Abraham, Manja D., Hendrien L. Kaal, & Peter D.A. Cohen (2002), Licit and illicit drug use in the Netherlands 2001. Amsterdam: CEDRO/Mets en Schilt. Pp. 23-32.
© Copyright 2002 CEDRO Centrum voor Drugsonderzoek.

[Press release] [Index] [Next Chapter]

Licit and illicit drug use in the Netherlands 2001

Foreword / Executive summary

Manja D. Abraham, Hendrien L. Kaal, & Peter D.A. Cohen


In the following texts we offer the results of our second national drug use survey and a detailed description of the methodologies we used to reach these results.

Without the assistance of many people this second survey of the Dutch population would 01not have been possible. The Ministry of Health, Welfare, and Sports funded our complex methodologies with elegance and curiosity. We thank the Ministry's Bob Keizer, Marcel de Kort, Fons Vloemans, Ruud Bik, Wil de Zwart, and Ton Cramer for their support.

Statistics Netherlands again co-operated with our study in very useful and interesting ways, and we thank Dirkjan Beukenhorst, Jos de Ree, Jeroen Winkels and José Geurts for their zeal, excellent expertise, and readiness to deal with the many problems we met along the way. Perhaps their most welcome support took the form of repeatedly reminding us that our response rates, however unsatisfactory, were the same as those found in all survey research in the Netherlands and had nothing to do with us!

Our field work partner, NIPO of Amsterdam, made our new Multi Method approach possible, and we thank Jacqueline de Wijs, Patrice Weijer and Henk Foekema for their hard and sustained effort, and of course for putting up with our monitoring of the field work every week (sometimes even twice a week!). We were very happy to be able to engage again the expertise of Roelf-Jan van Til of ABF Research in Delft to run the analyses necessary for our non-response survey.

The University of Utrecht also helped us a great deal by allowing Professor Harm 't Hart to be our methodological conscience and jack-of-all-trades. Harm 't Hart has been a great help in all our epidemiological work since 1987. His recent retirement did not stand in the way of making his vast knowledge and his delightful common sense available to us whenever it was most needed.

Last but not least we thank Arjan Sas of the University of Amsterdam for his many talents, above all untangling computers when they seemed to resist all efforts to tempt them into working for us.

July 2002

Peter Cohen, Manja Abraham, Hendrien Kaal

Excecutive summary


This report describes the results of the national population survey conducted by CEDRO in the year 2001 to measure the use of licit and illicit drugs in the registered population aged 12 years and over in the Netherlands. This survey is the second of its kind, after the first survey was conducted by CEDRO in 1997 (Abraham et al. 1999). Almost 18,000 respondents were questioned about some lifestyle variables, use of licit and illicit drugs and, in case of last year drug use, where these drugs were obtained.

Drugs included in the study are tobacco, alcohol, hypnotics, sedatives, cannabis, cocaine, amphetamines, ecstasy, hallucinogens, mushrooms, a range of opiates, inhalants, so called 'smart drugs' (newly added in 2001)[1] and performance enhancing drugs. In this report drug use prevalence rates are given for the population of the Netherlands but also for seven sub-samples. As in 1997, the population was divided by five different categories of address density, allowing one to show again that drug use varies considerably by address density. Samples of the populations of Amsterdam and Rotterdam were realised as in 1997.


The report is split in two parts in order to facilitate its use. Part I focuses on the survey itself: the methodology, response, representativity, non-response and mode analysis. Part II presents the outcomes of the survey: the drug use figures.

PART I: Survey Methodology

Chapter one presents an extensive overview of the methodology of the 2001 national survey and the pilots that preceded it. For the second time in the existence of this drug use survey important changes in survey methodology had to be introduced. A shift was made from Computer Assisted Personal Interviewing (CAPI) - used in 1997 and before - to an open choice methodology of data reporting, called Multi Method (MM). It was not feasible to repeat the CAPI method for the entire 2001 national survey. Because of the tight labour market it was very hard and, especially in Amsterdam and Rotterdam, almost impossible to recruit CAPI interviewers. Other aspects of the methodology, such as the questionnaire, sample protocol, and weighting method, remained the same.

Data collection

The MM drug use survey enabled respondents to fill in a paper and pencil version of the questionnaire, or to respond to the questions by computer. Respondents would typically receive:

  • An invitation to co-operate in the survey and an offer to receive a ƒ 25.- (€ 11.34 ) reward after returning the questionnaire[2];
  • A paper and pencil questionnaire;
  • A diskette based questionnaire for DOS or Windows machines; and
  • The URL of a website for downloading a web version of the questionnaire.

If after a period of around three weeks the respondent had not returned his or her data, a reminder routine would be started that consisted of several elements. One of the reminder routines was to find the respondent's telephone number and offer a CATI survey when the respondent was reached by phone. The other often-used reminder - in case no telephone number was available - was to send a second envelope with a diskette and paper and pencil questionnaire.


Stratified sampling of respondents was done exactly as in 1997, i.e. by first drawing a sample of municipalities per address density stratum, and then sampling respondents from a chosen municipal's registry of the population. Amsterdam and Rotterdam were oversampled. All sampling was done by Statistics Netherlands (CBS) in the same way as in 1997; the Amsterdam sample was drawn by the Amsterdam registry. Drawing independent samples from each of these categories enables one to look at drug use per address density category. More importantly, if drug use measurements are repeated, this type of sampling might show differential drug use developments between the density categories, increasing the understanding of drug use over time. In the course of 2002 a separate document will be published in which the seven samples will be used to see if drug use is developing in different ways per address density. The population sample was of persons of 12 years and older, with a boost for the age group 12-19.

In chapter 2, about response and representativity, the reader is presented with a series of analyses about data quality and response rates. For this national survey, the two cities included, a gross sample of 40,573 persons was approached. This resulted in 17,655 successful data sets, a valid response rate of 47.1 per cent. Chapter 2 shows in detail how the sample and the response fit the population on a series of variables (age group, marital status and gender).

To be able to assess the impact of mode shift from CAPI to MM, a randomly selected sub-sample of the gross sample was interviewed in the same way as done in 1997, with CAPI. Getting respondents to co-operate was most difficult in Amsterdam.

In chapter 3, the complications of non-response are discussed. Non-response is a problem of immense importance, and in spite of the MM approach - emboldened by a ƒ 25.- (€ 11.34) incentive per respondent for each self-reported data set - non-response has been growing. To have some overview of the effects of non-response a non-response survey was conducted, asking a sample of 10,352 persons from the 19,852 respondents that had refused to participate or had not been contacted, to answer a very short list of questions, including items as lifetime prevalence of cannabis use. The response rate of the non-response survey was 27.2 per cent (n=2,812). The non-response survey should answer the question if or to what extent non-response influenced the outcomes of the main survey. In previous years it was found that the non-response data did not suggest that the findings of the main survey underestimated drug use prevalence. This year, under more complicated survey circumstances, a similar conclusion was reached.

In chapter 4 the CAPI and MM methods are compared. In the NPO 2001 survey both CAPI and MM were used alongside; around 8,000 persons were approached for CAPI and around 32,500 persons were addressed following the MM protocol. Gathering data with both methods allowed comparing the CAPI data collection mode with the MM mode. Also, if the 2001 MM mode would prove unsatisfactory, it would still be possible to compare the 1997 national and Amsterdam data (all CAPI) with the national and Amsterdam data collected by CAPI in 2001.

Response rates for CAPI and MM were similar, 48.7 per cent and 46.8 per cent respectively. Because some non-response with MM could in reality be frame error, real response for MM is likely to be higher than stated above. Some differences were found in the type of respondents: e.g. juveniles were found to be more likely to respond to MM, while senior citizens are more likely to respond to CAPI.

In order to compare survey results between the two modes, data was weighted for oversampling effects, and the effects of selective non-response. It was found that CAPI resulted in slightly lower drug use prevalence estimates than MM. For a small number of drugs these differences are reversed. However, last month drug use prevalence figures do not show these differences. Furthermore, it was found that for some drug use prevalence estimates the 2001CAPI data yielded lower levels than the 1997 CAPI data.

On the basis of these comparisons, it is concluded that it is justified to combine two data sets that are collected by two different approaches (i.e. the CAPI and the MM sample) and regard the total sample as representative for the research population.

Part II: The Prevalence of Drug Use

Chapter 5 describes the prevalence of drug use expressed in core indicators. Drug use is described with three prevalence indicators (lifetime, last year and last 30 days). Lifetime and last month prevalence rates are also given for each drug per gender. Other core indicators are last year and last month continuation, and percentages of experienced users per drug. For alcohol, hypnotics, sedatives and cannabis, the number of days of use per drug, for the last 30 days prior to interview, are given. Incidence, mean age of initiation, are reported, concluding chapter 5 with mean age of current (last 30 day) users. All these indicators are given for each of the seven independent samples. Amsterdam is still by far the area where most drug use takes place. For instance, lifetime prevalence of cocaine in Amsterdam in 2001 is 10 per cent (up from 9.4 per cent in 1997) versus 5.2 per cent in Rotterdam (up from 3.4 per cent). It was found, as it was in the former survey, that the inhabitants of Rotterdam use slightly less drugs than those in the most populated areas in general.

Cannabis lifetime use in the Netherlands shows an expected increase, from 15.6 per cent in 1997 to 17.0 per cent in 2001. In the lowest address density municipalities, lifetime cannabis use is much lower, but increasing as well (from 10.5 to 11.4 per cent). For cocaine there also is an upward trend in the Netherlands, from 2.1 to 2.9 per cent, as there is for ecstasy, from 1.9 to 2.9 per cent. All classic party drugs show an increase, which is even true for alcohol (from 90.2 to 91.6 per cent).

Opiates lifetime use in the Netherlands shows a sharp decrease, from 11.7 per cent in 1997 to 8.2 per cent in 2001, partly due to an unexplained decrease in the use of codeine. Lifetime use of heroin increases from 0.3 per cent in 1997 to 0.4 per cent in 2001. The Amsterdam figure for lifetime use of heroin is 1.3 per cent. This figure is down from 1.8 per cent in 1997, but it remains by far the highest in all of the Netherlands. The trend in lifetime use of cannabis in the Dutch capital goes the opposite way (up from 36.7 per cent to 38.1 per cent) which shows once again that an increase in the use of cannabis type drugs does not trigger increased heroin use, despite this being forecast by hypotheses of cannabis as gateway drug.

In all of the seven samples except one, tobacco use is falling, which results in an average lifetime use of tobacco of 66.4 per cent in 2001 (down from 67.9 per cent in 1997). Also falling are the lifetime use figures for performance enhancing drugs - anabolic steroids being the most well known. Here one sees an already low figure of 0.9 per cent decreasing to 0.7 per cent.

Continuation rates indicate what proportion of lifetime users of a drug used this drug in the last year and the last month respectively. For instance, the national last month continuation rate for mushrooms is 3.5 per cent, suggesting very infrequent use. The last month continuation rate of heroin is 15.1 per cent, comparable to last month continuation rates for cannabis, cocaine and ecstasy. The national last month continuation rate for hypnotics and sedatives is around 46 per cent, and for alcohol it is 75.1 per cent. The rest of the indicators reported in chapter 5 - proportion of experienced users, frequency of use, mean age of first and current drug use - are not discussed in this short summary.

Chapter 6 deals with the same subject matter as chapter 5, organized in a different way. In chapter 6 the reader will find the lifetime and last month prevalence figures, broken down by age groups, ten in all. Drug use prevalence rates are reported for the following age groups: 12-15; 16-19; 20-24; 25-29; 30-34; 35-39; 40-49; 50-59; 60-69 and over 70. The age group 12-19 years of age was oversampled in order to be able to compare the drug use figures in this group with school survey data. Such a comparison of these data with the school survey has been conducted with the 1997 data (Langemeijer 1997; see also Gfroerer 1997). This showed that the data from the random population sample suggested lower prevalence rates than the data from the school survey. This comparison will be repeated with the 2001 data in the near future. In the same table one will find lifetime prevalence of the total sample, last month prevalence, last month continuation, percentage of experienced users and mean age of first use. These tables are created for a fast and informative overview of main findings per drug and age group.

An index of drug use is included, setting all figures of 1997 at 100. These tables allow observations about relative increase or decrease, per age group. This shows, for example, that the use of hypnotics amongst 12 to 15-year-olds has increased with 48 per cent, an increase higher than in any other age group. In the same age group, the use of sedatives dropped with 15 per cent and the use of cannabis dropped with 22 per cent. These tables allow very rapid observation of drug use development per age group, for each of the seven independent samples. The age group of 20 to 24-year-olds shows an increase in almost all drug use categories. Lifetime cannabis use rises with 32 per cent; lifetime cocaine, amphetamine and ecstasy rises with around 120 per cent; exception is the stability of opiates. Clearly, 20 to 24-year-olds are most active in visiting café's and dancings, and it is this age group that is most important when looking at drug use.

Chapter 7 reports the place of acquisition of the drugs that last month users have consumed. The questions about place of acquisition or purchase are relevant to examine one of the main hypotheses behind Dutch cannabis policy: allowing access to cannabis users decreases potential contacts with sellers of other drugs. Comparing the national 1997 data to the national 2001 data one finds for the age group under 18, that 38 per cent of cannabis transactions takes place in a coffeeshop, while for the older than 18 group this is 49 per cent. This is almost the same as in 1997. This means that more than half of all cannabis acquisitions take place outside the coffeeshop. Possibly it is the case that many cannabis users are low-level users who so rarely acquire cannabis that friends and relatives remain their most preferred access, irrespective of number or proximity of coffeeshops.

As in 1997, two per cent of all cocaine transactions took place in a coffeeshop; in 1997 55 per cent of all last year cocaine transactions took place in contacts with friends, and this percentage remains the same in 2001. Dealers operating from homes are the second largest source of last year cocaine, both in 1997 (20 per cent) as in 2001 (17 per cent).

Chapter 7 gives combined tables of place of acquisition for cannabis, cocaine, amphetamine, ecstasy, hallucinogens and mushrooms, for all last year users of the listed drugs. A separate table is created for place of acquisition of performance enhancing drugs showing that trainer, sports club or gym, are the most important source of these drugs (used by a very small proportion of the Dutch population, see chapters 4 and 5), followed by relatives and medical prescription. This is a remarkable shift from 1997, where trainers, sports clubs or gyms were mentioned by less than one tenth of all last year users.


  1. Smart drugs are a class of synthetic and natural supplements taken to enhance cognitive function. They are legally available and sold in 'smartshops'.
  2. The fieldwork organisation could also pay the reward into the account of an institution of choice if the respondent wished it to do so.