3 Prevalence of Illicit Drug Use
In this section we describe the data sources we have used to provide information on the number of people in Scotland using illicit substances. We summarise the data sources, then go on to discuss the assumptions associated with basing our estimates of the size of the market and the economic and social costs on these data sources. We describe this section as the 'prevalence model' in that we have constructed what we regard as the most likely scenario as regards to drug use in Scotland, split by age group, type or types of drugs used and severity of use.
In terms of severity of use, we split the population into two broad groups (or compartments). These are problem drug users and recreational drug users. The recreational drug use group only includes individuals aged between 10 and 64. We ignore any drug use by those less than 10 years or older than 64 years. While individuals of such ages may be using drugs, it is assumed that their contribution to the size of the markets or the social and economic costs would be negligible. Following the recommendations of the European Monitoring Centre for Drugs and Drug Addiction, the 2006 Scottish Prevalence Study estimated the prevalence of problem drug use across the 15 to 64 age range therefore for this study problem drug users range from age 15 to age 64 (thus defining all drug use by those under 15 as recreational use).
Clearly by categorising all of Scotland's drug users into two broad groups, we are in danger of ignoring much of the inherent variation amongst individuals' patterns of drug use. What we may have assigned as recreational ( i.e. non-problematic) use for one person may indeed cause them more problems than a similar level of drug use amongst someone we have placed in the problem drug user category. We also try not to attach any value judgements to the two categories, although the categories may have labels that, to some readers, appear judgemental. However in order to provide estimates of the size of the drugs market in Scotland or the social and economic costs we need to recognise that some people use drugs in different ways to others and we need to split the population into at least two different groups. We could attempt to give each group a non-descriptive name, such as Group A and Group B, but such an approach could be confusing.
3.2 Case Definitions
By 'problem drug use' we typically mean illicit drug use that would often result in the users requiring social or medical intervention into their lives. That would typically mean drug treatment, however we note that a substantial number of people who we would class as problem drug users are not currently engaging in treatment. Characteristically, drug users in this compartment would not be in employment and may have criminal justice issues and often other social / medical issues such as conditions relating to the injecting of drugs. We do, however, note that some users of drugs that we would naturally associate with problem drug use, such as heroin, may be using the drug in a less problematic way than typified above.
By 'recreational' drug use we mean illicit drug use, particularly the use of drugs such as cannabis, ecstasy and powder cocaine that would not typically be seen as causing major widespread problems to the users or society. Clearly there may be patterns of use, for example, of cannabis or cocaine that are indeed problematic, however, out of necessity we have classed all use of such drugs as recreational.
In terms of this study our definition of problem drug use has been led by the data sources available in Scotland on problem drug use, particularly the 2006 Scottish Prevalence Study. The Prevalence study defined problem drug use as illicit use of opiates and / or benzodiazepines and this study has used the same case definition. Therefore an individual who illicitly uses any opiate and / or benzodiazepine is defined as a problem drug user. Although we would also consider crack cocaine users to be problem drug users it was not possible to estimate the number of such users within the prevalence study. The way in which this issue was dealt with is discussed in a later section in this chapter.
Following on from this, we define illicit drug users as recreational drug users if their drug use does not include opiates or benzodiazepines.
These categories are not perfect, and may quite rightly be criticised as over-simplifications of Scotland's patterns of drug use. We accept this criticism; however we note that it is a necessary step within the wider analyses to estimate the size of the markets and social and economic costs. We do take steps to account for this over-simplification by including a range of consumption estimates within each group rather than just a point estimate of the amount of drugs used by, for example, recreational cannabis users.
3.3 Data Sources
The previous UK studies, based on English data, have based their estimates of the number of drug users on surveys, in particular surveys of those that have been arrested ( NEW- ADAM in Bramley-Harker (2001) and the Arrestee Survey in Pudney et al (2006)). Both studies then had to base their estimates on the number of drug users on information such as the probability of arrest, the probability that if they had been arrested they were using drugs and the number of arrests.
In this study, we do not follow the approach of the two previous studies for a number of reasons. The first reason is pragmatic; there is no Scottish survey that directly compares to either the NEW- ADAM or Arrestee Survey carried out in England or Wales. There may have been merit in using data collected through the small number of drug courts or arrest referral schemes in Scotland, but this was not thought appropriate because there are too few schemes and the data from such schemes is not believed to be representative of the whole of Scotland. Instead, we based our estimates on the national estimate of the prevalence of problem drug use and the Scottish Crime and Victimisation Survey ( SCVS) and the Scottish Schools Adolescent Lifestyle and Substance Use Survey ( SALSUS), which are two major studies commissioned by the Scottish Government to estimate the prevalence of drug use within the general population.
As with Pudney's approach we do attempt to cross-validate the results of these surveys with other available data, however we make no attempt at correcting for under-reporting or non-response. The reason for this is primarily because making such adjustments would have been out with the scope of this study. In addition we are not convinced that there is any significant under-reporting within the SCVS or SALSUS, at least for cannabis, amphetamine, ecstasy or cocaine use. In fact, there is a risk with schools surveys that respondents may exaggerate their drug use. In an attempt to detect this, SALSUS includes a fake drug however it is not clear how they deal with respondents reported to use this fake drug. It is easy to speculate that there could be under-reporting, however, there is no real evidence to back up that assertion. Studies have been done elsewhere that show discrepancies between self-reported drug use and other measures, for example saliva or urine testing (Fendrich et al, 2004; Harrison and Hughes, 1997), however, without carrying out these comparisons on the specific Scottish studies we are using data from, it would be wrong to conclude that there is actually any under-reporting. This could, of course, result in bias within our estimates, but we would prefer to state that we have made this assumption rather than try to account for such potential bias.
In terms of non-response bias, this could also influence the prevalence estimates from both SALSUS and SCVS. In relation to SALSUS, this would occur if those absent from mainstream schooling (for example those off sick, truanting or in alternatives to mainstream education such as residential schools or secure units, or indeed those in private education) had different patterns of drug use than those captured within the SALSUS sampling. Undoubtedly this could occur, but again without firm evidence as to the effect of such non-response we would prefer to acknowledge it in our assumptions rather than try to account for it. There may also be non-response bias in the SCVS. It is difficult to account for the perhaps significant non-response bias when considering heroin or crack cocaine users who will be in accommodation that is not entirely amenable to the typical household survey (including homeless accommodation, temporary accommodation or prison). We have therefore quantified the use of such drugs from means other than surveys, and have assumed that no problem drug users will be included in the SCVS sample. For the use of other drugs (such as cannabis and cocaine) there may be a degree of non-response bias, but again attempting to account for this would have been outwith the scope of this study.
Our definition of problem drug user is led by the available data, in this case, the national prevalence study (Hay et al, 2009) and DORIS, the Drug Outcomes Research in Scotland study 1. We have also obtained information from the Scottish Drug Misuse Database 2, but we have found that it was not particularly useful for feeding into the prevalence model for this study for a number of reasons that will be discussed in a later section.
We are using the national prevalence study and DORIS to estimate the use of all the different types of drugs used by problem drug users. So in effect, we are assuming that problem drug users (such as those that use heroin or crack cocaine) would be unlikely to be participating in household surveys such as the Scottish Crime and Victimisation Survey. We have therefore estimated the number of problem drug users using cannabis (and thus the amounts they use) from the national prevalence study and DORIS rather than assuming they use the same amount found Scotland-wide from the SCVS.
3.4 Problem Drug Users
To estimate the number of problem drug users who use different drugs we combine the results from the 2006 national prevalence study with information on the drugs used by problem drug users surveyed in the DORIS study. We first need to derive a total number of problem drug users, and define explicitly what we mean by problem drug use in this setting.
The first step is to derive an estimate of the number of opiate and/or benzodiazepine users in Scotland. The number of opiate and/or benzodiazepine users in 2006 was estimated to be 55,328 (Hay et al, 2009). For the purposes of this study we will refer to this population of drug users as problem drug users or PDUs.
To estimate the number of problem drug users who are using different drugs two different sources could be used; the Scottish Drug Misuse Database and DORIS. The Scottish Drug Misuse Database contains records of all new clients in treatment in Scotland for any given year and is currently moving towards recording information on all clients in treatment through the new SMR25 form. This form, amongst other things, asks patients about their current illicit drug use allowing up to six different drugs to be recorded along with the amount being used and the frequency of their use. The advantage of using the Scottish Drug Misuse Database is that their data is routinely available, it is up to date and has a large sample size compared to DORIS. Despite this, however, the nature of the database is not set up to record reliable information on secondary drug use as is shown below.
The DORIS study, on the other hand, is a cohort study following an initial sample of 1,030 drug users entering treatment in 2002 3. The cohort was followed up after 3 time periods; after 8 months, after 16 months and after 33 months. There are a number of advantages and limitations to the DORIS study that must be noted. Firstly, DORIS was set up to sample from a cross section of the problem drug using population in Scotland. It asked respondents in-depth questions on a wide range of areas relating to their life, from their illicit drug using profile to their uptake in health services and criminal history. Given that this study was for research purposes only it is felt that respondents may provide less biased answers than they would when answering questions for the SMR24/25 form when entering treatment. Even though this study only included individuals who had been in treatment at least once (in other words it excluded drug users who had never been in treatment), many of the questions asked in the first DORIS are about life before treatment. Therefore it is felt that using DORIS provides an adequate sample of problem drug users at all stages of their drug using careers.
However, as with any questionnaire that asks about behaviour in the last six or eight months there may be issues with recall. Further, there is a risk that participants may misunderstand certain questions either due to differences in the use of language by the interviewer and participant or due to the questions being too vague. Given the illegal status of drug use, DORIS participants are a covert population and so the sample size is limited at 1,030. It must also be noted that Aberdeen drug users are not well represented in this study due to the largest drug treatment agency not wishing to take part in the study.
The DORIS study included 1,001 problem drug users, identified as either using any opiates or benzodiazepine illicitly, or receiving substitute treatment for their opiate or benzodiazepine use, or had been identified via prison treatment. Out of the 1,001 problem drug users, 906 reported that they were using heroin. Thus we assume that 90.50% of problem drug users use heroin. If we apply this proportion to the estimated number of problem users (55,328) we estimate that there are 50,077 heroin users in Scotland.
Crack Cocaine Use
The approach described above needs to be adapted for crack cocaine users as there may be crack cocaine users who are not also opiate users. In this study we include crack cocaine use in our definition of problem drug use, such that, anyone using opiate or crack cocaine would be defined as a problem drug user. We can therefore either assume that there are virtually no individuals using crack cocaine who have not used an opiate, or we can try to estimate, from DORIS, the size of the problem user population that is using crack cocaine without using opiates.
We find from DORIS that there are only three people using crack cocaine but not using opiates. That could, however, be for a number of reasons. In particular, it could be because DORIS was designed to follow individuals from their first treatment episode and in Scotland the vast majority of treatment services are often accused of being opiate-centric. It could also be because the first sweep of DORIS is quite dated, since it was carried out in 2002 when crack cocaine use may not have been so prevalent. Unfortunately the 2006 national prevalence study did not find sufficient information to derive valid crack cocaine prevalence estimates. Although there are probably quite a few reasons why there are extremely low levels of crack cocaine use (without concurrent opiate use) in the DORIS sample, it could actually be possible that there are only a small amount of people in Scotland using crack cocaine who are not also using opiates. This is perhaps confirmed when an extract of the Scottish Drug Misuse Database is examined. Out of more than 9,000 individuals who are identified as using any opiate or crack cocaine, only 46 say they are using crack cocaine but not any opiate.
We therefore assume that the estimate of 55,328 opiate users is also a valid estimate of the number of opiate and / or crack cocaine users (or problem drug users) following the definition employed within this study. We can therefore estimate the number of problem drug users using crack cocaine in the same way as we estimated the number of opiate users using heroin. In total, 284 out of the 1,001 problem drug users stated that they were using crack cocaine. Thus we estimate that there are 15,697 problem drug users who are crack cocaine users.
We estimate the number of problem drug users who use amphetamines, benzodiazepines, cocaine (powder), ecstasy, and methadone (used illicitly) in the same way, by applying the proportions found in the DORIS problem drug using sample to the estimated total number of problem drug users. Table 3.4.1 summarises these estimates. As with all tables in this report, figures such as the number of days or the estimated prevalence of drug users are presented as whole numbers. All other figures are rounded to two decimal places. This can give the appearance of spurious accuracy, and hence should be interpreted with caution.
Table 3.4.1 Estimated number of users (who are categorised as problem drug users, by type of drug) using DORIS (n = 1,001)
PDUs in DORIS using drug (%)
There are some points worth noting from that table. In terms of powder cocaine, the 14,813 estimate describes only the number of 'problem users' who use cocaine powder. There will be other cocaine users in the recreational category. The same can be said for amphetamines, ecstasy, cannabis and benzodiazepines (although there might be some merit in assuming that benzodiazepine use is only by problem users). As we are assuming that all heroin, illicit methadone and crack cocaine use is done by problem users, the figures in the table provide what, for the moment, are our final estimates of the number of people in Scotland using those drugs. One final point to note is that the majority of problem drug users in DORIS are polydrug users. Therefore the estimates given above for each individual drug cannot be summed.
As previously mentioned, information from the Scottish Drug Misuse Database could be used to estimate how many people use crack cocaine but do not use opiates. We can also attempt to use the Scottish Drug Misuse Database data in the same manner as we used the DORIS data to try to estimate the numbers of problem users using different drugs - see table 3.4.2 below.
Table 3.4.2 Estimated number of users (who are categorised as problem drug users, by type of drug) using SMR25 data (n = 9,624).
PDUs in SMR25 using drug (%)
The two sets of figures are compared graphically in Figure 3.4.1 below:
Figure 3.4.1 Estimated number of users (who are categorised as problem drug users, by type of drug) using DORIS and SMR25 data.
Apart from heroin, there appears to be marked differences between the estimated numbers of problem users using specific drugs when deriving the estimates from the Scottish Drug Misuse Database data. One reason for the difference would be that the Scottish Drug Misuse Database asks about drugs used in the last 30 days whereas DORIS uses 90 days. However we suspect that much of the difference is because DORIS more accurately records secondary drug use than the Scottish Drug Misuse Database, either through more specific questioning about all illicit substances used or because an individual drug user would have less reason for not telling a DORIS interviewer about all aspects of their drug use. This is perhaps more acutely seen when looking at ecstasy, where almost ten times as many opiate users surveyed in DORIS admitted using that drug compared to within SMR25. Because of this, we intend using only the estimates derived from the DORIS data within this study.
3.5 Recreational Drug Use
We can estimate the number of drug users who are not classed as problem drug users from existing surveys.
We can use information from the Scottish Adolescent Lifestyle and Substance Use Survey ( SALSUS), which is augmented for the youngest individuals by information from a study on pre-teen drug use (McIntosh et al). For adults we use the Scottish Crime and Victimisation Survey. Although there is a slight overlap between the age groups of the two studies, with 16 year olds in both the SALSUS and SCVS studies, we have used the SCVS to estimate the levels of drug use by 16 year olds, primarily because the 16 year olds in SALSUS are possibly either quite old for their peer group within school (and thus appearing quite young for a 16 year old) or have been held back a year at some point. Therefore they may not be representative of 16 year olds in Scotland.
To reiterate, we assume that all heroin, crack cocaine or methadone use is by problem drug users and thus negligible within the non-problem use group.
We based our estimates of the number of adult recreational users using particular drugs on the Scottish Crime and Victimisation Survey. As previously noted, we do not make any attempt to correct for any possible bias introduced by either non-response or under-reporting. We have assumed that all heroin, crack cocaine and methadone use occurring in this 25 to 64 year old age group to come under the definition of problem drug use. As for other drugs, we simply apply the percentage used in the last year (for each individual age) to the population of Scotland who are that age to estimate the total number of people that age who are using a particular drug.
A similar approach is taken with the data from SALSUS and the preteen study to provide estimates of the number of people aged 10 to 15 using specific drugs. These estimates are used in conjunction with the data for 16 to 24 year olds to provide estimates of the number of younger people in the recreational group. Table 3.5.1 presents our estimates of the number of recreational drug users, obtained by combining the results of the preteen, SALSUS and SCVS studies. As with the problem drug use estimates, the estimates given below cannot be summed due to some individuals using more than one drug.
Table 3.5.1 Estimated number of recreational users (used in last year)
10 to 24
25 to 64
We have used these estimates relating to the last year within this analysis, primarily as we are interested in estimating the size of the market over a one year period, and also because questions relating to the frequency of use are framed in terms of the amount of times the drug is used over the past year.
We can summarise this information into a single table that presents the estimates of the number of drug user (either problem drug users or non-problem drug users).
Table 3.6.1 Total Number of Users
The previous sections have presented the prevalence estimates we intend using to inform both the analyses to estimate the size of the illicit drugs market in Scotland and the estimates of the social and economic costs. There are some assumptions associated with these estimates. One main assumption is that there is a negligible number of people who use crack cocaine but who do not use opiates. This may not be the case, but we have not found any data with which we could attempt to quantify this group of problem drug users.
We have taken the estimates from SALSUS and the SCVS at face value in that we have made no formal correction for potential biases introduced by either under-reporting or non-response. It could, however, be argued that using the national prevalence estimate in conjunction with DORIS would correct (at least in part) for under-reporting and non-response, particularly for heroin, crack cocaine, methadone and powder cocaine. There could be under-reporting in SALSUS and SCVS that is not accounted for. We have, however, taken the decision that it is not within the scope of this study to alter the results of these studies without some hard evidence that the results are affected by such biases (and if they were, then it would be up to the study teams / commissioners to deal with that problem). Thus we have not followed some of what Pudney suggests as methodological improvements, although to be fair, the issue of non-response in arrestee surveys is much greater and is corrected for within the source publication (by Pudney himself).
There is also the issue (perhaps more pertinent) that we could possibly be double counting people if the SCVS did adequately pick up problem users in their sample, particularly in terms of benzodiazepine and cocaine powder use.
We have also ignored an issue that Bramley-Harker explicitly tries to account for, that is differing patterns of drug use within prison. Pudney perhaps tries to account for this by splitting the population into household-based and non-household based groups, but does not explicitly model prison drug use in a separate category. We feel that the nature of DORIS (which includes a prison sample) and our interpretation of our national prevalence estimates, coupled with the typical pattern of heroin users serving relatively short sentences makes our approach robust to not specifically attempting to quantify the size of the market within prison. However, we have no firm evidence to back up that assertion.