Abraham, Manja D. (1999), The impact of urban residency and lifestyle on illicit drug use in the Netherlands. Journal of Drugs Issues, Summer 1999, pp. 565-586 (pre-publication version).
© Copyright 1999 Manja D. Abraham. All rights reserved.
The impact of urban residency and lifestyle on illicit drug use in the Netherlands
National drug use prevalence figures represent average rates of use among the general population. It is argued here that these numbers mask interesting and important differences between local areas. For example, municipalities and cities exhibit markedly different prevalence rates, as is illustrated by the Dutch case. In 1997, researchers asked approximately 22,000 respondents, selected from large samples of the populations of four major cities and five categories of distinct address density municipalities throughout the Netherlands, about their lifestyles and use of licit and illicit drugs. The data collected indicate divergence between drug use prevalence in urban and rural municipalities. Other indicators, such as age of first use, seem unaffected by address density. The survey data reveal distinctions between drug use prevalence in Amsterdam, Rotterdam, The Hague, and Utrecht, although the cities occupy the same level of address density. The core question addressed here is how can the discrepancies in prevalence rates be explained?
In many countries, drug use statistics are utilized as key elements in the formulation and adjustment of national drug policies (see e.g. Tweede Kamer, 1997). They are also of considerable interest to the general public. One way of measuring the extent of national drug use is through surveys of the general population. This paper is largely based on the results of such a survey, titled Licit and illicit drug use in the Netherlands, 1997 (Abraham et al. 1999). This survey provides prevalence estimates of both licit and illicit drug use, and is based on a nationally representative sample of the registered population aged 12 and over. The registered population is nearly the entire Dutch population, and even includes most homeless persons. A total of 21,959 respondents participated in this survey.
Earlier, between 1970-1991, six national household surveys were implemented in the Netherlands among persons of adolescent age and older (Korf 1995). These surveys were marked by a relatively small number of respondents, ranging from 910-1,123 persons out of in approximately 10 million (Dutch population of 12 years and older in 1970) to 13 million (Dutch population of 12 years and older in 1991) (Statistics Netherlands 1997). Given the small sample sizes involved, their results are limited for two reasons. Firstly, matching confidence intervals are large and as a consequence less reliable estimates can be made. Secondly, rarely used drugs (such as heroine or inhalants) require a large sample (see e.g. Sandwijk 1995).
Although there has been a paucity of adequate national data on substance use, more attention has been directed at those who are most likely to be involved with drugs, in this case Dutch youth. In the period from 1970-1997, eight school surveys were undertaken among Dutch schoolchildren (Kuipers et al. 1997; also see Kuipers in this issue). Unlike the national efforts described above, these surveys have included a substantial number of respondents (ranging from 6,451 to 22,971 schoolchildren).
The net effect of concentrating one's analysis solely on national figures is likely to suggest homogeneity of drug use patterns in a country. A closer, local, analysis demonstrates that there are significant differences in drug use prevalence rates among urban municipalities of varying densities, and these should not be overlooked. These comments apply not only to the Netherlands, however, as they reveal a universal concern when attempting to estimate drug use patterns in a national population. Given these observations, differences in the over-all degree of address density (as an indicator of urban residency) should be taken into account when comparing international prevalence data. Differences of address density can be found between countries as well as within countries. For example, Ireland is populated by 51 persons per square kilometer (in 1996) compared to Denmark by 122 persons (in 1996), and The Netherlands by 458 persons per square kilometer (in 1997). If we take a closer look on the Irish inhabitants, we see that the area around Dublin area is populated by over 1000 persons per square kilometer wereas almost all other districts have less than 50 inhabitants per square kilometer. (source: Grote Bosatlas 1998)
The core objectives focus of this paper are the delineation of differences in drug use prevalence rates among municipalities with different address densities. We explain these patterns in terms of urban residency and lifestyle.
The most recent national survey of licit and illicit drug use in the Netherlands was based on a national sample representing all persons 12 years of age and older who were listed in the Municipal Population Registry as of January 1, 1997 (for Utrecht, the target date was January 1, 1996). Because the survey is conducted among registered persons, some marginalized drug users, homeless and illegals are not included in the samples. However, high-school dropouts, often omitted in many similar surveys, are included. The population from which this sample was drawn consisted of 13,242,208 persons.
The study sample was comprised of nine sub-samples (see Table 1). Four of these sub-samples were drawn randomly from a list of registered persons in the large cities of: 1) Amsterdam; 2) Rotterdam; 3) The Hague; and 4) Utrecht. The remaining five sub-samples were selected on the basis of a two-stage stratified sample of the rest of the Netherlands. The selection process consisted of a number of related steps. First, all other municipalities were ranked in one of five strata, representing different levels of address density. Stratum 1 represented all municipalities with more than 2,500 addresses on average, per square kilometer. Stratum 2 encompassed municipalities with 1,500 to 2,500 addresses per square kilometer. The third stratum included areas with 1,000 to 1,500 addresses. Stratum 4 had 500 to 1,000 addresses per square kilometer. And Stratum 5 had, on average, less than 500 addresses per square kilometer. Within each stratum, a random sample of municipalities was drawn. Random samples represented all persons 12 years of age and older who were registered in the selected municipalities. Selective data can be provided for each of the separate sub-samples.
Later, in the analysis, we will classify the samples in 5 discrete categories. These strata are made up of persons living in: 1) municipalities with the highest address density (including the samples of the four big cities); 2) municipalities with high address density; 3) municipalities with moderate address density; 4) municipalities with low address density; and 5) municipalities with the lowest address density.
In order to provide figures for the entire Dutch population, the response data were weighted to produce a representative sample. Table 2 summarizes the weighted and unweighted background characteristics of respondents, including age, gender and social activity levels (the amount of time spent going out to bars, clubs and the like).
Altogether, a total of nearly 22,000 persons were interviewed. Respondents were questioned in a computer-assisted personal interview (CAPI). By this method, the interviewer questions respondents at their homes, guided by a laptop computer. The computer minimizes routing errors and instantly alerts for inconsistency of given answers.
The questionnaire asked about the use of various licit and illicit drugs, as well as respondent background and lifestyle characteristics. It began with questions about personal lifestyle issues. Examples of these lifestyle measures included the following questions: "How many evenings were not spent at home during the last week?"; "What is the frequency with which you went out to a club, disco or bar during the last four weeks?"; and "What is the frequency with which you went out to a theatre or ballet during the last eight weeks?" Following these introductory inquiries, the interviewer asks detailed questions about the subject's use of particular drugs (these cover, among others, the frequency and intensity of use and age of first use), including tobacco, alcohol, sedatives, hypnotics, cannabis, cocaine, ecstasy, amphetamines, hallucinogens, mushrooms, opiates, inhalants and performance-enhancing substances. The questionnaire concluded with a series of demographic questions to ascertain respondents' background characteristics. These included conventional variables such as income, education, age, gender and marital status, as well as others concerned with socioeconomic status. Most of the interviews were completed in 1997 and 1998, although, as noted, the Utrecht fieldwork took place in 1996. The survey was designed with the co-operation of Statistics Netherlands (CBS), and funded by the Ministry of Health, Welfare and Sports (VWS).
Although one of the best ways to assess the nature and extent of drug use is by completing confidential national surveys (Harrison et al.1995: 206), this method also has its limitations. For example, the data is self-reported. Asking people to report private or personal information, and specifically about drug use, can lead to overestimation, underestimation and/or selective non-responses.
National drug use prevalence figures
In the following discussion the author focuses on cannabis, ecstasy and cocaine, as these are the three most prevalent illicit drugs in the Netherlands. Data are also included about abstinence rates. The following concepts are utilized to measure different aspects of drug use:
When considered together, these key indicators reveal important aspects of drug use patterns in a population. They may be considered the core figures of drug use. For a more complete description of drug use patterns, they would have to be supplemented by other measures such as last year prevalence, incidence and the number of times used per month (Abraham 1998).
Dutch drug use in an national perspective
The core figures of cannabis, cocaine and ecstasy use in Holland are shown in Table 3. Not unexpectedly, the data reveal that cannabis is the most frequently used illicit drug in the Netherlands. In 1997, an estimated 15.6 percent of the Dutch population 12 years of age and older had tried cannabis at least once in their lifetimes, and 2.5 percent had done so in the month prior to the interview. Cocaine was used at least once in their lifetimes by 2.1 percent of the Dutch population, and its last-month prevalence was 0.2 percent. A lifetime prevalence of 1.9 percent was found for ecstasy (MDMA), while 0.3 percent had used the drug in the last month. Although cocaine and ecstasy lifetime prevalence rates are substantially lower than those for cannabis, this does not imply that other core figures for these drugs would follow a similar pattern. For example, cannabis and ecstasy exhibit almost identical levels of continuing use (15.8 and 14.0 % respectively), compared to a lower level for cocaine (10.0 %). 33.1 Percent of lifetime cannabis users had consumed the drug more than 25 times, compared to 25.4 percent of the ecstasy users and 22.7 percent of the cocaine users. On average cannabis is first used at age 19, while cocaine and ecstasy are used for the first time at age 23.
As mentioned before, survey research has been widely used in the Netherlands. Of course different general population surveys can not readily be compared with one another because of the varying target populations involved and methodological variations and inconsistencies. The surveys that have been completed do indicate a slight increase in lifetime drug prevalence in the Netherlands in the period between late 1970 and the early 1990s (Korf 1995).
When comparing these national surveys, several methodological differences are apparent. We list only a few of them. The first is the disparity in the sample sizes of the different administrations. The 1997 survey included 21,959 respondents, whereas each of the first three surveys only included between 910-1,123 respondents. In addition, the data collection instruments used differed in the surveys; the 1997 study featured face-to-face interviews, while the earlier ones had been either written or face-to-face administrations. The actual questions asked in these different applications also varied considerably. Finally, the target populations were non-uniform; the 1997 survey included all registered persons aged 12 and over, while the previous surveys included persons aged 15 and over, 16 and over and 18 and over.
Dutch drug use in an international perspective
In order to place these Dutch figures in an international context, they will be compared with those found in some other European countries and in the United States. Table 4 presents the lifetime prevalence of drug use (cannabis, cocaine and ecstasy) reported in recent national general population surveys completed in the Netherlands, Denmark, the former West Germany, Spain, Sweden, the United Kingdom and the United States (European Monitoring Centre for Drugs and Drug Addiction 1998).
The international drug use prevalence data indicate that the lifetime cannabis use rate in the Netherlands is lower than that found in Denmark, the U.K. and the U.S., but higher than is found in Sweden. Table 4 reveals that these rates vary considerably from a low of 9 percent in Sweden, to highs of 31 percent in Denmark and 33 percent in the U.S. Unfortunately these figures are not directly comparable because the Danish survey is limited to those 18-69 years of age, while the U.S. and Dutch figures include everyone 12 years of age and older. This means that the Danish figures are not really as close to those for the U.S. as they seem. The 12-17 year old group is included in the U.S. figures but not in those for Denmark. Ultimately, this has the net effect of inflating the Danish figures, because the 12-17 year olds are not likely to be experimenting with marijuana.
As noted in the Maris article in this issue, the U.S. drug czar was recently critical of Dutch drug policy in comments made before his visit there in 1998. General McCaffrey claimed that the Dutch drug use rates were higher than, or at least comparable to, those found in the U.S. (The figures presented above call into question this assertion). For example, the Dutch lifetime cannabis prevalence rate was less than one-half that found in the U. S., despite Holland's liberal policies.
Now it is not easy to compare between countries. We see for example that there are many distinctions between the various surveys implemented throughout Europe. One must be cautious when comparing statistics between countries that utilize different sampling methods (such as populations with non-comparable age compositions) and sample size. One must be aware of mode effects because different instruments are used to collect data (such as face-to-face interviews or written questionnaires). One must also ask whether the drug use questions are imbedded in a larger survey, such as in Great Britain's Crime Survey and Greece's Mental Health Survey, or are they the central concern being addressed, as with the Dutch sample.
All of these factors serve to constrain cross-country comparisons. This having been said, however, there are major differences observable in the collected data. Given these methodological and sample constraints, a tentative conclusion may be that the experimentation with cannabis, cocaine and ecstasy in the Netherlands does not occur more often than it does in other Western-European countries. A helpful overview of the problems associated with general population surveys of drug use in Europe may be found in Bless et al. (1997).
Prevalence and address density
Dutch drug use in national perspective
The next question we address asks how the Dutch drug use figures are related to address density. Reason for doing so is that for at least several European countries, urban illicit drug use prevalence rates are substantially higher than are their national figures (Hartnoll 1995). Also, in the U.S. a positive correlation has been found between the degree of urban residency and rates of illicit drug use (Substance Abuse and Mental Health Services Administration 1997).
We give the data for all five density categories, but mainly focus on the two density extremes: the highest density stratum (n = 12,796) and the lowest density stratum (n = 2,304). Table 5 displays drug use rates for cannabis, cocaine and ecstasy, as well as rates of abstinence. These figures compare the rates for those living in the density strata with each other, and finally, for the nation as a whole. Differences between the strata are tested by means of a Chi-square test (p<0.05).
Cannabis is the most widely used illicit drug in the Netherlands. In 1997, an estimated 15.6 percent of the Dutch population 12 years of age and older had tried cannabis at least once in their lifetime; this represents just over 2,000,000 persons. A distinct relationship was established between cannabis use and address density. When comparing drug use rates among the different strata, both lifetime and last-month prevalence rates are significantly highest in the highest density municipalities, and finally, by the lowest density municipalities. Current use varies from 4.9 in the highest address density municipalities to 1.5 in the lowest. This suggests that the lower the address density, the lower the current cannabis use prevalence. The distinctions between the density divisions are also found among continuing use and experienced use rates, although these are less profound. Among the highest density lifetime users, 19.4 percent consumed cannabis at least once during the last month, whereas in the least populous stratum 14.6. The mean age of first use, however, is nearly the same throughout the country (19 years).
Cocaine is the second most prevalent illicit drug, being used at least once in a lifetime by 2.1 percent of the Dutch population. As with cannabis, cocaine use is strongly related to address density. The distinction between the strata representing different levels of urban residency also holds for lifetime cocaine use prevalence, and experienced use rates, although not for continuation of cocaine use and the mean age of first cocaine use. Here, the lifetime prevalence rate is highest in urbanized municipalities (4.9 %) and lowest in the rural municipalities (1.0 %). Thus, in regards to cocaine, the higher the address density, the higher the percentage of lifetime and current use. The current continuation of cocaine is 10 percent nationwide, a rate that is consistent for all population strata. Consequently, there appears to be no specific relationship between cocaine use continuation and the degree of urban residency. Finally, the experienced use rates are not significantly (p<0.05) related to the degree of urban residency. In the highest density municipality 25.1 percent of the lifetime users had used cocaine more than 25 times, while in the lowest density municipality this percentage was 24 percent. The age of cocaine use onset is around 23 nation-wide, with only modest differences between varying density populations.
In 1997 ecstasy had been used at least once by 2 percent of the national population, and was used in the month prior to the interview by 0.3 percent. Ecstasy is now the third most prevalent illicit drug in the Netherlands. Prevalence rates vary widely in the different address densities, however. For example, the lifetime prevalence rate for the high address density municipality was 3.6 percent, while it was 1.2 percent for the lowest address density municipalities. Current continuation rates vary for each stratum, but are not necessarily higher in the highest address density municipalities. The same holds true for experienced use rates in the most urban municipalities, where 22 percent of the lifetime users took ecstasy 25 times or more. The comparable figures vary between 12 percent in high density to 41 percent in the low-density municipalities. The age of first ecstasy use also appears somewhat lower outside urban areas.
In summarizing this data, two conclusions may be derived. First, there is a consistent positive relationship between the reported prevalence of drug use and address density; the higher the density, the greater the use. National survey results suggest that this generalization is applicable to the use of all illicit drugs. Secondly, not all drug use indicators reflect this pattern. Think, for example, of variations observed with regard to "mean age of first use" and "continued use". Similarly, the prevalence rates for licit drugs such as alcohol, tobacco, hypnotics and sedatives, as well as those for drug abstinence, do not reflect this urban residency /high use pattern (Abraham 1999).
Clearly, we can express national figures without knowing the actual distribution of drug use in the Netherlands. Rather, these figures tend to mask important differences between municipalities with varying address densities. Given the significance of this phenomenon, any cross-national comparisons seem improbable, other than as very broad indicators of approximate levels of use.
Drug use, address density and lifestyle
Explanatory drug use prevalence variables
The previous section provided data establishing that there is a distinct relationship between drug use (cannabis, cocaine and ecstasy) and distinct address density levels. Now we address the interesting question of how these differences can best be explained. As mentioned before, we first examine the explicatory variables for cannabis, cocaine and ecstasy use, and then determine whether or not they are the same for each of the three population density divisions.
Our aim is to determine which explanatory factors relate to the use of each of the named drugs, and to express that relation as a measure. This can be achieved by means of a logit analysis, an approach that can be used for explanatory models of categorical phenomena. This is the natural complement of the regression model in case the regressand is not a continuous variable (for more information about logit, see e.g. Maddala 1983). We used a multinomial logit model to analyze categorical variables. The logit analysis applies a probability (t-test) to all explanatory variables in the model, which tells if the variable is significant, and if so, by how much. In order to compare the relations occurring among the strata, this analysis is performed separately for each stratum. Three ranked categorical dependent variables reflecting a level of drug use were constructed. These variables indicate the level of use of each of the three substances, cannabis, cocaine and ecstasy. The substance variables are ranked as follows: 0 = no substance use; 1 = lifetime substance use, no last year/month; 2 = last year substance use, no last month; and 3 = last month substance use. Then, nine categorical independent variables were selected: gender; age; marital status; income; education; composition of household; number of evenings spent at home during the last week; frequency of going out to a bar, club or disco in the last four weeks; and number of visits to the theatre, ballet, a restaurant, bar, club or disco (last four weeks).
Results of the logit analysis are summarized in Table 6 for the highest and lowest address density strata, and also national. Explanatory variables are only provided if they were significant when using a t-test p<0.001. The analysis suggests that there are three major explanatory factors, which are of significance for all of the examined drugs, including age, gender and the frequency of going out to bars, clubs or discos. Outcomes show, for example, that age is negatively related to both cannabis and ecstasy use. That is, the use of these drugs is more likely among young people than it is among older people. Gender is negatively related to all forms of drug use prevalence, meaning that men are more likely to use drugs than women. The frequency of going out to bars, clubs or discos is positively related with cannabis, cocaine and ecstasy use prevalence. One must exercise caution in the interpretation of the meaning of the conclusions, however, for there is no evidence that a causal relationship exists between these explanatory variables and drug use prevalence.
One of the limitations of the logit analysis is that we can not readily contrast the relationship strength between strata. This non-comparability is a byproduct of the variable prevalence rates that characterize different address densities. Consequently, we should be cautious when we argue, for example, that age is more robustly related to drug use prevalence in the high address density areas than it is in low-density locales. A high prevalence rate provides more observations and therefore enhances the probability that a significant relationship will be found, explaining the relatively high values found in the national model when compared to the lower values exhibited when the strata are handled independently.
Age and gender
Age serves as a first explanatory variable for cannabis and ecstasy use, but is not one for cocaine use. These findings partially support previous research among the AMSTERDAM population, where we learned that drug use prevalence is correlated with age. As expected, few elderly people report having used either cannabis, cocaine or ecstasy (Sandwijk et al. 1995), a result that has been consistently found in non-Dutch populations as well (Korf 1995 and Substance Abuse and Mental Health Services Administration 1997). A possible explanation then for differences in prevalence rates may be the younger average age of the highest address density stratum. Many young people move to the cities to study at high schools or universities. In 1997, the group of 18-34 year olds in the highest density municipalities comprised 35.3 percent of the cities' population, while in the lowest density municipalities, their percentage was 23.2 (source: Statistics Netherlands 1997). Thus, we know that the higher the address density of the municipality, the more young inhabitants there will be, and the higher the prevalence rates for substance use. However, this hypothesis may only be used to explain differences in cannabis and ecstasy rates, but not for cocaine, as there is no significant relationship between cocaine use and age. The hypothesis is easy to verify by looking at the distribution within the age cohort of young adults (for example the group of 18-34 year olds) and testing whether the distinction between strata still exists. In this way, percentages within this cohort neglect the relative numbers of the group 18-34 in each stratum in order to control for age. We consider the age cohort as a constant factor. Looking at Table 7, within given age groups, the relationship between the address density of the municipalities and prevalence rates holds for all drugs. Therefore, we have to reject the assumption that the presence of greater or lesser proportions of younger persons in a given strata explain distinct drug use prevalence differences observed between strata.
The evidence collected here indicates that the use of illegal drugs is closely related to gender, a finding that corroborates the conclusions of previous research (Sandwijk et al. 1995). Within the five strata, sex is equally distributed. In 1997, 48.4 percent of the highest density stratum was male and 51.6 percent female; in the lowest density municipality these percentages were only slightly different: 50.4 and 49.6 percent (source: Statistics Netherlands). The drug use distribution, controlled for gender, shows distinct rates per stratum (see Table 5). Consequently, the variable gender is not to explain the differences in prevalence rates.
A possible explanation for the variations may be found in the very abstract and complex concept of lifestyle. One key variable to test and verify that drug use can be partially explained by lifestyle is the frequency with which respondents visit clubs, discotheques or bars. This concern has already been addressed in previous research among the Amsterdam population and revealed that drug use may be largely determined by lifestyle (Sandwijk et al. 1988). Considering that this behavior correlates to address density and to drug use prevalence, lifestyle might help to explain the discrepancies in drug use prevalence between different address density locales).
Among current cannabis users, 44.8 percent frequently go out to bars, clubs or discos; the comparable figure is 12.5 percent for those who are no last month cannabis users. Similar distributions are found when comparing the activities of cocaine users (45.7 percent of current users go out often, versus 13.8 percent of current non users), as well as ecstasy users (51.9 percent of current users go out often to bars, clubs or discos versus 13.2 percent of the current non users. The logit analysis allows us to more closely examine this relationship. We found that, for all examined drugs, the frequency of visiting bars, clubs or discos is an important explanatory variable. Specifically lifestyle is the best explanatory variable, as opposed to the number of evenings spent out of the house, or an orientation toward "partying" behavior (e.g. going to movies, theatres, restaurants, and also bars, clubs or discos). An exception may be noted regarding cannabis use in the rural municipalities, where age is the most important explanatory variable, followed by an orientation toward an active nightlife. No significant relationship was found between lifestyle differences in low-density municipalities and cocaine or ecstasy use prevalence.
It is safe to say that the urban stratum has a relatively youthful population, and that a large proportion of these young people have active nightlives. Of course it is not unusual to see people in their 30s and 40s in clubs everywhere. For the highest density stratum generally 17.9 percent of the population report going out often; in the most rural municipalities it is only 12.4. A possible explanation for the discrepancies between drug use prevalence rates among various strata could be sought in the answers to the survey question regarding "the frequency of going out to bars, clubs or discos", since this is positively related to the prevalence of drug use. To find out if this is true, we focused on the group of frequent nightclubbers, defined here as persons who went out to clubs, cafes and discos more than 4 times in the month prior to the interview. Even among this active nightlife group there still exists a distinct relation between municipality and drug use prevalence rates. (Table 8) For example, the probability that such a person in a highest density stratum ever used ecstasy (10.4 %) is very different from the probability that a similar person in a rural municipality used it (5.6 %).
In summary, two factors may help to explain the observed relationship between drug use prevalence rates and address density. The first is that a much larger percentage of high-density urban dwellers have active nightlifes than is found among residents of less populous areas. The second is that these high-density urban also have relatively high drug use prevalence rates. These factors show the (expected) correlation between nightlife and high address density. It should be noted, however, that the answer is by no means confirmed yet.
A relatively large share of the background and socio-economic variables do not exhibit a significant relationship to drug use within the logit model. The variables of marital status, income, education and household composition do not help in explaining drug use prevalence. For that reason, they also do not help to explain the distinct relationship that often exists between drug use prevalence and address density.
Based on this review, we may conclude that there is no clear-cut answer to the question raised earlier regarding the significance of different drug use explanatory variables. Consequently, we have to continue to search for new factors that relate to drug use prevalence and address density.
Variation within address density strata
Drug use in the highest address density stratum
We have noted that drug use prevalence rates are not uniform throughout the Netherlands. This raises the related question of whether these rates are similar within a single address density stratum? We now take a closer look at the highest address density stratum that is made up of Amsterdam (n = 3,710), Rotterdam (n = 2,320), The Hague (n = 2,279), Utrecht (n = 2,198) and other high-density municipalities (n = 2,289) (see also Table 1). It should be noted that the city of Amsterdam has the highest average level of address density within this group (Statistics Netherlands (date).
Dutch drug use prevalence and other drug use statistics
The core figures for cannabis, cocaine and ecstasy use in the big cities are presented below in Table 9. The data provides the evidenced that prevalence rates are not equally distributed within a stratum. The following discussion is directed to interpreting and explaining these figures.
Cannabis, cocaine and ecstasy
The Amsterdam figures confirm that that city is indeed a special case and as such is not at all representative of the Netherlands as a whole. Notably, cannabis use is very prevalent in Amsterdam, with 36 percent of the population aged 12 and older reporting lifetime use. There are clear distinctions (significant Chi-square p<0.05) identifiable between the cities in terms of prevalence rates (lifetime as well as current), continuation rates and experienced use rates. These differences are not found for "mean age of first cannabis use", however, as this indicator seems to be equal in all of the Dutch samples.
Similarly, with regard to cocaine, the highest prevalence rates are found in Amsterdam. Cocaine prevalence rates (lifetime as well as current) are comparable for Rotterdam, The Hague and Utrecht, although there are significant distinctions between these rates and those found in Amsterdam. Cocaine continuation rates and experienced use rates are notably different in each city. The mean age of first cocaine use is about the same for all four cities.
When we turn our attention to ecstasy use, its prevalence is highest among the Amsterdam population, buoying the observation that that city's use patterns are clearly different from those in Rotterdam, The Hague and Utrecht. There are clear distinctions between the cities in terms of prevalence rates (lifetime as well as current), continuation rates and experienced use rates. In Amsterdam, users were introduced to this drug later in life than those in the other three cities.
When reviewing these drug use figures, we have observed both similarities and differences between the four large cities. Prevalence rates and use practices there follow distinct patterns, depending on the particular drugs used and the city in question. It is difficult to order the cities in such a way that drug use prevalence rates increase when the classification increases, as they did with address density. Thus, within a given density stratum, it is not simply the higher the address density, the higher the prevalence rates. Although Amsterdam (marked with the highest address density of all) always tops the list. The latter fact reinforces the notion that that city is indeed a special case. In terms of prevalence rates, the cities of Rotterdam, The Hague and Utrecht could be considered as one homogeneous group, and quite distinct from Amsterdam. These comparisons suggest that generalizing whole-population drug use figures distorts and covers over myriad variations that exist between different population elements.
In all four cities the population has a roughly equal composition, with regard to age and gender. The established relationships between age, gender and drug use prevalence are also equivalent in all four cities. For that reason, age and gender may not be seen as explanatory factors. Although the frequency of nightclubbing measure is positively related to drug use prevalence in all of the cities. The proportion of people in Amsterdam who go out often (24.7 %) is higher than in Rotterdam (14.5 %), The Hague (11.4 %) and Utrecht (23.5 %). Not surprisingly, Amsterdam also exhibits the highest drug use prevalence rates. Based on the frequency of nightclubbing behavior, however, Utrecht should show drug use prevalence rates at the same level as those of Amsterdam, but this is not the case. This indicates that the prevalence of nightclubbing is a necessary but not sufficient explanation of the different prevalence rates found among the different cities. Clearly, other variables must be sought that can better explain the clear distinction in prevalence rates that exists between Amsterdam and these other cities.
The samples here all represent big cities and, of course, each city has its own unique characteristics. Amsterdam, Rotterdam and Utrecht are all typically seen as student cities. Given this perception, we might expect them to have younger populations than are found in The Hague, which is the government center. Despite that expectation, the population composition for both age and gender is roughly equal in all four cities. This is not true for the lifestyles of these populations, however. The proportion of people who go out often is significantly higher in Amsterdam (24.7 %) and Utrecht (23.5 %) than it is in Rotterdam (14.5 %) and The Hague (11.4 %). We expect that these persons contribute significantly to the reported higher drug use prevalence that exists in Amsterdam and Utrecht.
All in all, it is not easy to explain the different city-related drug use prevalence rates. This should make us sensitive to the "location-specificity" of drug use figures, even within a single level of address density. One should be aware of this specificity (and therefore possibly "non-representativity"), especially when comparing the development in drug use patterns between countries or cities. Idealiter, one should control the surveys for address density and age.
One possible answer as to why drug use prevalence rates differ among various population concentrations may be sought in the phenomenon of "experimental behavior". There is a "gap" between the range of density strata's drug use prevalence rates and their more narrow dispersion among continuation rates in different strata. These variations suggest that experimental drug using behavior is not the same in every municipality. This observation is confirmed when we look more closely at the experienced use rates in Table 5 and Table 9. The figures indicate that the inhabitants of urban municipalities, and especially in Amsterdam, engage in much more experimental drug use than do those residing in the less populated areas. The percentages of lifetime users who consumed a given drug 25 times or more, are higher among Amsterdam residents than they are in the generic highest density stratum, and these are in turn different from these in the lowest density stratum. In looking at the total picture, however, the experienced use rates reveal a less pronounced relationship than was observed among prevalence rates.
National drug policy can not be seen as an explanatory factor regarding the spread and intensity of drug use in the Netherlands, because it officially applies to all address density municipalities and in each city. While the formal (de jure) policy theoretically applies to all areas, in its implementation (de facto) it can differ significantly. We can easily observe that local policies are often very different from one another. For example, Amsterdam has by far the largest number of coffee shops (297 in 1999, source: Bieleman 1999), whereas there are few or none in quite a number of villages and small towns.
It is also true that persons going out to clubs often do not limit these sojourns to their own home location. It is certainly conceivable that many persons living in the areas adjacent to large cities travel to the urban core for their entertainment and relaxation opportunities.
The paper has first addressed the national and international position of Dutch drug use prevalence rates. Secondly, it focussed on address density as a measure of urban residency. In accomplishing this, we have focused on drug use statistics gathered in several large Dutch cities, as well as those in areas of lower population density. Distinct differences and patterns were found. For example, cannabis use rates are closely tied to population density; the lowest rates were found in low-density communities, and the highest at the other end of the continuum. Despite such general findings, however, we still noted distinctions within specific population strata. Amsterdam's cannabis prevalence rate was 36.7 percent, for example, while that for the other high-density cities was 25.5 percent, and for the low-density municipalities it was 10.5 percent. The author concluded that other critical figures, such as those for continued use, experienced use and mean age of first use, are invariant for the level of address density, and, as a consequence, not necessarily related to prevalence rates. We discovered that there were significant differences between cities sharing a single address density category, however, and therefore this feature can only partially explain prevalence of drug use by address density. A further exploration leads to the third question addressed to in this paper: the impact of lifestyle on drug use. The data provides evidence that lifestyle is related to drug use prevalence rates. We still did not find a sufficient explanation for drug use prevalence. In order to be able to determine why there is a discrepancy between prevalence rates within various address density strata, we have to put more effort into explaining drug use in general.
Abraham, M.D. (1998), Drug use and lifestyle; behind the superficiality of drug use prevalence rates. Presentation held at the 9th Annual Conference on Drug Use and Drug Policy, Palma de Mallorca, October 2, 1998. Amsterdam: CEDRO.
Abraham, M.D., P.D.A. Cohen, and M.A.L. de Winter (1999), Licit and illicit drug use in the Netherlands. Amsterdam: CEDRO.
Bless R., Korf D.J. Riper. H., Diemel S. (1997), Improving the comparability of general population surveys on drug use in the European Union. Final report commissioned by the EMCDDA. Amsterdam: Amsterdam Bureau of Social Research and Statistics.
Bielemal B., Snippe J. (1999), Gedogen gewogen, evaluation of the Amsterdam coffeeshop policy. Stigting Intraval, Groningen: De Bie. (publication in Dutch)
European Monitoring Centre for Drugs and Drug Addiction (1998), Annual report on the state of the drugs problem in the European Union. Luxembourg: Office for official publications of the European communities.
Harrison, L.D., M. Backenheimer, and J.I. Inciardi (1995), Cannabis Use in the United States: Implications for Policy. In P.D.A. Cohen and A. Sas, Cannabisbeleid in Duitsland, Frankrijk en de Verenigde Staten., 179-276. Amsterdam: CEDRO.
Hartnoll, R. (1995), Multi-city study: Drug misuse trends in thirteen European cities. Pompidou Group. Strasbourg: Council of Europe press.
Korf, D.J. (1995), Dutch treat. Amsterdam: Thesis Publishers.
Kuipers, S.B.M., H. Stam, and W.M. de Zwart (1997), Jeugd en riskant gedrag 1996, Roken, drinken, drugsgebruik en gokken in het voortgezet speciaal onderwijs en spijbelopvangprojecten. Utrecht: Nationaal Instituut voor Alcohol en Drugs.
Maddala, G.S. (1983), Limited-dependent and qualitative variables in econometrics. Cambridge: Cambridge University Press.
Planije M.P., Mensink C., Spruit I.P. (1988), Zakboekje kenrcijfers Trimbos-instituut. Utrecht: Nationaal Instituut voor Alcohol en Drugs.
Sandwijk, J.P., Cohen, P.D.A., Musterd, S. (1988) Het gebruik van legale en illegale drugs in Amsterdam. Amsterdam: CEDRO.
Sandwijk, J.P., Cohen, P.D.A., Musterd, S. (1991), Licit and illicit drug use in Amsterdam. Amsterdam: CEDRO.
Sandwijk, J.P., Cohen, P.D.A., Musterd, S, Langemeijer M.P.S. (1995), Licit and illicit drug use in Amsterdam II. Amsterdam: CEDRO.
Statistics Netherlands (1997), CBSView. Voorburg/Heerlen.
Substance Abuse and Mental Health Services Administration (1997), National Household Survey On drug abuse: Population Estimates 1997. Department of health and human services. Rockville, DHHS Publication.
Tweede Kamer 1997 Letter from Dutch Health Minister to second chamber of the parliament regarding 'Progress report drug policy'. 1997-1998, 24077, nr. 57. The Hague. (publication in Dutch)
Last update: May 25, 2016