Consumption Patterns of Nightlife Attendees in Munich: A Latent-Class Analysis

ABSTRACT

Background: The affinity for substance use among patrons of nightclubs has been well established. With novel psychoactive substances (NPS) quickly emerging on the European drug market, trends, and patterns of use are potentially changing. Objectives: (1) The detection of subgroups of consumers in the electronic dance music scene of a major German metropolitan city, (2) describing the consumption patterns of these subgroups, (3) exploring the prevalence and type of NPS consumption in this population at nightlife events in Munich. Methods: A total of 1571 patrons answered questions regarding their own substance use and the emergence of NPS as well as their experience with these substances. A latent class analysis was employed to detect consumption patterns within the sample. Results: A four class model was determined reflecting different consumption patterns: the conservative class (34.9%) whose substance was limited to cannabis; the traditional class (36.6%) which especially consumed traditional club drugs; the psychedelic class (17.5%) which, in addition to traditional club drugs also consumed psychedelic drugs; and an unselective class (10.9%) which displayed the greatest likelihood of consumption of all assessed drugs. “Smoking mixtures” and methylone were the new substances mentioned most often, the number of substances mentioned differed between latent classes. Conclusion: Specific strategies are needed to reduce harm in those displaying the riskiest substance use. Although NPS use is still a fringe phenomenon its prevalence is greater in this subpopulation than in the general population, especially among users in the high-risk unselective class.

KEYWORDS:

• Substance use
• club drugs
• novel psychoactive substances
• NPS
• latent class analysis

With the growing popularity of rave and techno music and the ensuing resurfacing of ecstasy in the late 1990s and early 2000s came a great number of studies investigating the association of psychostimulant use and nightlife attendance. Results have been consistent over time and countries showing that illicit substance use is much more prevalent in nightlife attendees than within the general population (Künzel, Kröger, Bühringer, Tauscher, & Walden, 1997; Miller et al., 2015; Tossmann, Boldt, Tensil, & Tossmann, 2001). This was also verified in a 1997 study on the techno-scene in Bavaria (Künzel et al., 1997) with 12-month prevalence rates of ecstasy at 45.1% and amphetamines at 38.1% (Künzel et al., 1997) compared to rates of ecstasy at 0.8% and amphetamines at 0.4% in a general German population study (Kraus & Bauernfeind, 1997). This gave rise to a number of prevention projects aiming at reducing the associated harms and risks among users in the nightlife by providing information on safer use and encouraging reflection on ones' substance use rather than trying to prevent it altogether.

Although this subculture shows an affinity for illicit substance use, the population and their consumption patterns are far from homogeneous. A latent class analysis (LCA) on the 30-day substance use of MDMA/ecstasy users in the United States of America (USA) not only found three distinct patterns of substance use concerning the range of substances used but also found that these patterns related to the number of occasions on which MDMA had been consumed (Carlson, Wang, Falck, & Siegal, 2005). Accordingly, consumption of a limited range of substances was associated with less lifetime occasions of MDMA use than consumption of a moderate and wide range of substances. A more recent study observed three different consumption patterns (wide range, mainstream, primary cocaine users) in a sample of clubbers within New York City's club scene (Ramo, Grov, Delucchi, Kelly, & Parsons, 2010). The authors found that not only did class membership coincide with substance use, but was also linked to demographic variables such as education, sex, and sexual orientation, as well as psychosocial variables such as substance dependence and sensation seeking behavior. Fernández-Calderón and colleagues (2011) conducted a survey concerning substance use in Spanish underground raves. A cluster analysis based on the participants' drug use at their most recent rave attendance revealed two distinct groups of drug users that differed in the types of substances consumed and the number of substances consumed simultaneously.

Recent attempts have been made to quantitatively rank and compare the harms and risks of various licit and illicit psychoactive substances (Nutt, King, & Phillips, 2010; Nutt, King, Saulsbury, & Blakemore, 2007; van Amsterdam, Nutt, & Phillips, 2015; van Amsterdam, Opperhuizen, Koeter, & van den Brink, 2010) as well as their benefits (Reynaud, Luquiens, Aubin, Talon, & Bourgain, 2013). The overall rating scores varied vastly among substances demonstrating the diversity in harms associated with dissimilar substances.

Individual physical harms following the ingestion of MDMA can include tachycardia and hyperthermia as well as compulsive behaviors, and lockjaw (Gowing, Ali, Irvine, & Ali, 2002; Meyer, 2013; Rogers et al., 2009). Other popular amphetamines can bear the side-effects of loss of appetite, sleep disturbance (Kirkpatrick et al., 2012), increased sexual activity, increasing the risk of sexually transmitted infections (Darke, Kaye, Ketin, & Duflou, 2008) and serious cardiovascular disturbances (Kaye, Mcketin, Duflou, & Darke, 2007), whereas psychedelic substances, such as LSD and psychoactive, or magic mushrooms may induce bad trips and in rare severe cases may lead to temporary psychosis (Nichols, 2004). The occurrence of these harms, however, depend heavily on the individual user, the setting in which the substances are taken, the purity and dosage of the ingested substance and the combination of substances consumed.

This demonstrates the need to address the link between individual harms with the specific substances associated with them. Acknowledgment that different substance use patterns can evoke different risks and harms and the implications this may have on the aforementioned prevention strategies that are targeted at the club scene is of great importance to reduce the overall harms psychoactive substances cause and further tailor measures to reach high-risk users. Consequently, when referring to substance use in the nightlife setting and the risks associated with it the nuances and distinct consumption patterns that occur within this subgroup must be taken into account.

The emergence of novel psychoactive substances (NPS) has added a further aspect to this subculture. Within the past years, new substances have been appearing that seem to imitate the effects of known and established substances such as amphetamines and ecstasy/MDMA. While between the years 1997 and 2009, 110 new substances in total were registered by the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA, 2010), 101 new substances were registered in total in the year 2014 (EMCDDA, 2015).

The use of these substances is vanishingly rare in general population studies (Champion, Teesson, & Newton, 2015; Martinotti et al., 2015; Palamar, Martins, Su, & Ompad, 2015) and thus knowledge of consumption patterns is scarce. According to the Flash Eurobarometer 401 (European Commission, 2014), nearly two-thirds of participants that had used a NPS within the past 12 months had used it in the context of parties. Given the established link between psychostimulant use and the nightlife, the fact that these new substances often imitate established psychostimulants, looking toward the nightlife setting for more information on the patterns of NPS use seems obvious. One study, conducted in nightclubs in Rome, reported a 78% lifetime prevalence rate of NPS and “club drug” use among their study sample (Vento et al., 2014). Although NPS use has also been known to occur in other subcultures (Deluca et al., 2012) the study by Vento and colleagues (2014) suggests that the nightlife population should be strongly considered when investigating the use of NPS.

With the present study we aimed to: (1) detect subgroups of consumers based on 12-month consumption rates of various illicit substances in the electronic dance music scene of a major German metropolitan city, (2) describe the consumption patterns of these subgroups using information on the 30-day frequency of use, the combinations of illicit substances and those substances with alcohol and pharmaceuticals and (3) explore the prevalence and type of NPS consumption in this population.

Methods

Procedure

The study was conducted with the help of peer-to-peer prevention projects active at electronic dance events. Trained peers provide safer use information to patrons aimed at encouraging sensible drug use. The events were pre-selected as those with great likelihood of illicit drug-use among participants. As the main goal of these projects is to engage patrons in a dialogue about the risks and harms of drug use and to educate about strategies to reduce these risks, booths with information on various substances, fruit, earplugs and the like are set up within the venue. Interested patrons approach the booth in order to obtain information, fruit, or engage the peers of the prevention project in a dialogue. During field-work of the study, peers would invite patrons to participate in the survey. The questionnaires were filled out on location.

Sample

This study used a convenience sample. Questionnaires were completed and returned by 1849 patrons of electronic music events in Munich, Münster, and Erfurt. To control for regional fluctuations, only questionnaires from participants recruited in (and surrounding) Munich were used, n = 1741. Of these 1741 questionnaires, 25 were excluded from the analysis for claiming the consumption of “Relevin”, a non-existing drug used in surveys to discern overreporting of drug-use (Hibell et al., 2011), 68 were excluded for not having stated an age and/or sex, and an additional 77 cases of repeat participation were also excluded. The final analytical sample consisted of n = 1571 at 37 electronic dance events.

Material

A pen and paper questionnaire was employed. Apart from age and sex, 12-month prevalence rates were assessed by asking participants whether or not the following substances had been used within the past 12 months: cannabis, ecstasy, speed, cocaine, heroin, crystal methamphetamine, LSD, magic mushrooms, ketamine, GHB/GBL, research chemicals, bath salts, “smoking mixtures” (meaning a product designed to be inhaled or smoked, containing synthetic cannabinoids), medicines and natural drugs (e.g., salvia divinorum, kratom, ayahuasca). Furthermore, participants were asked how often they had consumed aforementioned substances in the past 30-days given five answer-options: never, 1–4 times, 5–10 times, more than 20 times and daily. Participants were also asked to specify the research chemical, bath salt, “smoking mixtures,” medicine or natural drug they had consumed. For the purpose of the analysis, the 12-month prevalence rates for research chemicals, bath salts and “smoking mixtures” were combined to indicate the 12-month prevalence rates of NPS consumption. They were then asked if they had ever combined two or more substances at one time (lifetime prevalence of polysubstance use), and if so, if they had ever consumed these specific combinations of substances: cannabis/ecstasy, ecstasy/speed, cannabis/cocaine, cannabis/speed, ecstasy/cocaine, ecstasy/LSD, cannabis/magic mushrooms, speed/cocaine, alcohol and illegal drugs, medicines and alcohol or illegal drugs, or other combinations. An open-ended question was posed in which participants were to name any NPS they had consumed. The identification and definition of the specific NPS named was then corroborated by Google search and classified using online forums (such as Erowid, Land der Träume) as well as the European Database on New Drugs (EDND). The differing groups of NPS are based on the underlying molecular structure of the substance. Although only assumptions can be made about the psychoactive properties of the substances based on this taxonomy, it is, at the time of writing the only known com-prehensive classification of NPS. The groups NPS were categorized into were: aminoindanes, arylalkylamines, arylcyclohexylamines, benzodiazepines, cannabinoids, cathinones, indolalkylamines, others, phenethylamines, piperazine derivaties, piperidines and pyrrolidines.

Analysis

Complex latent class analyses (LCA) were conducted to identify differing subgroups of patrons based on their 12-month prevalence of the substances described earlier. The analyses were conducted using MPlus 6.12 (Muthén & Muthén, 1998–2010). LCA for 1–7 classes were performed and fit indices were compared. Multiple starts were employed in order to avoid local maxima. Lo-Mendell-Rubin Adjusted Likelihood Ratio Tests (LMR-LRT) were employed in each analysis. The test establishes when a model should be rejected based on the number of classes, i.e., once the number of classes which fits the model best is achieved, each model with additional classes will be rejected and the test will fail to reach significance. In addition, low Bayesian Information Criterion (BIC), Akaike's Information Criterion (AIC), Sample-Size Adjusted BIC (ABIC) values were considered as best model fit (Yang, 2006). Entropy values were also taken into account when choosing a model. These range between 0 and 1, indicating the clarity of classification (Ramaswamy, DeSarbo, Reibstein, & Robinson, 1993), i.e., the closer the value is to 1, the clearer the classification. All indices were compared with the theoretical relevance of the class patterns, when choosing a model. For the final model, conditional response probabilities were used to determine participants' most likely class membership.

In order to test the relationship between class membership and 30-day consumption of substances, an omnibus one-way analysis of variance (ANOVA) was employed with subsequent linear regressions in order to test for between group differences. The predictor variable was class membership and a numeric criterion variable was created using the midpoint of the range of each 30-day frequency answer category. Participants that had taken part at the three first events (n = 40) were excluded from this analysis, as 30-day frequency had not been enquired.

In addition, Chi-square-statistics were applied in order to test the relationship between class membership and binary consumption indicators, i.e., lifetime prevalence of the consumption of two or more combined substances and the combination of substances (described above). Pairwise comparisons of the groups were carried out with logistic regression analyses.

The number of comparisons in this study calls for an adjustment of the alpha-level. As a conservative estimation might jeopardize the explorative nature of the study, p-values of 0.001 and lower were interpreted to be statistically significant. The analyses following the LCA analysis were conducted using Stata 12.1 software (StataCorp, 2011).

Results

Descriptive results

The sample's average age was 23.1 years (SD = 4.7) and it consisted of 38.4% female participants. Table 1 shows the 12-month prevalence rates for all assessed substances in the entire sample. In terms of 30-day substance use, cannabis was the substance most frequently used on a daily basis, claimed by 23.7% (n = 1.112) participants to answer the question. The average amount of days cannabis had been used within the previous 30 days was 12.2 (SD = 11.96). No other substance was taken more than an average of 4.5 days in the past 30-days (SD = 6.7). “Smoking mixtures” were the second most commonly used substance on a daily basis (1.1%, n = 558) with an average of one day of consumption within the past 30 days (SD = 3.8).

Table 1. 12-month prevalence (%) and 30-day frequency rate of substance use in total sample.

Substances	12-month prevalence	30-day frequency
	(%)	M (SD)
Cannabis	80.7	12.2 (12.0)
Ecstasy	63.5	3.8 (5.6)
Speed	59.5	4.5 (6.7)
Cocaine	40.3	2.3 (4.9)
LSD	34.8	1.8 (4.0)
Magic mushrooms	26.9	1.2 (3.1)
Ketamine	21.2	1.2 (3.6)
NPS^a	20.0	−.
Medicines	13.2	1.2 (4.0)
Crystal methamphetamine	12.2	0.9 (3.4)
GHB	10.6	0.7 (2.8)
Natural drugs	7.9	0.5 (2.3)
Heroin	3.4	0.2 (1.8)

^a Based on the accumulation of rates for bath salts, smoking mixtures and research chemicals. Frequency rate not computable.

Model selection

Table 2 shows the results of the fit indices of the analyses for one to seven classes. Comparing the four and five-class model, it was apparent that the fifth class was not a qualitatively distinct class, rather than displaying the identical response pattern at a greater conditional response probability. Based on these fit indices and theoretical considerations as well as opting for the most parsimonious model, a four-class model was chosen for further analyses.

Table 2. Model fit indices for LCA of 1–7 classes.

Model	AIC	BIC	ABIC	Entropy	LRT
1 class	16253.75	16323.42	16282.13	—	—
2 classes	12807.95	12952.66	12866.89	0.80	0.000^*
3 classes	12108.63	12328.37	12198.12	0.75	0.000^*
4 classes	12009.88	12304.65	12129.92	0.74	0.018^*
5 classes	11934.40	12304.21	12085.01	0.71	0.047^*
6 classes	11882.01	12326.84	12063.17	0.69	0.062
7 classes	11867.28	12387.15	12079.00	0.71	0.062

^* p < 0.05.

The four classes were labeled on the basis of their conditional response probabilities. Figure 1 illustrates the consumption patterns of the four identified classes. The first class, 34.9% of the entire sample, was labeled the conservative class. It was characterized by a low probability of having engaged in any substance use apart from cannabis in the previous 12 months. The second class, 36.6% of the sample, was labeled the traditional class and was characterized by the use of “traditional” club drugs such as cannabis, speed, ecstasy and LSD. The third class, 17.5% of the sample, was labeled the psychedelic class due to a greater probability of the use of “traditional” club drugs and additionally a higher probability of the use of psychedelic drugs such as LSD, ketamine, and magic mushrooms. Lastly, the fourth and smallest class, with 10.9% of the sample, was labeled the unselective class and displays higher conditional response probabilities on the use of all substances enquired, except for LSD.

Figure 1. Estimated conditional response probabilities for 12-month substance use per class.

Class differences

Class membership was not related to age, F (3, 1567) = 3.63, p = 0.013, but was related to sex as well as all consumption indicators. Consumption patterns and their relation to class membership are described in Table 3. Table 4 summarizes pairwise comparisons between classes based on regression coefficients (b) and odds ratios (OR). For example, participants of the traditional class consumed cannabis on 10.41 more days than those in the conservative class. Moreover, the odds of a member of the traditional class to have consumed alcohol and an illicit drug were 1.16 (2.16 − 1.0) times the odds of a member of the conservative class (or 16% higher), but 0.11 (0.89 – 1) times the odds of having mixed alcohol or an illicit drug with medicines. The conservative class had a significantly larger proportion of females compared to the traditional class and the unselective class. In terms of 30-day frequency rates, the conservative class consumed cannabis, ecstasy, speed, cocaine, and magic mushrooms significantly less frequently than all other classes. The unselective class consumed all substances at a significantly higher rate than all other classes, apart from cannabis, cocaine, crystal methamphetamine, and ketamine. In addition, the psychedelic class consumed cocaine, crystal methamphetamine, and ketamine significantly more often than the traditional class.

Table 3. Age, sex, and substance use patterns by class, including one-way ANOVA and chi-square statistics.

Classes	Conservative (n = 549)	Traditional (n = 575)	Psychedelic (n = 275)	Unselective (n = 172)	Model statistic
Age, M (SD)	23.4 (4.8)	22.7 (4.8)	23.4 (4.6)	22.4 (4.2)	F(3, 1570) = 3.62
Sex, % female	45.4	34.6	36.4	32.0	χ^2 (3, N = 1571) = 18.21)^*
30-day frequency rates, M (SD)
Cannabis	4.5 (7.5)	14.9 (12.0)	15.0 (11.7)	18.2 (12.0)	F(3, 1111) = 82.79)^*
Ecstasy	0.1 (0.4)	4.2 (5.0)	5.2 (5.9)	8.4 (7.9)	F(3, 977) = 85.29)^*
Speed	0.2 (0.9)	5.0 (6.3)	6.4 (7.2)	9.6 (8.7)	F(3, 909) = 80.14)^*
Cocaine	0.3 (2.0)	1.9 (3.3)	5.1 (7.1)	4.9 (6.4)	F(3, 720) = 46.30)^*
Heroin	0.0 (0.0)	0.0 (0.3)	0.3 (1.2)	2.4 (5.9)	F(3, 485) = 23.82)^*
Crystal methamphetamine	0.0 (0.0)	0.6 (2.5)	2.2 (4.8)	4.4 (7.0)	F(3, 533) = 33.71)^*
LSD	0.0 (0.3)	1.7 (3.6)	2.9 (3.7)	5.1 (6.8)	F(3, 713) = 50.34)^*
Magic mushrooms	0.1 (0.4)	1.3 (3.1)	1.9 (1.9)	3.8 (6.3)	F(3, 626) = 35.66)^*
Ketamine	0.0 (0.0)	0.8 (3.0)	2.6 (3.7)	4.0 (6.6)	F(3, 594) = 38.47)^*
GHB	0.0 (0.3)	0.6 (3.0)	1.2 (1.8)	3.6 (6.4)	F(3, 507) = 25.68)^*
Research chemicals	0.2 (2.0)	1.5 (4.4)	0.6 (2.5)	5.1 (7.2)	F(3, 581) = 33.29)^*
Bath salts	0.1 (0.6)	0.3 (0.9)	0.0 (0.3)	3.5 (6.6)	F(3, 514) = 36.23)^*
“Smoking mixtures”	0.1 (0.4)	0.9 (3.3)	0.1 (0.6)	5.0 (8.1)	F(3, 557) = 39.54)^*
Medicines	0.3 (2.2)	0.9 (2.9)	0.2 (1.0)	5.6 (7.7)	F(3, 566) = 45.86)^*
Natural drugs	0.1 (0.5)	0.7 (3.1)	0.0 (0.0)	2.7 (4.5)	F(3, 510) = 18.86)^*
Combination of substances, %
Lifetime prevalence	29.8	77.7	87.8	86—0	χ^2 (3, N = 1571) = 402.28)^*
Cannabis + ecstasy	6.9	42.1	42.6	57.6	χ^2 (3, N = 1571) = 252.93)^*
Ecstasy + speed	4.0	38.4	42.2	59.9	χ^2 (3, N = 1571) = 291.76)^*
Cannabis + cocaine	3.3	9.4	17.5	27.9	χ^2 (3, N = 1571) = 99.15)^*
Cannabis + speed	3.5	27.5	33.8	47.7	χ^2 (3, N = 1571) = 99.15)^*
Ecstasy + cocaine	1.8	7.0	13.8	23.8	χ^2 (3, N = 1571) = 98.13)^*
Ecstasy + LSD	1.5	5.7	19.6	28.5	χ^2 (3, N = 1571) = 160.63)^*
Cannabis + magic mushrooms	1.8	8.0	15.6	22.7	χ^2 (3, N = 1571) = 91.19)^*
Speed + cocaine	2.2	6.4	16.4	23.3	χ^2 (3, N = 1571) = 100.99)^*
Alcohol + illicit drug	26.7	44.1	46.9	48.8	χ^2 (3, N = 1571) = 48.36)^*
Medicines + alcohol/illicit drug	6.9	6.2	6.5	18.5	χ^2 (3, N = 1571) = 28.76)^*

^* p < 0.001.

Table 4. Pairwise comparisons of demographic variables and consumption indicators using linear and logistic regression.

	Traditional vs. conservative^a	Psychadelic vs. conservative^a	Unselective vs. conservative^a	Psychedelic vs. traditional^a	Unselective vs. traditional^a	Unselective vs. psychedelic^a
Age, b (CI)	−.67 (−1.24 – −0.14)	0.03 (−0.65 – 0.71)	−0.98 (−1.78 – −0.17)	0.72 (0.04 – 1.39)	−0.29 (−1.09 – 0.51)	−1.01 (−1.90 – −0.11)
Sex, female, OR (CI)	0.64 (0.50 – 0.81)^*	0.69 (0.51 – 0.93)	0.57 (0.39 – 0.81)	1.08 (0.80 – 1.46)	0.89 (0.62 – 1.28)	0.82 (0.55 – 1.23)
30-day frequency rates, b (CI)
Cannabis	10.41 (8.87 – 11.95)^*	10.46 (8.57 – 12.35)^*	13.72 (11.52 – 15.93)^*	0.05 (−1.74 – 1.85)	3.31 (1.19 – 5.43)	3.26 (0.87 – 5.65)
Ecstasy	4.15 (3.38 – 4.93)^*	5.12 (4.19 – 6.06)^*	8.31 (7.20 – 9.42)^*	0.97 (0.12 – 1.82)	4.16 (3.12 – 5.19)^*	3.19 (2.03 – 4.35)^*
Speed	4.80 (3.85 – 5.76)^*	6.20 (3.85 – 5.76)^*	9.35 (8.05 – 10.66)^*	1.39 (0.33 – 2.46)	4.55 (3.30 – 5.80)^*	3.16 (1.77 – 4.55)^*
Cocaine	1.64 (0.85 – 2.43)^*	4.78 (3.85 – 5.70)^*	4.64(3.59 – 5.70)^*	3.17 (2.20 – 4.08)^*	3.00 (1.94 – 4.07)^*	−0.13 (−1.30 – 1.04)
Heroin	0.33 (−0.31 – 0.37)	0.29 (−0.21 – 0.79)	2.43 (1.86 – 3.01)^*	0.26 (−0.27 – 0.78)	2.40 (1.80 – 3.00)^*	2.14 (1.44 – 2.85)^*
Crystal meth	0.61 (−0.01 – 1.23)	2.21 (1.36 – 3.05)^*	4.36 (3.45 – 5.28)^*	1.60 (0.71 – 2.48)^*	3.75 (2.80 – 4.71)^*	2.16 (1.04 – 3.27)^*
LSD	1.65 (0.99 – 2.30)^*	2.81 (2.08 – 3.55)^*	5.04 (4.19 – 5.89)^*	1.17 (0.41 – 1.92)	3.39 (2.52 – 4.25)^*	2.22 (1.29 – 3.15)^*
Magic mushrooms	1.19 (0.65 – 1.73)^*	1.88 (1.24 – 2.51)^*	3.69 (2.95 – 4.43)^*	0.69 (0.02 – 1.35)	2.50 (1.74 – 3.26)^*	1.81 (0.97 – 2.65)^*
Ketamine	0.76 (0.10 – 1.41)	2.64 (1.89 – 3.40)^*	3.99 (3.17 – 4.82)^*	1.89 (1.08 – 2.70)^*	3.24 (2.36 – 4.11)^*	1.35 (0.40 – 2.30)
GHB	0.51 (−0.03 – 1.04)	1.17 (0.05 – 1.89)	3.58 (2.78 – 4.39)^*	0.66 (−0.01 – 1.43)	3.07 (2.21 – 3.93)^*	2.41 (1.42 – 3.39)^*
Research chemicals	1.33 (0.55 – 2.12)^*	0.47 (−0.60 – 1.54)	4.96 (3.97 – 5.94)^*	−0.87 (−1.98 – 0.24)	3.62 (2.59 – 4.65)^*	4.49 (3.22 – 5.75)^*
Bath salts	0.16 (−0.28 – 0.60)	−0.06 (−0.65 – 0.54)	3.39 (2.73 – 4.05)^*	−0.22 (−0.85 – 0.41)	3.22 (2.53 – 3.91)^*	3.44 (2.64 – 4.24)^*
“Smoking mixtures”	0.83 (0.14 – 1.51)	0.09 (−0.84 – 1.02)	4.91 (4.00 – 5.82)^*	−0.74 (−1.71 – 0.23)	4.09 (3.14 – 5.03)^*	4.83 (3.67 – 5.96)^*
Medicines	0.56 (−0.13 – 1.25)	−0.15 (−1.11 – 0.81)	5.30 (4.38 – 6.22)^*	−0.71 (−1.71 – 0.29)	4.74 (3.77 – 5.70)^*	5.45 (4.28 – 6.21)^*
Natural drugs	0.66 (0.22 – 1.02)	−0.05 (−0.66 – 0.56)	2.62 (1.90 –3.34)^*	−0.71 (−1.36 – −0.07)	1.96 (1.22 – 2.70)^*	2.67 (1.82 – 3.53)^*
Combination of substances, OR (CI)
Lifetime prevalence	8.19 (6.22 – 10.80)^*	16.89 (11.19 – 25.49)^*	14.40 (8.98 – 23.10)^*	2.06 (1.36 – 3.12)^*	1.76 (1.09 – 2.83)	0.85 (0.48 – 1.50)
Cannabis + ecstasy	9.77 (6.76 – 14.13)^*	9.96 (6.63 – 14.96)^*	18.24 (11.66 – 28.52)^*	1.02 (0.76 – 1.36)	1.87 (1.32 – 2.63)^*	1.83 (1.25 – 2.69)
Ecstasy + speed	14.95 (9.46 – 23.65)^*	17.48 (10.72 – 28.50)^*	35.76 (21.17 – 60.40)^*	1.17 (0.87 – 1.57)	2.39 (1.69 – 3.39)^*	2.05 (1.39 – 3.01)^*
Cannabis + cocaine	3.06 (1.77 – 5.28)^*	6.24 (3.55 – 10.96)^*	11.42 (6.42 – 20.31)^*	2.04 (1.34 – 3.10)^*	3.73 (2.42 –5.77)^*	1.83 (1.16 –2.89)
Cannabis + speed	10.57 (6.46 – 17.30)^*	14.25 (8.46 – 24.01)^*	25.42 (14.71 – 43.91)^*	1.358 (0.99 – 1.84)	2.40 (1.69 – 3.42)^*	1.78 (1.21 – 2.63)
Ecstasy + cocaine	4.03 (1.99 – 8.14)^*	8.64 (4.24 – 17.63)^*	16.87 (8.23 – 34.56)^*	2.14 (1.34 – 3.43)^*	4.19 (2.60 – 6.74)^*	1.95 (1.20 – 3.19)
Ecstasy + LSD	4.12 (1.88 –9.00)^*	16.52 (7.74 – 35.29)^*	26.94 (12.44 – 58.34)^*	4.01 (2.53 –6.36)^*	6.54 (4.04 – 10.60)^*	1.63 (1.04 – 2.54)
Cannabis + m. mushrooms	4.69 (2.34 – 9.38)^*	9.99 (4.94 – 20.22)^*	15.81 (7.69 – 32.48)^*	2.13 (1.37 – 3.32)^*	3.37 (2.11 – 5.38)^*	1.58 (0.98 – 2.56)
Speed + cocaine	3.08 (1.59 – 5.97)^*	8.76 (4.55 – 16.86)^*	13.56 (6.92 – 26.57)^*	2.84 (1.79 – 4.51)^*	4.41 (2.71 – 7.16)^*	1.55 (0.96 – 2.49)
Alcohol + illicit drug	2.16 (1.65 – 2.82)^*	2.42 (1.76 – 2.82)^*	2.61 (1.81 – 3.77)^*	1.12 (0.83 – 1.51)	1.21 (0.85 – 1.72)	1.08 (0.73 – 1.59)
Medicines + alcohol/illicit drug	0.89 (0.54 – 1.47)	0.95 (0.51 – 1.74)	3.08 (1.80 – 5.25)^*	1.07 (0.58 – 1.95)	3.46 (2.04 – 5.89)^*	3.25 (1.73 – 6.11)^*

^a Refrence category)

^* p < 0.001.

In terms of combining two or more substances, the conservative class was less likely than all other classes to have combined any two or more substances within their lifetime. In addition, the psychedelic class showed a greater likelihood to have consumed more than one substance simultaneously than the traditional class. The conservative class was less likely to have tried any of the suggested substance combination than any of the other classes, except for the combination of medicines and alcohol or illicit drugs, respectively. The psychedelic class was significantly more likely than the traditional class to have combined cannabis and cocaine, ecstasy and cocaine, ecstasy and LSD, cannabis and magic mushrooms, and speed and cocaine, whereas unselective class was significantly more likely than the traditional class to combine any of the suggested substances, apart from the combination of alcohol and illicit drugs. In addition, the unselective class consumed the combination of cannabis and ecstasy at a greater likelihood than the psychedelic class.

NPS use

The most commonly cited NPS were the mentions of “smoking mixtures,” meaning synthetic cannabinoid products to be smoked or vaped (18 mentions in total) and the synthetic cathinone methylone (11 mentions in total), as well as ketamine derivate methoxetamine (MXE; 10 mentions in total).

Due to the small number of mentions, no further inferential statistics were employed to detect class differences. Table 5 illustrates the mentions of NPS distributed across the latent classes. Participants in the traditional class were most likely to name an NPS they had consumed accumulating in 49 mentions of a new substance. Twenty-four separate substances were named, more than in all other latent classes. The most frequently named substances in this class were synthetic cannabinoids in the mention of “smoking mixtures” followed by the synthetic cathinone methylone and the arylcyclohexylamine MXE.

Table 5. Frequency of NPS mentioned by class.

Conservative class			Traditional class			Psychedelic class			Unselective class
Substance	N	type	Substance	N	type	Substance	N	type	Substance	N	type
AM-2201	1	cannabinoids	“smoking mixtures“	7	cannabinoids	2C-B	3	phenethylamines	“smoking mixtures”	7	cannabinoids
“Smoking mixtures”	1	cannabinoids	methylone	6	cathinones	mephedrone	3	cathinone	MDPV	5	cathinones
Mephedrone	1	cathinone	MXE	4	arylcyclohexylamine	2C-E	2	phenethylamines	MXE	4	arylcyclohexylamine
			4-FA	3	phenethylamines	methylone	2	cathinone	2C-E	3	phenethylamines
			25C-NBOMe	2	phenethylamines	MXE	2	arylcyclohexylamine	4-FA	3	phenethylamines
			2C-B	2	phenethylamines	bath salts^2	1	cathinones	methylone	3	cathinone
			2C-D	2	phenethylamines	2-FA	1	phenethylamines	2C-B	2	phenethylamines
			2C-P	2	phenethylamines	25C-NBOH	1	phenethylamines	2C-I	2	phenethylamines
			3-MMC	2	cathinones	25i-NBOH	1	phenethylamines	3C-E	2	phenethylamines
			6-APB	2	arylalkylamines	2C-I	1	phenethylamines	4-AcO-DMT	2	indoalkylamine
			MDPV	2	cathinones	3-MeO-PCP	1	arylcyclohexylamine	NBOMe^1	2	phenethylamines
			mephedrone	2	cathinone	4-FA	1	phenethylamines	bath salts^2	1	cathinones
			NBOMe^1	2	phenethylamines	“smoking mixtures”	1	cannabinoids	25C-NBOMe	1	phenethylamines
			2C-C	1	phenethylamines	NBOMe^1	1	phenethylamines	25i-NBOMe	1	phenethylamines
			2C-I	1	phenethylamines				3-MMC	1	cathinones
			4-AcO-DMT	1	indoalkylamine				4-HO-MET	1	indoalkylamine
			5-APB	1	arylalkylamines				6-APB	1	arylalkylamines
			5-MAPB	1	arylalkylamines				butylone	1	cathinones
			5-MeO-MiPT	1	indoalkylamine				DMT	1	indoalkylamine
			DMT	1	indoalkylamine				DOI	1	phenethylamines
			JWH-122	1	cannabinoids				DOM	1	phenethylamines
			JWH-210	1	cannabinoids				AB-FUBINACA	1	cannabinoids
			kratom	1	plants and extracts				UR-144	1	cannabinoids
			UR-144	1	cannabinoids
Total	3			49			21			47

¹ Mention of unspecific NBOMe-derivate.

² Not a specific substance named, bath salts typically contain synthetic cathinones.

Participants in the psychedelic class named phenethylamines most frequently. The most frequently named single substances were mephedrone as well as 2C-B followed by MXE, the synthetic cathinone methylone, and the phenethylamine 2C-E.

Within the unselective class, “smoking mixtures” were mentioned most often followed by the synthetic cathinone MDPV, and MXE. Phenethylamines were the most frequently mentioned group of NPS.

Discussion

The sample of this study was drawn from a population whose affinity for psychoactive substances is well-known (Calafat et al., 1999; Tossmann et al., 2001; Winstock, Griffiths, & Stewart, 2001). Within this selected population distinct substance use patterns were found indicating that different approaches of prevention and intervention may be needed.

The largest groups that were found were the conservative and traditional class characterized by their relatively limited substance use in terms of the range of substances that were used. Although the conservative class' prevalence of cannabis use was above that of the general population (Pabst, Kraus, Gomes De Matos, & Piontek, 2013), the use of illicit substances being limited to cannabis seems to reflect a pattern found outside of the nightlife population. Cannabis is the most widely used illicit substance in Europe (EMCDDA, 2015; Vincente, Olszewski, & Matias, 2008) and in Germany with prevalence rates substantially higher than any other illicit substances (Pabst et al., 2013). In order to reduce harms associated with substance use among nightlife attendees it may therefore be sufficient to concentrate on reducing the risks of cannabis.

The traditional class, named for their use of “traditional” club drugs, is the largest class within the sample. White and colleagues (2006) interviewed regular ecstasy users, nearly all of which named a wide range of perceived benefits of the use of psychostimulants, particularly those of ecstasy. In addition, psychological harms pertaining to dependence, psychosis, or depression were rarely acknowledged. Although ecstasy was the second most often used substance within the present sample, most displayed a weekly consumption at most. However, a study by Parrot (2015) suggests that even low frequency consumption of stimulants (including nicotine, MDMA, amphetamines and mephedrone) is physiologically and psychologically harmful, especially when used over a prolonged period of time. This suggests that the traditional class is at a greater risk for problematic substance use than the conservative class.

In addition to the traditional club drugs, members identified in the psychedelic class consumed hallucinogenic and dissociative substances such as LSD, magic mushroom, and ketamine. Some of these substances are associated with synesthetic effects in context with music (Luke & Terhune, 2013; Nichols, 2004). The psychedelic class may therefore be specific to this context. Considering the sample was limited to electronic dance music events, other types of consumption patterns requiring different prevention and intervention approaches are likely to occur in other settings. The link between substance use and nightlife is not exclusive to the rave and electronic dance music scene but can be found in other music genre scenes as well (Lim, Hellard, Hocking, & Aitken, 2008; Van Havere, Vanderplasschen, Lammertyn, Broekaert, & Bellis, 2011). Further research is needed to identify potentially high-risk substance use patterns in other nightlife contexts and outside of the nightlife and consequently extend prevention toward this population.

The unselective class displayed the riskiest consumption pattern of the sample as well as the greatest need for support when reducing the risks and harms of substance use. In addition to a wide range of substances used, the prevalence of simultaneously using multiple substances was highest within this group. This substance use pattern indicates little reflection on the behavior. Given one of the main purposes of the prevention initiatives is to engage patrons in self-reflection on their substance use the unselective class may be underrepresented in this study. Only patrons that were interested in the offers of the peers were approached to take part in the study- thus less reflective users may not be concerned with the prevention projects at all. Although the unselective class represents a small subset of an already distinct population there exists a great importance for indicative, tailored prevention strategies and harm reduction measures, due to the class' high risk of adverse consequences of substance use.

As one main aim of the prevention projects is to engage the patrons in a dialogue about their consumption, a measure to prevent problematic use or the development of substance use disorders should be considered. The implementation of a screening process within these dialogues could be useful in identifying those at high-risk for developing substance use disorders or those most likely to suffer the consequences of risky substance use. The use of screening tools has been shown to be effective in recognizing such individuals amongst young people (Stockings et al., 2016). Referrals to treatment or brief intervention can follow such screenings.

A further additional measure to prevent harms in this population may be pill-checking. This method has found great approval in countries where this is possible (Brunt et al., 2017; Hungerbuehler, Buecheli, & Schaub, 2011; Wiese & Verthein, 2014) as it provides a further opportunity of educating the consumer on the harms and risks of the substance they are to ingest. However, these tests are not legally possible in Germany.

A further objective of this study was to investigate the role that NPS play within this population of substance users. Generally, NPS use was more prevalent in the observed sample than in the general population (Pabst et al., 2013; Werse, Kamphausen, Egger, Sravari, & Müller, 2015), although it played a substantially smaller role than other established substances. Of the NPS, synthetic cannabinoids were used most frequently on a daily basis. Considering that cannabis was the only substance to be used daily at a greater likelihood, this is not as surprising. Studies on the motivations for “legal high” consumption have reported the use of “smoking mixtures” to substitute cannabis use; either in order to circumvent drug-tests, as a legal alternative for cannabis or due to a supply shortage of cannabis (Werse & Müller, 2010). However, other NPS were used as an addition to known illicit substances rather than a substitute for these (Werse & Morgenstern, 2012). Given the numerous mentions of separate substances with a relatively low 12-month prevalence rate of NPS use, acquiring greater insights into the motivations for the use of specific substances would benefit stakeholders involved in reducing the harms of these substances.

The differences in the classes were also apparent in the use of NPS. The smallest class, the unselective class, named the second most NPS further indicating the high-risk consumption pattern of this group. The consumption of NPS affords particular harms such as the unknown content of substances, i.e., the active ingredients contained in one product, as well as the uncertain potency of the substances. In addition, emergency personnel are not always able to appropriately respond to intoxications as the consumed substances are unknown. Furthermore, there is no information on the long-term effects of these substances (Zamengo, Frison, Bettin & Sciarrone, 2014). Harm reduction strategies discussed earlier in relation to the unselective class, drug-checking, drug user empowerment and outreach at nightlife location have also been established as providing ample opportunity for reducing risks in relation to NPS (Móró, 2014). Considering the high prevalence of illicit substance use, especially in the unselective class, outlawing NPS may not be the most effective way of preventing harms resulting from their consumption.

The prevention projects active in the nightlife involved are generally well received by patrons (Boiler, Voorham, Monshouwer, Van Hasselt, & Bellis, 2011; Calafat & Juan, 2009; Hannemann & Piontek, 2015). This indicates a wide reach and a good opportunity to implement further measures. However, prevention in a nightlife setting has not been sufficiently nor methodologically soundly evaluated (Bühler & Thrul, 2015; Calafat & Juan, 2009; Stockings et al., 2016). It is therefore not possible to make comprehensive claims on their efficacy. However, personal brief interventions or multicomponent programmes as suggested above have been able to produce good results (Bühler & Thrul, 2015). More research is needed that meets methodological requirements in order to fully assess the efficacy of prevention in the nightlife setting.

One limitation of this study lies in the subject matter. The drawbacks of self-reported substance use have been well established (Del Boca & Darkes, 2003; Tourangeau & Yan, 2007). However, to counteract social desirability no further personal information was established other than sex, age, and hometown. A further drawback with self-reports is when this happens within the context of substance use. Miller and colleagues (2015) concluded that substance use was underreported in clubs when comparing self-report measures with biological markers.

Additionally, substance use patterns include more than the types of substances used and the combinations in which there are used but also the route of administration. Although this was assessed relating to NPS consumption in the original questionnaire, due to the low response rate of this question it was not included in the analyses. However, it should be further assessed whether the riskiest consumption patterns are also riskiest in terms of routes of administrations (i.e., intravenous substance use).

One problem that arises with the study of NPS is that there is no reliable definition of what constitutes an NPS. Some studies regard any yet legal substance as an NPS; others regard crystal meth as a new substance. With legal changes across Europe in response to the surge of new substances on the scene, the definition remains unstable. Therefore, all mentions of a new substance used by the participants were subject to interpretation. Some participants mentioned well-established substances such as cannabis and MDMA as new substances, whereas the analysis only included non-established substances (see Methods section). Particularly, because the investigation was conducted over an extended period (23 months), the interpretation needed to be less conservative. What are considered established substances by some today may not have been at the time the responses were given. Finally, it needs to be considered that latent class analyses do not provide a perfect classification of individuals. Class membership was based on posterior probabilities and thus, a certain amount of uncertainty must be taken into account. One of the strengths of the study also lies in the assessment of NPS use. All NPS names were researched individually in scientific sources as well as informal sources (such as internet forums). As the mentions of the substances were researched over an extended period of time, substances not yet known amongst the scientific community or in forums at the time of mention could later be included in the study. By not restricting the responses to a limited number of known NPS, there is the possibility of an entirely new substance being named. Given the quick pace and the sheer amount at which new substances are emerging on the drug-scene, open-ended questions seems the most effective method of documenting NPS use.

In conclusion, this study exemplified the heterogeneity of a population notorious for its affinity for substance use. Although risky patterns of substance use were detected in this sample, there were substantial differences between the participants that prevention and policy makers should be aware of. In addition, this study illustrated the use of NPS within Munich's nightlife. While still a fringe phenomenon, its use was particularly prevalent among the high-risk users of the unselective group. Safer use measures, beyond those of education, should be considered in order to reduce the potential harms these substances have on their consumers. Further research is needed in order to tailor evidence-based prevention measures and policies.

Declaration of interest

Ludwig Kraus and Daniela Piontek have received a grant from Lundbeck GmbH for a research project on alcohol epidemiology related to this study.

Funding

The data collection for this article was made possible through the support of the German Federal Ministry of Health (Phar-Mon: IIA5-2513DSM300) and through the great commitment of the prevention project MINDZONE, in particular Dirk Grimm and Sonia Nunes, in Munich, as well as the projects Eve&Rave in Münster and Musikprojekt: Drogerie in Erfurt.