To map the patterns of drug use generally and as they relate to NPSs, we devised a twofold process.
First, we present a LCA and regression modelling of the BYDS data. These were used to classify different patterns of drug-taking behaviour in BYDS participants. Additional analyses using the latent class data to inform the subsequent qualitative analyses are also summarised. These additional analyses included provisional regression models to identify risk factors associated with types of drug-taking behaviour. The way in which the LCA informed the sampling frame for qualitative data collection and the integration of regression analyses to the qualitative component is also presented.
Second, using the narrative data we develop a taxonomy of use and discuss how four distinctive drug use typologies emerged. Characteristics of these groups and how they relate to the various NPSs used are presented, with a discussion of how the qualitative analysis informed the next stage of quantitative modelling. Specifically, the themes identified in the qualitative analysis were aligned with BYDS quantitative variables. These BYDS variables were then used as predictors of substance use in the quantitative models and are presented in Chapter 4.
Latent class analysis of the Belfast Youth Development Study
Determining the number of drug classes
This study was a secondary analysis of data from the BYDS, a longitudinal study of substance use during adolescence (see Chapter 1 for more information). The data extraction plan was prepared and confirmed between October 2015 and November 2015 and then finalised in December 2015 through consultation with the knowledge exchange (KE) group. A draft analysis plan that addressed each of the research components was negotiated and completed by the team in January 2016 and a number of strategies were implemented over the next month to prepare the groundwork and data for the first stage of the analytical plan. We hypothesised in our application and follow-up responses that a number of homogeneous subpopulations would emerge in the data and would include a class of NPS users given the adequate numbers of NPS use in the sample. As depicted in , the first stage of this strategy was to apply LCA to the data to identify the number and nature of classes based on ‘ever use’ of the drug variables included in our data. These included alcohol, tobacco, cannabis, ecstasy, speed, LSD, cocaine, heroin, other pills, poppers and NPSs. The actual questions used to assess drug-taking behaviour are shown in Appendix 2, Table 13.
Representation of basic latent class structure: traditional and NPS drug use.
As is standard in LCA, models with two to six classes were estimated using MLR.91 To help prevent solutions based on local maxima and reduce the risk of missing the model with the best fit, initially 100 random sets of starting values were specified, with 20 final-stage optimisations. The models were then compared in terms of relative fit by using information theory-based fit statistics (), namely the AIC, BIC and ssaBIC. For these three fit measures, lower values are representative of better model fit.
Fit statistics for the models with one to six classes
The LMR and bootstrap likelihood ratio test were also used to compare models with increasing numbers of latent classes. Model estimates for both three- and four-class models were examined and the four-class solution was subsequently selected based on the interpretability of the results. After extensive exploratory modelling work we were able to determine that a four-class model was the best-fitting model according to the fit indices and information criteria; this provided us with unique and interpretable clusters of individuals with similar profiles within the latent SGs. Average latent class probabilities for most likely latent class membership are shown in Appendix 3, Table 14. shows that a four-class solution provides good cluster delineation (see Appendix 3, Table 15).
Probability of substance use in each of the four latent classes identified in the BYDS data.
Therefore, a four-class solution was adopted in the present research to account for the different typologies of drug taking. The four classes of drug use identified were as follows:
Alcohol – high probability of alcohol use; a p-value of < 0.5 for all other substance use types.
AT – high probability of AT use; a p-value of < 0.5 for all other substance use types.
ATC – high probability of ATC use; average probability (p < 0.5) for all other substance use types. The probability of poppers (p = 0.44) and cocaine (p = 0.43) use, although a p-value of < 0.50, is noteworthy for this group.
Polydrug – high probability of alcohol, tobacco, cannabis, ecstasy, speed, cocaine, poppers and NPS use. Probability of a p-value of < 0.5 for all other substance types.
The percentage of participants reporting ever using drugs is shown by latent class in . The latent class probabilities for each substance type are shown by latent class in Appendix 3, Table 15. The sample sizes of the four latent classes were as follows: class 1 (alcohol; n = 532), class 2 (AT; n = 926), class 3 (ATC; n = 367) and class 4 (polydrug; n = 214).
Percentage of participants reporting having ever used drugs by latent class and substance type
Predictors of drug use patterns (provisional analysis)
To inform the design of the qualitative analysis, predictors of latent class membership were examined using multinomial logistic regression. A backwards stepwise regression approach with Holm–Bonferroni corrections applied to adjust for multiple comparisons was used. In all models, sex and free school meal (FSM) status were controlled for. Separate models were fit to examine the relationship between latent class membership and family/relationships, school, leisure, friends’ substance use, drug abuse, delinquent/criminal acts and psychosis variables. All predictors in the models were binary coded (0 = no; 1 = yes). Polydrug users were the reference group in this analysis. This analysis is considered provisional and was designed solely to inform the content and structure of the interview schedules. The provisional results are presented in full in Report Supplementary Material 2, Tables 39–50. More refined models, which were built on the basis of the qualitative findings, are discussed in Chapter 4.
Informing the qualitative analysis
The aim of the qualitative component was to explore in detail the life experiences of drug users and, in particular, how NPS use related to their overall drug portfolio. If a unique NPS-only class had emerged from the LCA, the aim could have been achieved by sampling from this class. However, because this class did not emerge, an alternative sampling strategy was devised to examine NPS use qualitatively with adequate control for polydrug use. Sampling from only polydrug users would have made it difficult to distinguish between factors related to NPS use and those related to polydrug use. As shown in and Appendix 3, Table 15, not all members of the polydrug group were NPS users. Therefore, in the qualitative analysis sampling, polydrug users were split into those who reported NPS use and those who did not. A small proportion of the class 3 (ATC) users reported using NPSs (see and Appendix 3, Table 15). We wanted to explore this rare group of users further to shed light on why users who are more selective about substance use choose to try NPSs. Importantly, although the majority of ATC users did not use NPSs, the vast majority did report using cannabis. To distinguish between life experiences related to cannabis and to NPSs, individuals who reported using cannabis but not NPSs were also sampled from to serve as a comparator group.
In summary, to distinguish between factors related to NPS use and other drug use, sampling was based on four groups (). Polydrug users were split into those who used NPSs and those who did not (polydrug and NPS users; polydrug users, no NPS). The ATC group (ATC and NPS users) were also sampled from, and a control group was formed of cannabis users who did not use NPSs (ATC no NPS). This allowed a distinction to be made in the analysis between NPS-specific experiences and those associated with cannabis. The specific criteria used to form the groups were as follows:
Sampling strategy for the qualitative analysis.
polydrug users and NPS – responded ‘yes’ to using mephedrone or legal highs; responded ‘yes’ to using all of the following: cigarettes, alcohol, cannabis, ecstasy, poppers and cocaine
polydrug users, no NPS – responded ‘yes’ to using cigarettes, alcohol, cannabis, ecstasy, poppers and cocaine
ATC and NPS – used mephedrone or legal highs, used cannabis; did not meet criteria for polydrug and NPSs
ATC user, no NPS – did not use mephedrone or legal highs, used cannabis; did not meet criteria for polydrug and no NPSs.
Qualitative interviews with BYDS participants commenced using the sampling framework derived from the LCAs. As regression analyses ran concurrent to data collection, emerging significant variables in the multinomial logistic regressions were passed to the researchers conducting the interviews for additional probing, in case the themes emerged in the participant’s narrative.
Belfast Youth Development Study participants were selected and invited for interview based on the LCA and sampling framework presented in . The qualitative sample was supplemented with individuals recruited from (1) drug and alcohol services and (2) the prison estates (see Chapter 2 for further information). Demographic information on the recruitment groups is shown in .
Sample information for interview participants
Based on the narrative data, and utilising our specific coding framework, we constructed a taxonomy of users (; see Chapter 2 for detail on construction). This allowed us to draw together the multiple lines of investigation. The rationale was that such data-driven classification would help us to identify subsets of participants.
Taxonomy of groups from qualitative classification
Following the development of the taxonomy groups, the data were interrogated using text searches and matrix code queries. This process of interrogation was to explore shared/differing characteristics in and across group taxonomies.
Group 4 (dependents) was by far the largest group and had variability, which from both a theoretical and empirical perspective necessitated subdivision into SGs.
The narratives provided dynamic data on the life cycle of participants’ use. For those who abstained from use or had never reported NPS use, data on motivations to resist/desist were also included. provides detail on interview participants by group.
Sample information for interview participants by group
As noted in Chapter 2, risk and protective factors were used as a shared theoretical lens between the quantitative and narrative data. To recap, longitudinal and other studies have identified a number of well-established individual, family, school and community risk factors that are associated with alcohol and drug use throughout the life course (see Chapter 2).69–73 The overall larger BYDS research programme quantitatively examined these interactions in depth for a large number of respondents.73 We do not lay claim to offering the same here. Rather, from a qualitative perspective the narratives articulate what these risks and protective factors looked like through the lived experiences of this group of participants. These are presented alongside information on the drug repertoires of the taxonomy groups (see Appendix 4 for detailed information).
Group 1: limited experimentals
Drug repertoire
The limited experimentals were participants who feature on the least severe end of the drug-using continuum. Members reported substance use that was generally limited to alcohol and cannabis; substances generally included alcohol and cannabis with one instance of experimenting with cocaine (n = 1) and poppers (n = 1). Although they had limited experience with substances, limited experimentals are nevertheless an important group to include as a comparison. Interestingly, none of those who fell into the limited experimentals group (n = 8) reported using any NPSs. All initiated with alcohol, with an average age at initiation of 15.6 years (range 13–18 years).
Moving beyond substance use profile, other shared characteristics for this group related to the risk and protective factors as described by participants; these emerged in the narrative interviews naturally, without prompting. Risk factors to experimental or opportunistic use of substances were evident in narrative accounts. Peer group pressure was important for both alcohol and drug use among this group. Noted, too, was the impact of changes in friendship groups, which resulted in exposure to different substance-using peers/acquaintances. Exposure to older drug-using peers was also noted. Half of the group mentioned some pressure to try substances at various times and set this alongside a sense of normalisation in the use of alcohol or cannabis among their peers. During adolescence, friends are crucial to a young person’s sense of belonging and drug-using behaviour.92,93 Commensurate with transitional time points in the life course, key events such as going on a holiday with friends in late adolescence marked a period of excessive alcohol use and experimentation with other substances (n = 2). In addition, attending university and nights out with new peers while living away from home contributed to increased substance use (n = 2).
Protective factors
Given that this group had limited substance use experience, limited experimentals appear not to have been significantly affected by risk factors such as negative peer influences and traumatic critical incidents, possibly owing to some of the protective factors they highlighted. For example, when younger, limited experimentals were characterised by having family environments where parental monitoring and control were evident. Parental monitoring can limit substance use by allowing parents to intervene in situations where their child may be at risk.94,95 The narrative interviews also revealed that limited experimentals expressed fear of getting caught using substances by their parents. One can speculate that the protective factors in place for traditional illicit drugs were also present for NPSs. The narrative interviews consistently indicated a strong influence by peers in both positive and negative ways. Indeed, the influence of peers as they relate to drug taking has been extensively documented in the literature.96,97 From our narrative interviews, two-thirds of limited experimentals made note of group affiliation in a pro-social peer setting as protective factor, mitigating excessive substance use. Around half of participants in this group reflected that their peers did not use alcohol and/or drugs and that they generally did not have the opportunity to use substances extensively in their social environment. In essence, such positive peer influence was protective.98–100
In a general sense, a normative developmental trajectory was evident and most participants in this group successfully transitioned to emerging adulthood101 and aged out of any significant substance use without intervention. This finding is consistent with the evidence from longitudinal studies that show that the vast majority of those who engage in delinquent behaviours as adolescents age out of these behaviours.73,102 Positive role transitions,103,104 defined as changes in relationships, education, work and outlook on life, were evident in this group. Over half were in a romantic partnership and tended to report having a partner who did not use alcohol or drugs and disapproved of substance use. There was a clear sense of the influence of financial independence as respondents matured. They noted a desire to invest in a car or home and to avoid spending money on alcohol or drugs. Another influencing factor in this group appeared to stem from witnessing adverse consequences of substance use, such as seeing people in the community who have used substances, losing friends to substance use or having a family member who has a substance use problem.
Group 2: past recreationals
Drug repertoire
Past recreationals had wider drug repertoires than limited experimentals. Nevertheless, the past recreationals group featured participants whose substance use was still largely opportunistic and reasonably limited; almost all had aged out of these behaviours at the time of interview. Less than half reported NPS use and this was largely confined to use of mephedrone pre legislative ban. Use of NPSs by these respondents tended to be opportunistic in the same way that their use of traditional substances had been. The availability of mephedrone among those who reported use was a significant motive for use: none of those in this group actively sought out the substance; it was made available by a peer, thus resulting in experimentation. None reported mephedrone use following the ban in 2010 and only one member of the group reported using any stimulant-type NPSs that emerged following mephedrone [one participant had a single experience with ‘China White’ (local term used for a specific synthetic stimulant)]. Experiences with SCs were generally isolated incidents, with the exception of one participant who reported numerous experiences with ‘trippy’ (local term used for SCs) both in England and Northern Ireland. Pre-ban use among this group points to factors such as legality and/or availability of NPSs as influential. All past recreationals tended to initiate with alcohol earlier than the participants in the limited experimentals group, with the average age at initiation being 14.1 (range 11–16) years. Going to university was a risk factor for almost one-third of this group (4/14 participants), particularly for those who attended university outside Northern Ireland. This opportunity for leeway and lack of parental monitoring that university attendance offered was seen as key in terms of increased alcohol use and experimentation with drugs. When members of this group attended university the protective influence of parents was largely removed.
Risk factors
Parents were discussed as risk factors for four of the participants in this group (in terms of their substance use and lack of supervision); for the majority, parents were more of a protective influence. Being under the influence of alcohol was a risk factor to drug use for almost half of this group. Some studies have suggested that delaying and limiting alcohol use is important in preventing drug use in young people.105 Almost one-third of this group (4/14 participants) made reference to peer pressure and engaging in alcohol or drug use to fit in. Romantic partners were a risk factor for some of this group in that having a partner who used drugs during teenage years resulted in participants’ use of drugs also. The role of partners in substance use has been well documented; according to the social development model, if there is a strong emotional bond with a partner this can have a negative or positive impact depending on the substance use habits of the partner.106
Protective factors
Despite the wider drug repertoire, this group described many of the risk and protective processes described by the limited experimentals group. For example, half of the group (7/14 participants) described fear of disclosure to parents or ‘strict’ parenting as something that served as a protective factor when it came to drug use. In contrast with the parenting described by the limited experimentals group, the parenting described in the past recreationals group tended to protect by limiting rather than preventing substance use. Other important established protective factors were apparent in the form of involvement in sports while growing up. The positive effects of engagement in sports and other pro-social activity has been very well evidenced.69 Interestingly, Iceland, which has experienced declines in substance use by children, has based its approach to substance use prevention in children on an evidence-based approach that specifically encourages uptake of sports and other activities. Having peers that had also aged out of drug use and/or having a romantic partner that had either aged out or was a non-user was highlighted as a protective influence and discussed by almost half of the sample (6/14 participants). Previous research has shown that behaviours such as substance use generally peak in adolescence,73,107 declining with biological and social maturity.70 Being surrounded by fewer substance-using peers presents fewer opportunities for use, less peer pressure to use and greater stigma associated with use. In addition, Moffitt70 argues that as adolescents mature the consequences of substance use are no longer perceived as rewards but instead as punishments; typically at this stage important aspects of life such as family, work and community become a key deterrent to substance use. This was described without prompting in the narrative interviews of the past recreationals group; normal markers of successful assumption of adult roles were evident and served a protective role in late adolescence/early adulthood. Six participants made note of employment, financial responsibilities and changing priorities as indicators that marked a transition point in ageing out of the drug use process. Of those respondents in this group who had children (n = 4), all described their children as having a protective influence, and some of those who did not have children speculated that marriage and children would mark a transition point, including in other behaviour such as drinking and ‘going out’. Listening to stories from friends or parents or observing fewer health patterns of alcohol or substance use discouraged more chaotic patterns of consumption and therefore acted as a protective influence for almost one-third of this group (4/14 participants).
Group 3: contemporary regulars
Drug repertoire
The third group in our taxonomy, contemporary regulars, consisted of 10 participants. The average age of substance use initiation in this group was lower than in groups 1 (limited experimentals) and 2 (past recreationals): 13.5 (range 9–15) years. Alcohol was the primary substance with which this group initiated; however, cannabis, SCs and solvents were also reported as substances initially used. Half of the participants in the contemporary regulars group were single and half were in a relationship. Participants in this group generally had much wider repertoires than those in groups 1 and 2 (see Appendix 4). All the contemporary regulars reported alcohol and cannabis use and, in contrast to groups 1 and 2, all reported use of a novel psychoactive substance. All but two reported using ecstasy and all but one reported cocaine use. More experimental drug use is also observed among this group, with reports of ketamine and GHB use, for example, as well as NPSs beyond mephedrone [e.g. China White and MDMI (definition unknown; discussed as a novel psychoactive substance in narrative accounts]. This willingness to experiment with a wider range of substances is not observed in groups 1 or 2. Participants in the contemporary regulars group generally had much more extensive drug repertoires and lengthier drug careers than those in group 2. Importantly for some participants in this group, the primary driving force behind the transition from traditional illicit drugs to NPSs was market factors. We hypothesise that this group could be considered to be non-marginalised availers (SG2) because their transition to and from NPSs was a functional one. It is important to note that almost half of participants in this group highlighted their concern that the ‘moreish’ properties of mephedrone had created in them the potential to escalate from ‘recreational’ to ‘problem’ use.
Risk factors
Whereas being under the influence of alcohol was highlighted as a significant risk factor for group 2, the past recreationals group’s drug use was not accidental, opportunistic or a result of a lapse of judgement while under the influence of alcohol. As with groups 1 and 2, the narrative accounts of over half of this group included information on how their parents attempted to act in a protective capacity during their teenage years. However, rather than serving to lessen drug taking, it resulted in a shift towards concealing their drug use. This is interesting as the parental control was reported as having a protective effect by groups 1 and 2 and may be reflective of different attitudes and beliefs held by the groups. For example, Tilton-Weaver et al.108 have theorised that negative peer influences could be reduced by parental monitoring, provided parents are viewed as legitimate authorities over friendships.109 Alternatively, if youths feel controlled and intruded on this could lead to an escalation of delinquent behaviours such as substance use or increased orientation towards substance-using peers.110,111
Protective factors
Reference to protective factors for the past recreationals group was centred around harm reduction as opposed to either preventing use (i.e. group 1) or limiting use (i.e. group 2). Decisions to use drugs appeared premeditated and informed and the protective influence of peers, in contrast to groups 1 and 2, was centred around harm reduction techniques for safer drug use – all contemporary regulars used drugs with peers and there are fewer references to non-using peers than in the other groups. Half of this group discussed the protective influence of peers in terms of offering advice/support to promote safer drug use. Furthermore, some participants in this group engaged in risk justification/neutralisation by either measuring the harms of drugs against the harms of alcohol or placing their more chaotic drug taking as a past feature of their drug career.
Although not always preventing drug taking, the protective influence of romantic partners who did not use drugs was still referred to by half of the past recreationals group. This influence resulted in some participants in this group abstaining and reporting signs of ageing out at the time of interview, whereas for others the influence of non-using romantic partners resulted in changes around the way participants used drugs, such as more infrequently, to a lesser extent or only on occasions when it could be concealed from non-using partners. One-third of the past recreationals group also discussed the protective influence of employment; this they presented as a measure of the functionality of their behaviour, that is, their drug use never resulted in them missing days from work.
Group 4: dependents
The dependents group, consisting of 52 participants, formed the largest group in our taxonomy. Dependents included individuals who self-reported dependency on one or more substance at some stage of their drug career. It is important to note that this was based on self-report of dependency and not verified independently by treatment data. However, in all narratives features of dependent use, for example in terms of tolerance and withdrawal, were noted.
The literature of drug dependency clearly evidences a sex disparity112 as well as reasons why females are less likely to report dependence.113 We posit that female users were likely under-represented in this grouping and we are cognisant of the accumulating epidemiological clinical and qualitative research that suggests that what determines whether or not drug abuse and dependence develops or progresses further is often sex specific or sex sensitive.112
Generally, the dependents group is characterised by having participants with more chaotic lives. Almost all were not in a relationship at the time of interview. The average age for substance use initiation was the earliest of the four groups, at 12.9 years (range 8–16 years). The majority initiated with alcohol (n = 36) and six participants initiated with cannabis. For a minority of the group it was not clear from their narratives which substance was used first; six participants reported synchronous onset with multiple substances, for example alcohol and cannabis, one participant reported using alcohol and mephedrone at the same age, one reported using alcohol and SCs at the same age, one reported using alcohol and ecstasy at the same age and one reported using cannabis and prescription medication at the same age.
Although all participants in the dependents group reported issues with dependency at some stage or another, patterns and levels of use within this group varied greatly. Some of the more established users reported long histories of drug use with multiple dependencies at various stages, whereas other, younger participants in this group reported dependency on a single substance [e.g. SCs, ‘synthetic’ acid/2,5-dimethoxy-4-iodophenethylamine (2C-I)]. Although all participants reported dependency, they were at very different points in their drug use career. Twenty-nine participants reported dependency on a single substance and 18 participants reported problems with multiple substances. Four participants from the dependents group reported never having used NPSs.
Subgroups
Given the significant variation in the dependents group in terms of participants’ dependency and relationship with NPSs, the group was subcategorised. SG1 comprised those for whom NPSs were their primary and only dependency. SG2 comprised those with dependency on a traditional illicit substance and for whom NPS served only to temporarily displace/substitute the traditional illicit substance (i.e. no dependency reported on NPSs). SG3 comprised individuals with multiple reported dependencies including dependency on NPSs.
Subgroup 1: generation new psychoactive substances
Drug repertoire
For SG1 (n = 14) the average age at initiation with any substance was 13.4 (range 11–16) years, with seven participants reporting alcohol as their first drug and two reporting cannabis as their first drug. For the other five participants it was not clear from their narrative which substance was used first; three reported using alcohol and cannabis at the same age, one reported using alcohol and mephedrone at the same age and one reported using alcohol and SCs at the same age.
Participants in group 4 were categorised into SG1 if their primary and only dependency was NPSs. The majority of the group (12/14 participants) had used SCs, with 11 participants reporting dependency on SCs. Five participants reported the use of mephedrone, with two reporting dependency. Three participants reported the use of other NPS powders, three reported the use of synthetic ecstasy (referred to as ‘herbal pills’) and one reported the use of NPS hallucinogens (e.g. 2C-I, 2C). The participant who mentioned using 2C, 2C-I reported dependency. For details of other substances, see Appendix 4.
Risk factors
When examining common characteristics of this group, peers stood out as the most common influential factor in terms of substance use, including NPS use. All participants reported that their friends used NPSs at the time of initiation. Watching friends and/or family using prior to first experience with NPSs served as a risk factor in this group. Community influences, including comments such as ‘everybody’s taking it’, and infiltration of NPSs into areas where young people lived and at sites where young people congregated were significant risk factors. Evidently the point in time at which NPSs emerged on the scene was significant given the age of participants in this group and the desire to experiment with substances. Notably, a proportion of participants in this group reported use, specifically of SCs, moving from social use with peers to use alone, which was an indicator of quick acceleration from experimental/social use to a more dependent pattern. The lack of knowledge about NPSs and naivety around what substances they were taking was a risk factor for this group, reflected in interchangeable terminology (e.g. ‘herbal’, legal highs, trippy) and elements of confusion when describing what they were using. Some reported not realising the signs of growing dependence on NPSs until they experienced withdrawal (e.g. sweating, vomiting, not being able to sleep, ‘craving the high’).
Early negative experiences with the education system was a predominant, common characteristic of this group. Seven participants in this SG were expelled from school/attended an alternative education programme. The remainder reported disengagement with education, with two participants leaving higher education as a result of their NPS use. Seven participants reported using substances in their school environment. Just as lack of school attachment is associated with substance use,114 so too are the reactions of the schools and employers reported here by the generation NPS group. Excluded from school, these children were removed from a relatively pro-social environment, leaving them with more free time in which to engage in substance use behaviour.115
Going to live with other family members (commonly a grandparent) or leaving home as a result of other family issues, for example parental alcohol use, parental mental health problems and conflict between teen and parent, were also common. Parental substance use can lead to child substance use: the child observes and then imitates their parent’s behaviour.116 In addition, in households where parents are problem users, there tends to be less routine, more conflict and fewer feelings of warmth and caring,117 further increasing the risk of substance use problems for their children. Indeed, lack of a structured environment was evident in the narratives: some participants reflected on hiding use from parents, yet it was clear that a number of parents were aware of substance use behaviour and did not seem to intervene effectively, either ignoring the behaviour or offering advice but with no sign of discipline or monitoring/control of behaviour. Parents splitting and conflict between parents were also factors for a significant number in this group. Parental separation has been found to be a key risk factor for early substance use, even after controlling for environmental and genetic factors.118 A small proportion reported using NPSs with a partner and two out of the three females in the sample highlighted male partners as facilitators of their use in terms of initiating substance use. Eight of the sample reported that NPSs functioned for them as a means to achieve escapism, help with anxiety/depression and a distortion of reality.
Protective factors
In contrast to groups 1 (limited experimentals), 2 (past recreationals) and 3 (contemporary regulars), discussion of any kind of protective influence was rare in the narratives of this generation NPS group. Being aware of deaths in the community, effects of their use on family members and peers stopping use were all factors highlighted by those who sought help with their dependency or ceased use by their own accord.
Subgroup 2: availers
Drug repertoire
Participants reported dependency on one or more traditional illicit substance and no dependency on NPSs. NPSs were ‘availed’ of by this group when their drug of choice was not obtainable; thus, NPSs only temporarily displaced traditional drugs, much like those in the contemporary regulars group. The average age at initiation in this SG was 12.6 (range 8–15) years. In total, 12 participants initiated with alcohol, two with cannabis and one with alcohol and cannabis at the same age. In terms of drug-taking profile, participants in SG2 resembled those in group 3 in terms of willingness to experiment with a range of drugs; however, SG2 differed in that participants reported use of heroin, which group 3 participants did not, and were more likely to report use of prescription medication.
Risk factors
The majority of participants in SG2 discussed mental health issues in the context of their drug narratives. Issues associated with mental health included depression, anxiety, panic disorder and schizophrenia. One-third of the group reported self-harm and suicide attempts. Of course, there is a much more complex interplay of factors regarding dual diagnosis, particularly in relation to the extent to which mental health problems contribute to substance use and the degree to which substance use exacerbates mental health problems.119 All participants in this group used substances with peers and, for many, peers increased risk in terms of initial access to drugs. Half of the group reported excessive use of alcohol during their drug careers. Most of this group reported limited engagement with education (n = 11); this ranged from lack of interest in school to expulsion and removal from mainstream education (n = 5). The majority of this group reported contact with the CJS, with contact largely related to drug-related offences, including possession, supply, crime to fund use and crime committed while under the influence. This crime and substance use nexus is believed to further escalate both behaviours, leading to more persistent offending and greater substance dependence.120
Protective factors
This group was interesting in that almost half of the participants (n = 7) described the influence of parents as largely protective and many of them questioned their motivations for drug use and the trajectory of dependency that followed, although, given the multitude of risk factors identified in their narratives, it is perhaps unsurprising that this protective influence was not enough on its own to limit or prevent their substance use dependency. Some attributed decision-making around drug use to factors associated with personality, for example impulsivity and desire to take risks. However, most of those who reported stable upbringings also acknowledged that they self-medicated to cope with their mental health issues and/or reported experiencing the death of a close family member or friend during teenage years. Almost one-third of SG2 participants discussed employment in a protective capacity and, for those outside prison, they seemed, like contemporary regular group participants, to measure functionality in terms of not missing days at work as a result of substance use.
Subgroup 3: persisters
Drug repertoire
The persisters SG includes the most chronic substance users (n = 18). All participants in SG3 reported multiple dependencies, including dependency on NPSs. The average age at initiation in this group was 11.7 (range 10–16) years. A total of 12 initiated with alcohol, three with cannabis, two with alcohol and cannabis at the same age, one with alcohol and ecstasy at the same age and one with cannabis and prescription medicines at the same age.
Participants in this SG, as with those in SG1, initially believed that SCs offered a safer alternative to the traditional illicit drugs that they were using in a harmful way. Comorbidities and vulnerabilities among this group made the potent effects of SCs particularly appealing and so resulted in sustained use. This SG was particularly well placed to track the evolution of SCs over time. Ironically, legislation designed to reduce the harm of SCs was counterproductive and resulted in the development of more potent and particularly risky third-generation legal highs.
Risk factors
Almost all documented risk factors were at play for this group and included negative experiences at school, adverse life/childhood experiences, mental illness, contact with the CJS, complex family issues, social deprivation and homelessness. In terms of education, 17 participants in this group reported significant disengagement with education, including being expelled (n = 12). One anomaly was a male participant from our service-using sample who attended school (and is now at university) but reported problems throughout school as the school environment exacerbated his underlying anxiety problem leading to more excessive substance use at that age.
A large proportion of participants in this SG reported an unstable home environment while growing up. Ten participants reported that one or both parents used substances themselves and in three cases a parent was instrumental in facilitating substance use (selling, providing substances). There was clear evidence of a lack of parental monitoring owing to parental substance use and issues such as strained relationships in the home or other adverse experiences relating to family life. More than half of the participants in this SG disclosed adverse life/child experiences (e.g. around one-quarter mentioned physical or sexual abuse). Unfortunately, those who have experienced such abuse are more likely to suffer negative consequences from using drugs.121 Five reported the death of a close family member and two reported the death of close friends.
Generally, having children was not a protective factor for this SG. Four participants had lost contact with their children/had children removed by social services, one of whom was regaining contact with his sons who are now adults. For the other three participants in this SG, substance use ceased/decreased during pregnancy/early years, but participants later resumed previous patterns of use.
For almost half of the participants in this SG, mental illness (anxiety, depression) was reported. Evidence of self-medication, in most cases through the use of SCs, was particularly highlighted by participants who had had adverse experiences or periods of homelessness.
As one might expect, all participants in this SG had contact with services at some stage. Five participants had contact with social services; for two this spanned more than two generations. Two participants reported being in care as a child. Consistent with the features of SG3, people who were in care as children tend to use drugs more regularly and use higher-class drugs than those who were not in care as children, making them a key group to monitor for substance use problems.122,123
Seven participants in this group had experienced homelessness at some stage and three were residing in hostels at the time of interview. Homelessness did not emerge as a risk factor for any of the other groups or SGs and it is likely that the severity of SG3 participants’ drug dependency led to them losing their home and made it difficult to rehouse them; this may also have escalated their drug dependency, taking drugs in an effort to cope.124,125
Contact with the CJS was common in this SG, with 15 participants reporting contact dating back to their early teens. Two of our female participants reported prostitution. There was evidence of contact with drug and alcohol services and multiple substance dependencies prior to NPS use for the majority of this SG.
Protective factors
The narratives of SG3 participants included references to risk factors; any kind of protective influence was lacking. For five participants in this SG, children were initially described as protective influences. These respondents reported abstaining from drugs when their children were born; however, the breakdown of relationships with significant others, or the removal of children from care, precipitated relapse. Nine participants described rebuffed efforts of family and friends to intervene and offer help and support. Several respondents reported creative, if unsuccessful, efforts by family and friends to assist in drawing attention to the effects of NPSs when participants were under the influence; for example, some participants were video recorded and the footage was shown to them when the participant was sober in a bid to deter them from future use. Employment was highlighted as protective for some in terms of either abstaining or reducing use, and job loss was described as a risk factor for relapse or returning to heavier drug use. Four of those from the prison sample highlighted prison itself as a protective influence and expressed concerns that they would return to using drugs on release.
Subgroup 4
Subgroup 4 did not report any use of NPSs while being dependent on other substances. They indicated that other substances were preferable and they chose not to use NPSs. The data on this group were very limited. Although this group was retained for transparency in the analysis, they are not reported on in detail.
summarises the risk and protective factors that emerged from the narratives by SG.
Groups by repertoire and risk/protective factors. a, In comparison with other SGs in group 4.
Chapter summary
The LCA revealed that only one class of substance use [class 4 (polydrug)] was characterised by NPS use. The multinomial regressions (see Appendix 5, Tables 18–20) showed that, based on the relative magnitude of effect sizes and number of significant differences between polydrug users and the other groups, there is a clear relationship between number of substances used and cumulative exposure to risk factors. In terms of exposure to drug-taking risk factors, polydrug users had the highest levels of exposure, followed by ATC users, then AT users, with the alcohol users showing the lowest levels of exposure.
The four-group taxonomy detailed patterns of substance use throughout our qualitative sample, ranging from those with very limited drug use repertoires to those reporting chronic use and dependency. Our data confirm the increasingly numerous messages from the literature that NPSs are best conceptualised as having a place within a polydrug use trajectory. As noted, commensurate with this hypothesis, our LCA did not generate a distinct NPS group, nor did it map neatly onto our qualitative group classifications. Rather, different NPSs made their way for different reasons and to different extents into the taxonomy groups 2–4.
Consistent with existing literature on trajectories of drug use, the taxonomy groups derived from the narratives clearly evidenced incrementally less favourable outcomes and multiple risks and adverse incidents as we progressed from group 1 to group 4. Common risks were early age at onset of substance use and length of substance use career. For example, age at onset progressively reduced as we moved through the taxonomy. The limited experimentals’ mean age at onset was 15.6 years, the past recreationals’ was 14 years, the contemporary regulars’ was 13 years and the dependents’ was 12.1 years. In addition, current and past extensiveness of substance behaviours appeared closely linked to early-life difficulties as reflected in the narratives (e.g. problems at school and the presence of criminal involvement in youth).
Dynamic changes such as important life events or ‘turning points’126 that can serve to protect or mitigate risks also showed a decreasing gradient as we moved through groups numerically. That is, such protective influences diminished rapidly by group to the extent that group 4 participants had few protective factors noted. Marriage, functional romantic relationships and a desire for financial security featured much more strongly in the narratives of the first two groups. Relationship breakdown, childhood in care, relapse or remission of physical and/or mental illness and bereavement appeared to have a firm influence on a person’s trajectory. The relative impact of these can be seen when looking across the four trajectories. Similarly clear was the cumulative effect of exposure to multiple adversity, which was very apparent in the female participants, especially the seven who fell into the dependent group. This summary of our groups through the narratives builds a picture of how risk factors were inter-related and tended to cluster together. For example, adverse childhood experiences appeared to be related to other family-level risk factors, for example poor parenting skills and parental substance use and mental illness.
A growing number of researchers are critical of too heavy a focus on risks at the expense of describing the specific pleasure and functions of specific drugs.127,128 Whereas this chapter has focused on risk, the pursuit of pleasure as it relates to NPS use of varying types is discussed in Chapter 5.
