A structured review of reasons for ecstasy use and related behaviours: pointers for future research

Background While the health risks of using ecstasy warrant intervention development, a recent meta-analysis of determinants of ecstasy use identified a number of lacunae in the literature. Specifically, no studies were included that address behaviours other than 'using ecstasy' (e.g. 'trying out ecstasy' or 'ceasing ecstasy use'). However, because meta-analyses aim to integrate study results quantitatively, the resulting rigid exclusion criteria cause many studies to be discarded on the basis of their qualitative methodology. Such qualitative studies may nonetheless provide valuable insights to guide future research. To provide an overview of these insights regarding ecstasy use, the current study summarizes and combines what is known from qualitative and exploratory quantitative literature on ecstasy use. Methods The databases PsycINFO and MedLine were searched for publications reporting reasons for ecstasy use and related behaviour, and the results were structured and discussed per behaviour and compared over behaviours. Results Two main categories of reasons were found. The first category comprised reasons to start using ecstasy, use ecstasy, use ecstasy more often, and refrain from ceasing ecstasy use. The second category comprised reasons to refrain from starting to use ecstasy, use less ecstasy, and cease using ecstasy. Reasons for related behaviours within each of these two categories appear to differ, but not as substantially as between the two categories. A large number of reasons that were not yet explored in quantitative research emerged. Conclusion The current summary and combination of exploratory studies yields useful lists of reasons for each behaviour. Before these lists can inform interventions, however, they beg quantitative verification. Also, similarity of determinant configurations of different behaviours can be assessed by addressing determinants of several behaviours in one study. Another important finding is that meta-analytical integration of the literature may overlook important findings and implications. Thus, qualitative reviews remain useful instruments in setting the research agenda.


Background
Although evidence that ecstasy use may be damaging to health accumulates [1,2], its prevalence persists [3,4]. Theory-based behavioural interventions have successfully generated behaviour change in other areas [5,6], and may likewise have beneficial effects when applied to ecstasy use. However, development of an effective intervention requires knowledge about which modifiable determi-nants need to be targeted [7]. A recent meta-analysis of quantitative studies on ecstasy use and related behaviours, which aimed to provide this knowledge, concluded that much research is still necessary [8]. Specifically, of all behaviours relevant to ecstasy use (e.g. trying out ecstasy, applying harm reduction strategies, ceasing ecstasy use) only the broad behavioural category of 'using ecstasy' had been addressed by the included studies. Because metaanalyses aim to quantitatively integrate the literature, they generally exclude studies that do not report certain statistics. For example, the meta-analysis about ecstasy use included only studies that "assess quantitatively the relationship between determinants and behaviour or intention" [ [8], p. 110]. These restrictive inclusion criteria led to the exclusion of all qualitative and exploratory studies into the reasons for ecstasy use, while paradoxically, it is exactly this exploratory methodology that renders these studies particularly valuable in setting the research agenda. The current review sets out to inform future research into ecstasy use and related behaviours by summarising this qualitative and exploratory quantitative literature.
Although the aforementioned meta-analysis did result in a list of determinants that seem relevant for ecstasy use (i.e. attitude, subjective norm, perceived behavioural control, moral norm, anticipated regret and habit), consideration of the most salient underlying beliefs indicated that some of these determinants of ecstasy use may prove exceptionally hard to modify [8]. For example, although ecstasy users had a higher descriptive norm than nonusers (i.e. users perceived there to be more ecstasy use at dance events), this difference appeared to reflect an underestimation of actual ecstasy use prevalence on the part of non-users. In this case, therefore, correcting this erroneous belief would lead to more ecstasy use, and an intervention aiming to decrease descriptive norms among ecstasy users would have to present incorrect information about the prevalence of ecstasy use (i.e. portray ecstasy use prevalence as lower than it is). The difficulties of intervening on the reported determinants seem to be underlined by a recent evaluation of an intervention among ecstasy users, where the authors concluded that a brief motivational intervention was no more efficient than the informationonly control condition [though use decreased in both conditions, which most control participants attributed to the self-assessment at baseline; [9]]. In addition, none of the reviewed studies specifically addressed the initiation or cessation of ecstasy use or the application of harm reduction strategies. All studies addressed the broad behavioural category of 'using ecstasy' [e.g. by comparing users with non-users, or examining the intention to 'use ecstasy'; [8]], which may be problematic because determinants of related but different behaviours such as these are assumed to differ [7,10]. Finally, it seemed that the sum-marized research had not addressed a number of potentially relevant determinants [8].
Thus, so far, only one of several relevant behaviours has been reviewed; the identified determinants appear hard to modify; and relevant determinants may have been omitted. The current paper investigates whether studies into other ecstasy use-related behaviours or addressing other determinants do exist, but were excluded by the rigid exclusion criteria of the meta-analysis drawing these conclusions. Specifically, the current paper reports a structured review of all qualitative studies and exploratory studies, which did not report an association with intention or behaviour but did report reasons for performing (or not performing) an ecstasy use-related behaviour (e.g. trying out ecstasy, ceasing ecstasy use, getting ecstasy tested, or drinking water during use). The aim of this overview is to guide further quantitative research into ecstasy use, eventually enabling development of effective evidence-based interventions.

Methods
Relevant literature was identified through the databases PsycINFO and MedLine. These were accessed through the Ovid SilverPlatter WebSpirs interface (version 5.12). At the 21 st of August 2008, the query "(("ecstasy" OR "mdma" OR "xtc" OR "methyldioxymethamphetamine" OR "party drug" OR "party drugs" OR "club drug" OR "club drugs" OR "dance drug" OR "dance drugs") IN TI, AB) AND (LA = "English")" was entered, which searched for all English records that contain ecstasy (or a synonym) in their title or abstract. This query yielded 4574 hits (1232 from PsycINFO and 2021 from MedLine). The phrase "NOT ("mouse" OR "mice" OR "rat" OR "rats")" was added, which eliminated 1321 hits. The remaining 3253 entries were downloaded and imported into a reference management program [11], which automatically identified and deleted 741 duplicates (defined as entries with the same title and year of publication). The titles and abstracts of the 2512 remaining records were manually inspected for relevance, and all publications reporting reasons for an ecstasy use-related behaviour were acquired. The acquired publications were examined in more detail, and if upon closer inspection a paper turned out to not report any reasons for ecstasy use or a related behaviour, it was excluded accordingly. In total, 2490 publications were excluded.
Most excluded records described biological studies [e.g. [12,13]], followed by a large number of publications describing prevalence of drug use, sometimes combined with demographic variables [e.g. [14,15]]. A number of studies also investigated consequences, perceived effects, or risks of ecstasy use [e.g. [16,17]]. Unless these were reported in response to a question about reasons, or con-sidered as reasons by the author of the original paper, these were not considered as reasons in the current paper either. Also, studies not reporting original empirical data were excluded [e.g., discussions or reviews, see for example [18,19]]. Finally, studies that did not investigate the target population (i.e. young recreational ecstasy users in western society) were excluded [e.g. [20,21]], as it has been shown that factors influencing behaviour can differ between populations [22][23][24].

Included studies
Details of the included 22 publications  are provided in Table 1. The included studies examined several behaviours, and given the qualitative or exploratory quantitative nature of these data, this renders presentation of all results challenging. In 2006, Baylen and Rosenberg [47] reported a review with a goal similar to the current goal, and with resulting data that were structured similarly. As the way in which they report their results seems very useful for the current purposes, this approach will roughly be followed. Specifically, we will report the percentages that were reported in each study in a table, grouped by reason category (rows) and behaviour (columns). Two groups of behaviours share a number of reasons, and the results pertaining to these groups of behaviours are therefore presented in the same table. Table 2 contains four "more-use behaviours": behaviours leading to consumption of more ecstasy (starting use, "using ecstasy" in general, using more ecstasy, and not ceasing ecstasy use). Table 3 contains three "less-use behaviours": behaviours leading to consumption of less ecstasy (not starting use, using less ecstasy, or ceasing use). In addition to these seven behaviours, a number of studies addressed reasons for combining ecstasy with other drugs and applying harm reduction strategies [25,28,29,35,36,40,46]. Because of the multitude of different behaviours, these reasons will not be tabulated but rather discussed in the text. Tables 2 and 3, but exclusively to ease presentation and discussion, as the qualitative nature of this review prohibits quantitative integration of the results. These categories were established by first entering all reasons into an online database [where they are available to the interested reader; see [48]]. Then, the first author clustered those reasons that appeared similar, eventually reaching a list of relatively distinct categories. The clustering does not suggest that the grouped reasons are psychologically similar, but rather only serves to ease presentation and discussion of the results. Therefore, the original descriptions for each reason are listed in the first column of every table. In addition, interested readers can consult the original (unclustered) lists of reasons, which have been made pub-licly available [48]. The descriptions are provided in the same order as the corresponding reporting frequency (if available). When several reasons within one category were extracted from one study, the descriptions and reporting frequencies are separated by commas.

More-use behaviours
A number of patterns emerge when all studies into starting use, "using ecstasy" in general, using more ecstasy, and not ceasing ecstasy use are combined (see Table 2). First, like studies in the meta-analysis, most currently reviewed studies focused on the behaviour 'using ecstasy'. However, there is still a lot of data on other behaviours, such as starting to use ecstasy, using more ecstasy, and not ceasing ecstasy use. The reason categories in which the reasons for these behaviours fall have been summarised in Table 4. In this table, a tentative attempt has been made to indicate each reason category's relevance, based on the frequency with which the reasons in a category were endorsed by participants. When at least one reason in a reason category was endorsed by more than half of the participants, a reason category was considered very relevant (indicated in the table by '+'); when reasons in a reason category were reported by less than half, but more than one in ten participants, a reason category was considered moderately relevant ('±'); and when reasons in a reason category were reported by one in ten participants or less, the category was considered minimally relevant ('-'). When no frequency information was available (i.e. with interview studies where only quotes were provided), the symbol 'N' is used, and in the one case where the only study suggested that a reason was irrelevant, the symbol '8' is used.
Three reasons from Table 2, specifically fear of health risks, noticing mood/affective/cognitive changes in oneself, and one's own or another's bad experience, were each reported as reasons for using more ecstasy by 1% of the participants of study F [46]. The current authors assumed that this reflects measurement error, and upon being contacted, the original authors confirmed that they share this assumption. These reasons have therefore been omitted from Table 4.
When this information is combined for all more-use behaviours, it becomes clear that although some reason categories are equally relevant for all behaviours, there are also differences. Understandably, curiosity only seems relevant for starting ecstasy use; denial of negative effects only seems relevant for not ceasing use; and tolerance is only relevant for using more ecstasy. However, other differences are less intuitive. For example, social influence, and ecstasy's ability to provide energy and enhance social interaction and sensory perception, do not seem to play a big role in starting to use ecstasy.  Tables 2 and 3 to refer to this study, ** = probably primarily these region(s) (internet study), US = United States, UK = United Kingdom, EUR = Europe, RDS = respondent driven sampling (e.g. snowballing), NR = not reported

Less-use behaviours
When looking at behaviours that health promoters would generally construe as the desirable behaviours, it is clear that most research focussed on using less ecstasy and ceasing ecstasy use (see Table 3). Although the reasons for using less ecstasy and ceasing ecstasy use were quite similar, different reasons were reported for not starting ecstasy (see Table 4). For example, the cost of ecstasy does not deter potential users, but it does cause users to use less or even cease use, and other people's bad experiences can be a reason to use less ecstasy or cease altogether, but was not reported as a deterrent by non-users. Fearing or minimising health risks was reported for all three behaviours, although markedly less frequently as a deterrent to not start using ecstasy. When comparing these reasons with the 'more-use behaviours', it became clear that there was very little overlap. It seems that people have different reasons for starting ecstasy use and for not starting ecstasy use, and yet different reasons for using less ecstasy and for ceasing ecstasy use, and yet again for using ecstasy.

Combining drugs and applying harm reduction strategies
In total, seven studies reported reasons to combine ecstasy with other drugs, to apply harm reduction strategies, or to refrain from these behaviours. Five studies reported reasons to combine ecstasy with other drugs, and all reasons fell in one of two categories [29,35,36,40,46]: to enhance the ecstasy experience, or to minimize the comedown. To enhance the ecstasy experience, ecstasy was combined with ADHD medication, amphetamine, benzodiazepines, ketamine, LSD, marijuana, and Viagra (studies A [40], E [35], O [36] and R [29]). To minimize the comedown, ecstasy was combined with alcohol, antihistamine, benzodiazepines, cocaine, heroin, ketamine, marijuana, oxycodone-containing analgesics, rohypnol and valium (studies A [40], E [35], F [46], O [36] and R [29]). Interestingly, studies also reported that people refrained from combining with other drugs to maximize the ecstasy experience (studies A [40] and K [28]). Other reasons to refrain from combining were to minimize health risks (studies E [35] and K [28]) and after having heard about people dying from ecstasy use (study K [28]). To minimize health risks, participants also pre-or postloaded with vitamins, 5-Hydroxytryptophan (5-HTP), or selective serotonin reuptake inhibitors (SSRIs; studies O [36], Q [25] and R [29]). Preloading was also reported to enhance the ecstasy experience (studies Q [25] and R [29]), and postloading to minimize the comedown (studies Q [25] and R [29]), and one study reported that participants drank water during use to minimize the comedown (study Q [25]).
Most other harm reduction strategies were applied to minimize the potential for negative or adverse outcomes or health risks, namely drinking water and chilling out during ecstasy use (both from study Q [25]; chilling out means taking breaks from dancing), purchasing fewer ecstasy pills per occasion and limiting one's supply, only using when in a positive mood and with friends, and the more altruistic behaviours of guiding initiates and monitoring others (all from study E [35]). Then, a number of behaviours served to deal with the uncertain contents of ecstasy pills: using only after someone else had tried the ecstasy (study E [35]), obtaining pills from a reliable source (studies K [28] and O [36]), and purchasing pills in bulk (study E [35]). However, one study also reported that participants could avoid getting their ecstasy tested because they considered the uncertainty (as to the pill contents) 'part of the process' (study E [35]). Finally, one study reported a user who liked to drive under the influence of ecstasy because he enjoyed the experience (study E [35]).

Discussion
The papers included in this review contain valuable information. Compared to the synthesis of quantitative literature [8], the studies included here have indeed addressed more behaviours and more potential determinants. Though a minority of reasons reported here has already been quantitatively studied, those quantitative studies have only examined their relevance for the behaviour 'using ecstasy', and the results show that reasons for different behaviours (e.g. 'using ecstasy' and starting or ceasing ecstasy use) differ. This means that an intervention targeting the important determinants for 'using ecstasy' may be unable to effectively influence other behaviours (such as starting or ceasing ecstasy use). This also means that development of evidence-based interventions addressing these other behaviours first requires studies that map the determinant configurations for those specific behaviours.
Unfortunately, this review cannot inform intervention development, because this review only provides an overview of how frequently a reason was reported, and comparison of the frequency with which reasons were reported with the effect sizes found in the meta-analysis for the behaviour 'using ecstasy' showed that frequently reported reasons can correspond to beliefs that were not associated to frequency of use (i.e. beliefs held equally strongly by users and abstainers or by heavy users and light users). For example, in this review, using to 'enhance energy and dancing' was mentioned as a reason for using ecstasy by between 39% and 91%, yet in the meta-analysis [8], the belief that ecstasy helps to stay awake was associated to ecstasy use with a trivial effect size (i.e. Cohen's d < .2 [49]). A second limitation of this review is the fact that with two exceptions [26,39] all included studies have been performed in the US, the UK and Australia. It remains to be seen whether these conclusions apply to other countries such as the Netherlands. Third, following from the qualitative methodology, no conclusions can be Decreased drug effects (e.g. tolerance) or decreased appreciation of drug effects (needing to take more tablets than used to F ; I need to take more to get the same effects I ; increased tolerance K ) 35% F ; 9% I ; NN K Denying or forgetting negative effects (thinking "that won't happen to me" or "well, everybody else is doing ecstasy, and they're not having any problems", upon hearing about negative consequences K ; unpleasant experiences are forgotten and the excitement of going out and socialising takes over L ) NN K ; NN L Desire to be on the same level as friends (i.e. to be intoxicated in the same way) (to get into the spirit of the party H ; it makes you sad if you're at a club and you see everybody else is having all this fun and you're not K ; having a shared experience, desire to be on the same level as friends L ; social pressure ("you see friends having a great time and you want to join") O  21%, 23% T Note: superscripted letters denote studies as listed in Table 1, NN = no numbers (frequency or percentage) reported Table 2: Reasons and reported frequencies in each included study for starting ecstasy use, using ecstasy, using more ecstasy, and not ceasing ecstasy use. (Continued)    drawn except that reasons for related but different behaviours differ; nothing can be said about the degree to which they differ. This conclusion does, however, imply that determinant configurations (i.e. relative relevance of each of the determinants of a behaviour) of related but different behaviours differ as well.
Thus, there is a need to find out whether and to what degree determinant configurations for ecstasy use-related behaviours differ, ideally by comparing one or more behaviours (e.g. trying out ecstasy, using ecstasy and ceasing use) in one study. If these determinant structures do indeed differ, interventions should target different determinants depending on the specific behaviour that is targeted. In addition, future studies should measure the beliefs underlying the reasons for each behaviour. Ideally, for each behaviour, beliefs potentially underlying all reasons that have been studied (i.e. that have been marked in Table 4) are quantitatively examined. That way, over time, a clear picture will emerge as to the relative relevance of each of these reasons.
Just after the current review was completed, two new manuscripts were published that also addressed reasons to refrain from trying out ecstasy [50,51]. Vervaeke, Benschop and Korf conducted a factor analysis and found support for three factors: fear of the effects, rationality, and lack of opportunity [50]. Rosenberg, Baylen, Murray, Phillips, Tisak, Versland and Pristas used a different method and distinguished eight factors: harm to thinking, school, work, or athletic performance; ecstasy use is contrary to values/self-image; fear of failing a drug-test; fear of effects on body; difficulty with acquiring ecstasy; fear of dangerous outcomes; no enjoyment expected from ecstasy; and fear of loss of control. Most reasons underlying these factors reflect reasons from the earlier studies that were included in this review, but additional reasons to refrain from starting ecstasy use are also reported: uncertainty about pill contents, medical reasons, no access to ecstasy, already using another substance, and not using on principle [in [50]], and against religion, fear of damage to reputation, want to be a role model, don't know where to get it, and fear of losing control [in [51]]. Especially interesting are the different factor structures revealed by these two studies. This may be attributed to their different methodologies of constructing the factors, or to the different locales (Dutch versus American).

Conclusion
The results of this review provide a clear agenda for the research needed to develop evidence-based interventions addressing ecstasy use. Worth noting in this respect is that many studies reported overlapping reasons, both within and between studies. For example, it is unclear whether, and if so, to what degree, the reasons "help enjoy the com-pany of friends", "enhance socialising", and "being together with other people" reflect similar determinants. Ideally, a number of largely orthogonal beliefs can be identified [by studies such as [50] and [51]], the relevance of each of which can then be established for each behaviour. As multiple theoretical frameworks seem to apply to ecstasy use [8], it seems advisable for future studies to include variables specified by different theories so that it can be determined whether and how the relevant beliefs underlie these variables.
Finally, the combination of this study and the meta-analysis [8] has important implications. First, by virtue of their strict quantitative approach, meta-analyses provide only a very narrow view into the literature, excluding many studies that may provide valuable pointers for future research. By considering these excluded studies, qualitative reviews remain very valuable tools in synthesising the state of the literature. Second, conclusions from such qualitative reviews need to be quantitatively verified. As was also the case in the current review, results from qualitative research may not be corroborated by quantitative data. Thus, a balanced synthesis of the state of the art requires both metaanalytical and qualitative reviews.