The presented results suggest that drug use is high among staff at licensed premises in Stockholm. However, staff in 2007/08 reported lower rates of self-reported drug use, both life-time use and more interestingly, last year use, than staff in 2001.
The 2001 sample and 2007/08 sample also reported different rates of behavioral intention and dissimilar attitudes towards drugs. A higher number of the 2007/08 sample responded that drug-intoxicated guests should always be asked to leave licensed premises. In addition, a significantly higher number in 2007/08 reported that they would call the police if they saw someone take drugs at the licensed premise where they worked. Finally, the staff in 2007/08 reported significantly lower rates of observed drug-intoxicated guests, observed drug offers and drug intake.
Comparisons among staff that are last year drug-users and non drug-users verify that drug-using staff observe more drug use among guests, have a more liberal attitude towards drug use, and are less likely to intervene than non drug-using staff. The data suggest that in order to be more effective in reducing drug use among guests at licensed premises, drug use among staff should be targeted as well. Hence, staff at licensed premises are an important population to target in club drug prevention programs.
Staff at licensed premises had a much higher life-time and last year prevalence of drug use as compared to the general population in Sweden. For example, in 2003, 17% of Swedish 16-24 year olds reported ever-use of illicit drugs  compared to 57% of staff in the same age group in our survey 2007/08. Possible explanations for high self-reported drug use among our surveyed staff may include environmental factors such as greater access to drugs at licensed premises than other workplaces, and stressful work shifts with late hours. Another possible explanation could be that individuals with sensation-seeking personality may be more attracted to work in the nightlife setting . However, despite the fact that the study participants reported a high prevalence of drug use the majority supported Sweden's strict drug laws (Table 6).
It was reported that the four most commonly used drugs in 2001 and 2007/08 were cannabis, cocaine, amphetamine, and ecstasy (Table 4). The rates for these four drugs were lower in 2007/08. Specifically, amphetamine had decreased the most and cocaine the least. This is in accordance with reports from the Police Authority and Custom Control Department, showing that the availability of cocaine has increased and prices have decreased in Stockholm . The most popular types of club drugs used can vary in different settings, cultures, and countries. Researchers from other countries report that other types of drugs, for instance ketamine and GHB, are being used as club drugs [35, 36]. The participants of our study reported very low rates of these club drugs. Here we measured staffs' self-reported drug use and their observations of drug use at licensed premises. Other researchers have studied prevalence rates for drug use among guests using self-reports and biological assays. For example, in a Swiss study conducted at dance music events, the rate of attendees ever-use of ecstasy and cocaine was 40% and 36% respectively . Furthermore, researchers in the US found that 25% of the guests in the club setting used illicit drugs .
The objective of this study was to examine self-reported drug use among staff at licensed premises, types of drugs used, attitudes towards drugs and observed drug use among guests, and not to study the effects of the "Clubs against Drugs" program. The nature of a multi-component intervention such as the "Clubs against Drugs" complicates the explanation of these findings. Nevertheless, in the absence of other reasons for the decrease in drug use, we propose that it might be possible that the "Clubs against Drugs" program may have contributed to this result. It should be kept in mind that the intervention program was quite extensive and that no competing activities have transpired with the targeted licensed premises during the intervention period. The intervention strategies were implemented immediately following the first measurement in 2001. The policy work component of the intervention focused the most on preventing club drug use among staff. In 2007, at the time for the second measurement, more than 150 owners and managers had been policy-trained and over 400 doormen had passed the two-day drug-training course.
There are limitations to this study that constitute possible threats to the validity of our findings. As mentioned earlier, the use of a non-randomized cross-sectional design reduces our ability to interpret the results causally. The absence of a control group and national comparison data are further limitations. However, it is important to note that when comparing the two sample groups (Table 1) they are very similar in most demographic characteristics.
Another concern to be addressed is whether or not the reported lower rates of drug use at licensed premises could be explained by displacement of problems. The reduction in reported drug use could partly be the result of staff that use drugs choose to work at licensed premises not involved in the drug prevention work, or in another city. Additionally, guests that use drugs may have chosen to go to other licensed premises. But, it seems unlikely that this is the main explanation for the results, as these licensed premises in downtown Stockholm are very popular workplaces as well as popular for guests to visit. A multi-component prevention program in Australia, also supports the idea that guests choose to stay at their favorite establishment even after the implementation of intervention programs .
There are also some strengths of this study. The response rate was very high, all of the staff attending server training agreed to participate. The highest internal drop-out rate, at both measurements, was approximately 6-7% for questions on self-reported drug use. The same procedures for sampling and data collection were used in 2001 and in 2007/08. Since both measurements were conducted during a one year period, seasonal variations of drug use would not explain the results. Even though random sampling was not possible, we would like to argue that the sample used is the best available representation of staff from licensed premises with late open hours in central Stockholm.
The majority of drug prevention activities are focused on school prevention programs. However, this study emphasizes the importance of also developing prevention programs within other arenas. The nightlife scene is a high-risk setting for club drug use. Previous results published by our research group,  as well as results presented herein, indicate that the intervention might have reduced the rates of drug use at licensed premises in Stockholm.