of the sample. One should ideally investigate a random sample of experienced users to be sure that the research findings have some wider validity. However, I see as an even more dangerous pitfall the perspective from which the questions are asked. If one sees cannabis use principally as an expression of mental illness, one will ask questions about signs of mental illness. These signs will always be found (in any group), and unless very careful research is done among perfectly matched contrast groups, one will arrive at erroneous conclusions. Another possibility is to view cannabis use as behaviour that is integrated into normal life styles and subject to normal control mechanisms. Finding out how people control their use is then an option. This means that the perspective one views the consumption of cannabis from, is very important for the type of questions, and thus for the type of answers they lead to.

But once one has a set of reasonable and neutral questions about cannabis use, research among experienced users can add tremendously to knowledge about the functions and conse-quences of its use among different subgroups and cultures. If we conduct this research well, we may even be able to detect difference or similarities between countries. These comparisons may provide essential information about the still unknown role of drug policy in the level and types of cannabis use. Our Centre for Drug Research has initiated a comparative cannabis use research project, in which our investigations into the population and a community-based sample of experienced users will be replicated in Bremen (Germany) and San Francisco (California).

As we know from research in the past, only a small minority of cannabis users have serious longer lasting difficulties with cannabis use. This means that if we only do research among experienced users, we may find very little evidence of long term problems related to this use. That is why we have to know something about the group that runs into difficulties. What are their difficulties? Are they due to cannabis use, or were they there before? If they were there before, did cannabis use diminish these problems, or did it enlarge them? If the problems were enlarged, was this related to the drug, or to a complicated mixture, including the social risks of using drugs in a specific context? In other words, we have to conduc careful research among cannabis users who come into contact with the treatment system in order to find out what drives them into treatment, and on how they differ from regular users in the general population. We also need to find out whether treatment for these users is so rare because cannabis use is marginalised in most countries, and treatment is therefore not seen as a 'normal' option. It may very well be that the more cannabis use is accepted, the more of a rise in treatment demand there is apt to be. This is not because more users need it, but because more users perceive this option as being within the range of 'normalcy'.

For the next part of my presentation, I will draw from the knowledge we have gathered in the Netherlands in the three areas I have mentioned, and I will suggest an overall interpretation as regards the selection of relevant policies that can be derived from this research .

Cannabis use. Monitoring for health policies

In the Netherlands, we have some of the same problems as other countries in relation to drug use data: we have too little of them. For instance, we would have liked to have precise and valid data about drug use in the population in the period when our national harm reduction policies were initiated, the early seventies. If we had this data, and if we had repeated our measurements in the same way till now, we would be able to see drug use trends over time. If the same data had been available for other countries, we might even have developed hypotheses about differences or similarities between trends in different countries. We might even have had some modest knowledge about the role drug policies play in the determination of drug use prevalence.

Unfortunately, data about drug use in the population are so scarce and intermittent, and so incomparable between countries, regions or cities, that up till now we have not been able to create a sound theory about the role drug policies play in drug prevalence.

Korf has made an effort to contribute to this kind of theory by trying to make some sense out of the different data we have on Germany and the Netherlands. By comparing data on Amsterdam and Hamburg, he concludes that 'the coffeeshops increase the chance that people in their twenties and thirties start using cannabis or, more so, continue to use it' (Korf 1995, 91)[1]

This hypothesis is very interesting, and it is a pity we cannot verify it by comparing continuation rates and initiation ages in a precise enough way. The research and sampling methods that were used for obtaining the data on the city of Amsterdam and the limited data on the province of Hamburg, on which Korf founds much of his hypothesis, are simply too dissimilar to do so.

But imagine if we had all these data. And we were able to comprehend how drug policies interplay with drug use prevalence, how relevant would this be? Would we be able to use this data and such theory for health policies? I think we would, but not easily. The main difficulty is that health is not an objective entity, and the data and the theory we need for health policies may be vastly different across different definitions of health.[2]

In order to give drug use monitoring a place in policy making for public health, we have to understand that in addition to (comparable) indicators about drug use prevalence, we need some comparable definition of health in relation to drug use.As long as we omit doing so, our discussions are focussing on different issues,and we will have the sensation to be trapped in the dark. If drug use in itself is considered unhealthy or a threat to health, we are discussing a different vision on health than if some well defined potential consequences of drug use are considered a threat to health.Clarity about this is vital for any sort of discussion. Differences in perspectives on health will be reflected as well in types of drug control policies, and will give rise to completely different sets of needs around data.

In this presentation, I would like to concentrate on cannabis use, and discuss some data we have for cannabis use in Amsterdam. Since Amsterdam is so conspicuous in the international arena for drug policy, we may use this city as a playground, not only for drug policy experimentation, but also for drug use research, and most of all, for the type of theory we can develop around drug use in general.

I will show you some data for the prevalence of cannabis use in the general population, and some data about other drugs these cannabis users have experience with. Then I will show you some data we gathered on a group of experienced cannabis users within the general population. And I will conclude with some data about treatment clients who entered outpatient clinics for problems with cannabis.

Cannabis use in the general population in Amsterdam

Coffeeshops started to open in Amsterdam in the early 80's after a period of rather easy accessibility of cannabis in the late seventies via 'house dealers'. These house dealers operated in youth clubs, and were allowed to sell cannabis products to youngster who frequented these clubs. In Amsterdam, city subsidised clubs like Paradiso and the Milky Way were the best known ones.

So when we conducted the first local household survey on drug use prevalence in 1987, the population in Amsterdam had easy access to cannabis for at least ten years. We do not know the level of cannabis use before 1987, so it is impossible to say what the level of use was at the time of the change in the Dutch Opium Law in 1976, which formally turned cannabis use and possession into a misdemeanor.

We repeated these large household surveys in 1990 and 1994, and one is ongoing now in 1997, so we know a bit about the developments. The interesting thing is that between 1987 and 1994, cannabis use in the city remained very stable.[3] Lifetime experience did increase due to a generation effect from 23 to 29% of the population. Last 12 month prevalence and last 30 day prevalence however remained stable at about 10% and 6% of the population. The average age of onset of cannabis use has remained stable as well, at about 20, and the age when last 30 day cannabis use occurs is roughly between 15 and 35. Outside this age cohort it is rare, even in Amsterdam, where access to cannabis is fully open.

One of the many health risks that have been discussed for cannabis is its gateway function to harder drugs. This risk is one of the few that can be really tested on large populations of users.

We have done some secondary analyses on our household survey data in an effort to establish how many of the cannabis users in Amsterdam have lifetime experience with drugs like heroin, cocaine and ecstasy. Since have no lifetime heroin data for 1987, we only analyzed the 1990 and 1994 data for this purpose. Heroin experience was noted among 4.2 % of the cannabis users, cocaine experience among 21.7% and xtc experience among 7.9%. The problem with lifetime prevalence figures is that they include all types of experience, even one time experimental use. Looking at last 30 day experience is a far more reliable figure for more regular drug use experience. Last 30 day experience with cocaine among cannabis users is 2% , with heroin it is 0.2% and with ecstasy about 1.5%. This shows that even among the subgroup of cannabis users in the population, 98% of the cannabis users had not use any other drugs in the month before they were interviewed. Even if these figures might change somewhat over the next few years, we still see a huge majority of cannabis users opting for non-use of other drugs. Of the 2.368 cannabis users we found in the two household surveys in 1990 and 1994, there were only two persons who had taken heroin or cocaine more than 20 times in the thirty day period preceding the interview.[7] We can thus conclude that the alleged health risk of other drug use in cannabis users in the Amsterdam population is limited.

I do not want to discuss the question of whether other drug use is indeed a health risk. I myself think it is not, since this other drug use almost always remains within clear limits for most users, but this is a topic for a different discussion. We can see, however, that a policy like the one in Amsterdam that does not criminalise personal drug use and small personalized drug dealing of any type, does not lead to high prevalence figures of recent drug use, irrespective of which drug is observed. We should be prepared to accept the notion that drug control policy in the Netherlands is of limited influence on drug use prevalence.

In household surveys, one can only measure prevalence data and some other related data. For other health risks possibly related to cannabis use, one of the methods involves in depth research among a representative group of experienced cannabis users.

We have conducted this research by interviewing over 600 experienced cannabis users in three cities (Amsterdam, Utrecht and Tilburg). In Amsterdam we did a follow-up on 216 experienced cannabis users who were also in the 1994 household survey. Here I will only use some data from the Amsterdam segment of our cannabis use data. The 216 experienced users are a random sample, so there is good reason to assume that the results we measure in this sample are generalisable to all the community-based experienced cannabis users in the city.[8]

Of all the lifetime cannabis users in Amsterdam, a large minority of 43% becomes an experienced user, i.e. has used cannabis over 25 times. We asked these users 70 pages of questions divided into 12 topics : 1) initiation of cannabis use, 2) level of use through time, 3) patterns of use through time, 4) quitting and diminishing of use ,5) the use of other drugs and combinations of drugs, 6) buying cannabis, 7) contexts of cannabis use, 8) advantages and disadvantages, 9) prevalence of effects of use (more than a hundred potential effects are mentioned), 10) attitudes about cannabis and other users, 11) cannabis dependence both from a subjective angle and according to DSM-IV, and 12) use of cannabis at work.

Of course, what the risks are of cannabis use is not an objective problem. It has been battleground for ideological positions around drug use in general, and cannabis use in particular. Risks are definable in the realm of physical or mental functioning, and even on the level of the human cell. They can also be defined in matters of behaviour, which is more the type of risks we dealt with in our survey. How to measure the prevalence and relative importance of all the possible culturally determined health-related phenomena? This is an unsolved problem, and one of the reasons why scientific research is of limited value to international drug policy development ,which is unable to deal with complex issues like the one we are discussing here. It is more of value to national drug policies, where more restricted and homogeneous notions of health problems are defined. Research can also clarify local impacts on health risks by way of rigid and systematic comparative studies between different cities or regions, as we are now projecting in Bremen, Amsterdam and San Francisco.

Another more theoretical problem related to the notion of risk is that we have to measure the risk of some behaviour against its positive yield. The risk of breaking a leg skiing is huge compared to the risk of breaking a leg taking a walk in the park. But we accept some high risks if the benefits of the behaviour are high as well. So, on the basis of risk assessment, few skiers will exchange skiing for taking a walk in the park. Risk is always a relative matter, and as such a very complicated topic for research.

Also, health is not an objective entity. What we in Amsterdam consider healthy or unhealthy reflected in the wording of our questionaire. It is also reflected in answers we get, so to a certain extent even our outcomes are determined by our local bias. This can not be prevented. Every questionnaire is a reflection of political or professional preoccupations.[4] Sociologists ask completely different questions than psychologists or psychiatrists. This is very clear from the enormous difference between the topics of the recent Kleiber et al. study of cannabis users in Germany and our own.[5] Kleiber lives in a political and professional world where psychological and psychiatric questions are considered relevant. This means that in his design health is more defined in terms of scores on psychological scales than in ours. However, in our own user survey we opened the possibility for each respondent to insert his or her own definition of problems and/or health around cannabis. We introduced numerous open questions to tap into the notions of users themselves. Now, do not expect me to say that this is a sufficient precaution against local bias. Just as local researchers have their own implicit or explicit notions about drug harm, so have local users. National debates about drugs tend to influence the images local users have of themselves and of drug-related health issues and therefor on what needs reporting when asked about health. Of course we also introduced some fixed questions about health risks, and I will present to you here some of the results in relation to health I will present to you here.

One might consider a type of cannabis use the respondent describes as being high all day as a potential health risk. Of all the experienced users, 12% reported having been high all day during their period of heaviest use. Being high all day during the first year of regular cannabis use was reported by 4% of all the respondents. However, this intensity of use was hardly found in the 12 months before the interview (that is, with just 1 respondent). Another health risk was related to traffic accidents under the influence of cannabis alone, or in combination with alcohol. We asked all the respondents a question about any type of accident, even very small and inconsequential ones.

Out of 216, 12 respondents (or 6%) reported having had some sort of traffic accident under the influence of alcohol and cannabis, or cannabis alone.

It is quite striking that experienced users should mention so many disadvantages of cannabis, something we had also noted in our investigations of cocaine users. Our question about disadvantages is fully open, which means respondents can choose what terms to use to describe the disadvantages. We asked them to mention the most important four disadvantages of cannabis.

Cocaine users often formulated the disadvantages of cocaine in terms of physical health, but one of our most unexpected findings was that the cannabis users do not. The most often mentioned disadvantage was that cannabis makes one "vague and/or inactive", as was noted by 77 respondents or 38%. The second most often mentioned disadvantage was that it takes away one's capacity to focus on tasks, as was mentioned by 18 respondents or 9%. The third disadvantage was that cannabis can make one paranoid or confused, as was mentioned by 17 persons or 8%. 16 respondents noted as a disadvantage that cannabis makes you introverted. The fifth most often mentioned disadvantage was the only one that pertained to physical health, i.e. the disadvantage of smoking of tobacco in a cannabis joint, as was mentioned by 16 persons.

In some countries, we observed some preoccupation with a phenomenon called flashback. It involves a temporary return into a state of intoxication, without any use of cannabis. We asked all our respondents whether they had ever experienced a flashback. Just over a third of them had flasbacks, and we asked all of them to describe them. From these descriptions, it is clear that most respondents define a flashback as a very strong memory or association, either positively or negatively charged, and associated with some cannabis experience. A flashback in the sense of reintoxication without any use of cannabis is very rare.

Another potential health risk is dependency. Do cannabis users experience dependency? In order to find out, we included the seven DSM-IV indicators of possible dependency in our questionnaire. As you know, the DSM diagnosis of dependency is made if one answers yes to three or more items on a 7 item list. We wanted to maximize the possibility of finding traces of dependency, so we asked our respondents to answer these DSM-IV questions with their total use career in view. (We did not ask our respondents whether they had experienced the DSM-IV items in the last week, or in the last 12 months as is normally asked, but if they had ever had these experiences). Over an average period of cannabis use of 12 years among these 216 experienced cannabis users, just 53 (25%) had had assembled the necessary experience on 3 items or more that qualifies as DSM-IV dependency. We do not know whether these experiences occurred at the same time or during different periods. These 53 persons had an average career of almost 10 years of cannabis use.

It is hard to say if this is a lot or not. We can only know this if we compare these lifetime DSM-IV items in a similar group for other drugs such as alcohol. Unfortunately, we can not draw this comparison.

Of the 53 respondents who had ever had had the experience that met with 3 or more DSM-IV items during any period of their use career, 13 had considered some form of help for their cannabis use problems. However, at the time of the interview 47% of these 53 persons noted being abstinent no use in the last 3 months) 25% were low level users, 17% were medium level users and just 11% high level user.[6]

When we asked whether our respondents had had any contact with the drug treatment system during the last 2 years before interview, 11 (5%) answered affirmatively. Only one had indeed gone into treatment because of cannabis. However, during the total use career of these 216 persons, 19 (or 9 %) had, at some moment, been considering entering some drug assistance programma because of their use of cannabis.

This means that forms of dependency or other problems that necessitate drug treatment were a low risk among our sample of experienced cannabis users.

Some treatment data

In Amsterdam we have one treatment facility for cannabis users, mostly out patients. According to their registration data in 1990 and 1994, the same years we did our household surveys, the number of treatment clients has increased from 50 in 1990 to 176 in 1994. Since we have shown that the use of cannabis in Amsterdam was the same in 1990 as in 1994, this increase in treatment demand can not be related to increased use.

Frequency of treatment contacts per client in Amsterdam
1990 1994
n % n %
10 or less 14 28 122 78
More than 10 36 72 34 22
Total # of clients 50 100 156 100
Referred by clients' own volition 65 37
Court 15 5
Doctor 12 27
Other 8 31
Total 100 100

Although it is very difficult to give valid answers to the question of where the increase in treatment demand came from, I will try. One reason might be that at the time of a social normalization of cannabis use, we also see a normalization of cannabis problems. This means that subjective notions about problem-related cannabis use are no longer seen as very deviant by cannabis users, which might make it easier to apply for assistance in times of problematic cannabis use. This hypothesis is no more than a speculation. Another reason might be that the publicity about the cannabis treatment possibility at the local treatment center attracted more problem users to go into treatment. Or, this publicity may be related to more referrals to that clinic. The latter hypothesis can be checked.

In this table, we clearly see that the voluntary treatment clients dropped from 65% in 1990 to 37% in 1994. This means of course that the referrals of clients by others increased enormously. We may have found here the simple explanation of why the proportion of clients who have 10 or less treatment contacts increased so much, from 28% in 1990 to 74% in 1994. If people are referred by others, they are less motivated to remain in treatment than if they come out of their own volition.

The sharp increase in clients, and the sharp decrease in client contacts may be the result of a changed route into treatment. This means that we have to be very careful with comparisons between the treatment data of 1990 and 1994.

Of course, looking at the current cannabis use prevalence in the city of Amsterdam among 6% of the population above the age of 12, means there were 36.000 cannabis users in Amsterdam in 1994. If in 1994 the out patient clinics attracted 176 persons with cannabis problems, many of whom did not come from Amsterdam but from the suburbs, we can not say that cannabis use in the city often results in a treatment demand for cannabis. The rate is about 5 per thousand. This is a very low risk rate indeed.


Summarizing our results, we might say that between 5% and 25% of all the experienced users have at some time during their use career met with some problems or health risks (as we see them) related to their use of cannabis. A large majority has not. Combining our household survey data with our user survey data, we can easily see that most of the risks this minority met with were overcome without any professional help. Cannabis users now constitute a stable 6% of the population, and to a high degree cannabis use is only temporary, taking place as it does between the ages of 15 and 35. Under a system that does not try to prevent the availability of cannabis, current use is very low. As regards health policy, this could lead to the decision that money would be better spent on the prevention of specific risks during a use career than on the prevention of the use itself. This conclusion is valid, even if cannabis consumption rates would slowly climb over time, since the risks of cannabis use are so limited. Under these conditions, there is no reason to try to limit the use itself. Combinations of research projetcs like the one I have just described may play a role in efforts to get a clear view on the specific risks that a minority of users will meet. Once again, these risks may not be the same in different cultures or political systems and they may change over time.


  1. Dirk J. Korf: Dutch Treat. Thesis Publishers Amsterdam 1995.
  2. P. Cohen: The relationship between drug use prevalence estimation and policy interests. In: Gerry Stimson (Editor), Estimating the prevalence of problem drug use in Europe. Lisbon: European Monitoring Centre for Drugs and Drug Addiction. 1997
  3. Sandwijk, P., P. Cohen, S. Musterd and M. Langemeijer: Licit and illicit drug use in Amsterdam. Report of a household survey in 1994 on the prevalence of drug use among the population of 12 years and older. University of Amsterdam 1995.
  4. See for a discussion on the relation between types of data that are perceived as needed, and type of drug control system, P. Cohen: The relationship between drug use prevalence estimation and policy interests. In: Gerry Stimson (Editor), Estimating the prevalence of problem drug use in Europe. Lisbon: European Monitoring Centre for Drugs and Drug Addiction. 1997
  5. Dieter Kleiber, Renate Soellner and Peter Tossmann, Cannabiskonsum in der Bundesrepublik Deutschland. Berlin: Freien Universität, 1997.
  6. High level use is more than 10 grams of cannabis products a month. Medium use is between 2.5 and 10 grams a month; low level use is less than 2.5 grams a month.
  7. See for an extensive presentation of data describing other drug use by cannabis users "Cannabis use, a stepping stone to other drug use? The case of Amsterdam".
  8. Peter Cohen and Arjan Sas: Cannabis Use in Amsterdam. 1998 CEDRO.