Legal Minors Who Inject: Differences in Socio-Demographics and Treatment Needs Compared to Adults in a Swedish National Sample of People with Injecting Drug Use

Abstract Background Injection drug use among legal minors is under-researched. Although the population may be small in absolute terms, treatment needs may be greater than for those who began injecting as adults. Such knowledge may help tailor services more effectively. Previous research tends to use selective samples or focuses solely on medical indicators. The present study uses a larger sample drawn from national register data in Sweden over a 9-year period (2013–2021) to analyze differences in medical and social treatment needs between people who began injecting as legal minors and their older counterparts. Method Data on first-time visitors to needle and syringe programmes (n = 8225, mean age 37.6, 26% women) were used. Historical socio-demographics and presenting treatment needs were compared between those with a debut injecting age under 18, and those who began injecting as adults. Results The prevalence of injecting before 18 years was 29%. This group had more negative social circumstances, such as leaving school early, worse health, and greater service consumption, compared to those who began injecting as adults. In particular, they had been subjected to a greater level of control measures, such as arrest and compulsory care. Conclusions The present study shows that there are important health and social differences between those who inject prior to 18 and those who begin injecting as adults. This raises important questions for both child protection services and harm reduction approaches for legal minors who inject, who still qualify as ‘children’ in a legal and policy sense.


Introduction
Injecting drug use among legal minors is under-researched. We use the term 'legal minors' to foreground the important legal differences between those under and those 18 or over. Few countries have population size estimates for legal minors who inject, and surveys with people who inject drugs (PWID) often exclude those under the age of 18 (Barrett et al., 2013). Where estimates exist, they point to a population that may be small in actual size but with potentially extensive treatment needs. As noted by Watson et al. (2015, p. 371), 'regardless of population size, young people who inject drugs are at risk of potentially serious harms' which can have detrimental effects on their developmental trajectory making their treatment needs potentially different to adults. Yet services targeting legal minors who inject are uncommon globally (HIV/AIDS Alliance Ukraine, 2015). Existing harm reduction interventions focused on injecting drug use tend to be designed for adults, while issues regarding injecting may not factor into adolescent drug treatment services. It is possible that opportunities for both harm reduction and drug prevention services are being missed, given the acute and developmental issues associated with an early injection debut (Abelson et al., 2006;Busza et al., 2013;Bryant 2014;Degenhardt et al., 2008;Fuller et al., 2001;Miller et al., 2006). Knowledge on the debut age of injection for legal minors as well as how their life situation and treatment needs differ to adults with injecting drug use may help tailor services more effectively.
While there are some studies that have compared people who began injecting before they turned 18 to people who began injecting as adults (Battjes et al., 1992;Busza et al., 2013), these tend to be based on selective samples, such as participants recruited from street-based projects. Furthermore, the majority of these studies are located within the field of public health and medicine, often focusing on drug or alcohol use as a risk factor, e.g. for HIV or hepatitis C (HCV) transmission, and also predominantly from North America. The present study aims to contribute to this gap in the literature by using a national register of people attending needle and syringe programmes (NSP) in Sweden to compare both historical socio-demographics and medical and social treatment needs between those who began injecting before turning 18, and those who began injecting as adults.

Literature comparing adolescent and adult injectors
Behavioral survey data have indicated that between 1% and 3% of US adolescents in grades 9-12 (age 15-18) report injecting (Brighthaupt et al., 2019;Earlywine et al., 2020;Rajan et al., 2018). Higher rates of injecting drug use are found among those who used drugs in the past year (Wu & Howard, 2007) and in specific contexts, including among adolescents in drug treatment (Thurstone et al., 2013), and psychiatric care (Ilomäki et al., 2005). Intake data at Swedish institutional care homes, meanwhile, show that 2% of boys and 5% of girls report injecting (SiS, 2019). Previous intake data have shown this to be closer to 10% (SiS, 2010), while among 476 people in care homes surveyed in 2008, 18% of under 18s reported injecting in the previous six months (Richert, 2012).
Cross-sectional research from the US and Canada shows that a significant proportion of PWID (ranging from one third to two fifths) began before the age of 18 (Adams et al., 2020;Battjes et al., 1992;Miller et al., 2006). A Swedish study of 720 PWID surveyed in 2007-2008 found that 37% reported injecting debut before the age of 18 (Britton et al., 2009). Based on a smaller sample from the same database used in the present study, 20% of entrants to an NSP had an injection debut at or before the age of 16, with a further fifth initiating between the ages of 17 and 19 (Kåberg et al., 2020).
Mean ages of initiation into injecting recorded in the literature vary considerably, ranging from 14 to 23 depending on location, cohort (especially age), context, and historical period (Abelson et al., 2006;Crofts et al., 1996;Degenhardt et al., 2008;Fuller et al., 2001;Hadland et al., 2010;Martinez et al., 1998;Roy et al., 2002;Wilson et al., 2016). A mean age of injection debut of 20-21 years has previously been identified in Sweden (Britton et al. 2009;Kåberg et al., 2017;Stenström, 2008). For youth in institutional care homes who report injecting, the mean injection debut age is 16 for boys and 15 for girls (SiS, 2019). Note, however, that these means are based on a highly selective sample and cannot be generalized to all people who inject.
Some comparative studies have looked at young people who inject drugs compared to those who use drugs but do not inject (Kerr et al., 2009a;Liu et al., 2014;Martinez et al., 1998;Roy et al., 1995), while others have specifically compared younger versus older people who inject (Abelson et al., 2006;Battjes et al., 1992;Busza et al., 2013;Bryant 2014;Degenhardt et al., 2008;Fuller et al., 2001;Miller et al., 2006). Among young people aged 12-17 in in-patient treatment with both substance use disorder and depression symptoms, those who inject were more likely to have attempted suicide (Liu et al., 2014). Some studies that use comparison groups of people who use drugs but who do not inject have shown that childhood trauma (Kerr et al., 2009b;Martinez et al., 1998), sexual abuse (Strathdee et al., 2001), homelessness and experiences of serious assault (Fuller et al., 2002) are all associated with early onset of injecting. Younger injectors have been found to be more likely to be female (Degenhardt et al., 2008;Fuller et al., 2001); to have first had sex at an earlier age, to have sold sex, and to have been arrested ; to report more risky injecting practices such as frequency of sharing, to have less knowledge about health risks and to have had less contact with treatment for their drug use (Bryant, 2014); and to be HCV positive (Dalgard et al., 2009;Day et al., 2005).
Most studies, however, do not use the legal age of majority, i.e. 18 years, to delineate the comparison groups. Some studies use older cutoff ages (Degenhardt et al., 2008;Fuller et al., 2001;Kerr et al., 2009b), whereas adolescents are defined as under 21 in some cases (e.g. Fuller et al., 2001Fuller et al., , 2005, and as under 18 in others (Battjes et al., 1992;Busza et al., 2013). Those that have compared under 18s versus those over 18 (or in some cases those aged 16 and under versus those 17 and older) have supported the above findings but have also contributed additional findings specific to under 18s. For example, among a cohort of street-recruited young people aged 29 and under (n = 542), females were more likely to have begun injecting before the age of 16 (Miller et al., 2006). Those who began at 16 or under had longer injecting careers, were more likely to be HIV negative at intake, but also more likely to be HCV positive. In an Australian survey study of PWID (n = 336), those who began injecting on or before 16 were more likely to have had injecting drug use in the family, to have an unstable income, to have left school earlier, and to have had a shorter pre-injecting drug using career (Abelson et al., 2006).
To the best of our knowledge, there are no national or large-scale population studies based on registry data that have compared people who began injecting before turning 18 (legal age of majority) to those who began injecting as adults. Given that legal minors appear to have different social circumstances surrounding their initiation into injection behavior, as well as potentially different presenting needs, comparative study is warranted using more extensive data.

The present study
Using national register data on all adults attending NSP in Sweden in the period 2013-2021 (n = 8225), which includes age of injection debut, we compare people with injecting drug use before the age of 18 (IDU < 18) to those who began injecting at age 18 or older (IDU ≥ 18). Our main research question is: Are there differences in socio-demographic factors and presenting treatment needs between IDU < 18 compared to IDU ≥ 18. We hypothesize that IDU < 18s will: i) have different socio-demographics; ii) have different treatment needs, compared to IDU ≥ 18s. For the latter question, we compare dynamic treatment needs for a younger cohort of people aged 25 or under at point of entry to the NSP, in order to ascertain whether IDU < 18 have different presenting treatment needs to those who began injecting as young adults. A subsidiary research question concerns the prevalence of IDU < 18 in the total dataset. Here we also aim to provide descriptive data on age of injection debut and historical trends concerning age of debut and time between debut and NSP entry. Understanding the age range for injection debut, the delay in seeking treatment, and any changing historical patterns may be important for the design of early intervention for injecting drug use, but also child protection services more generally, as well as for legal and policy frameworks for harm reduction services, which are often designed for adults.

The Swedish context
There is currently no reliable population size estimate of PWID in Sweden. A 2012 estimate has placed the number at 8,000 (PHAS, 2015, pp. 10 & 11), based on 'people with substance use diagnoses combined with health care diagnoses typical of people who inject drugs' (p. 10). At that time, NSP had not opened nationwide and there was no national register. Since 2013, over 8,000 people have registered in the national needle exchange database. In 2019, a Delphi process estimated that the number of PWID may be 21,000 (Blach et al., 2021). As a consequence of the lack of an overall estimate, there is no estimate of the prevalence of injecting among under 18s, though it can reasonably be assumed to be a relatively small group. Annual, national school-based surveys, which are the main method of understanding prevalence of drug use among under 18s in Sweden, do not record the method of drug intake, including injecting (CAN, 2021).
In the past, Sweden's drug policy has been antagonistic to harm reduction, but this has recently been changing with attention to harm reduction in national policy, increased attention to overdose mortality, and an expansion of NSP nationally (Regeringen, 2021, pp. 59-73). Harm reduction approaches for those under the age of 18, however, are not currently prominent. In both national policy and local drug services targeting under 18s, there is a strong focus on early identification and prevention of any drug use (typically cannabis) and parental use (Regeringen, 2021, pp. 42-53;Socialstyrelsen, 2019;SKL, 2018). Young people defined as having problematic substance use prior to 18 are the responsibility of municipal social services, with a requirement for collaboration with the relevant regional health authority. It is mandatory for all professionals working with children to report high-risk behavior such as injection drug use to social services (Regeringen 2016), and this is in turn likely to result in compulsory institutional care, regulated by specific legislation (SFS 1990:52). All interventions targeting legal minors using drugs in Sweden are typically abstinence-oriented. There are no dedicated harm reduction interventions focused on children and adolescents (though some youth services may well bring harm reduction principles into their work). Harm reduction related to minors is not mentioned in official policy documents or guidance at national or local levels (Socialstyrelsen 2019; SKL 2018) other than to restrict the use of NSP to over 18s only (SFS 2006:323, 6 §).
On the other hand, child protection laws, as well as the UN Convention on the Rights of the Child (UNCRC), applies to legal minors. The UNCRC was incorporated in its entirety into Swedish law in 2018 (SFS 2018(SFS :1197. The UN Committee on the Rights of the Child has long recommended appropriate harm reduction services for children and adolescents who use drugs, both as part of their right to health (Art 24 of the Convention), but also their right to protection from drugs (Art 33) (Barrett, 2020, pp. 129-132). What this means in practice, of course, depends on local contexts and differing interpretations, not least as the two articles can be viewed as creating opposing positions (Barrett, 2020, pp. 103-111).

Materials and methods
The present study is part of a wider project studying injecting drug use among legal minors, their characteristics, treatment and support needs, as well as policy and practice in three high-income countries: Sweden, Switzerland, and Wales. Ethics approval for the research project was granted by the Swedish Ethical Review Authority (Etikprövningsmyndigheten) (case number 2021-03361).

Participants
A total sample and a sub-sample were used in the current study. The total sample (n = 8225) comprised adults at first entry into an NSP in Sweden. The mean age at entry was 37.6 (range 18-73) and 74% were male and 26% female. The majority (61%) were born in Sweden, though 26% were missing country of birth data. The sub-sample comprised only those who were age 25 or under at entry (n = 1265; 65% males and 36% females).

Data and measures
The data comes from the national quality register InfCare NSP and covers a 9-year period: 2013-2021. InfCare NSP was first introduced in Stockholm in 2013, but had imported some data from the Skåne region pertaining to the period 1986-2013. Thus, the national data was deemed to be more reliable from 2013 when all regions used the same system. InfCare is used at 23 of 25 NSP sites across Sweden, but nearly half the data pertains to Sweden's two largest cities: Stockholm and Gothenburg. The two sites that are not covered by InfCare are very small sites with low numbers of visitors. In 2021, InfCare NSP covered >95% of all NSP clients in Sweden, thus coverage is viewed as nationally representative (Lindqvist et al., 2021). To access the NSP, clients must by law provide proof of identity and the inclusion criteria are age ≥18 years (≥20 years before March 2017) and active injecting drug use, both at point of treatment entry.
The measures are drawn from InfCare's 34-item entry questionnaire, which is completed on initial registration to the NSP. The questionnaire includes: 1) static, historical variables, such as sex, minimum educational level, and age of drug debut and age of injection debut (the key exposure); 2) dynamic variables, such as accommodation, health situation, and injection practices. In the present study, presenting treatment needs are thus defined as those needs surveyed by the entry questionnaire, with dynamic treatment needs being those that are amenable to change at or around the point of treatment entry; and 3) previous contact with statutory control systems, such as the police or compulsory care, and support systems, such as other drug treatment services. These items provide both a historical description of the level of consumption of services, but also a picture of potential treatment needs.

Procedure
All first-time visits to the NSP are registered in InfCare NSP, using the Swedish unique personal identity number. At first admission, NSP staff administer the entry questionnaire (i.e. entrance to the register occurs the same day as entrance to the NSP).

Analytical plan
We firstly analyzed the prevalence of injecting before the age of 18 (IDU < 18) in the total sample. Data on age of injection debut were missing in 10% (n = 794) of cases. Comparison of cases with missing data to those with valid data on age of injection revealed no significant differences in terms of sex, educational level, or current accommodation or employment. Thus, the data was deemed Missing-At-Random (MAR). Additionally, eight cases with an injection debut age less than 10 years old were deemed to be data errors and coded as missing. Historical trends were analyzed using one-way ANOVA unweighted linear sum of squares. Using the total sample, we ran bivariate comparative analyses of the static/historical variables. For analysis of dynamic variables, we used the sub-sample of those who were age 25 or under at intake. This was to be able to determine whether the IDU < 18 group had different presenting treatment needs to those who began injecting as young adults (IDU ≥ 18). The dynamic variables give a relatively current picture of people's life situation prior to NSP entry and we wanted to minimize conflating age-related issues that may occur when using the total sample's presenting needs (i.e. with an average age at NSP entry of 37.6). By selecting a younger sub-sample, a comparative analysis can reveal differences that are developmentally proximal to injection initiation, which will be more relevant for understanding the differential effects of early injection behavior on presenting treatment needs. The variable debut drug included many response categories, so those used by less than 2% of the sample were collapsed into an 'Other' category. Accommodation types were also collapsed into a dichotomous 'Stable/unstable' variable to enable analysis, with own accommodation coded as stable but all other types (e.g. shelter, street, lodging, and treatment home) coded as unstable.

Prevalence and historical trends
The prevalence of IDU < 18 in the total dataset was 29% (n = 2158). By default, age of injection debut for the IDU < 18 group ranged from 10-17 years. The mean for this restricted age range was 15.3 (percentiles: 25 = 14, 50 = 16, 75 = 17). The mean age of injection debut for the total sample was 22.4 (s.d. 7.7 years). The proportion of IDU < 18 showed a significant linear decline from 35% in 2013 to 24% in 2021 (F = 40.2, df = 1, p < 0.01) (Figure 1). Similarly, the mean age of injection debut showed a significant linear increase in the same period, from 21.5 to 23.7 years old (F = 51.7, df = 1, p < 0.01). Analyzing injection debut by age cohorts 1 found a small, but significant difference (Brown-Forsythe = 49.5, p < 0.01), with older age cohorts having a lower mean age of injection debut (Figure 2).
The mean number of years between injection debut and NSP entry for the total sample was 24 for IDU < 18 and 11 for IDU ≥ 18 (t = 40.5, df = 3417, p < 0.01). The mean for each year in the period 2013-2021 for IDU < 18 was between 22 and 26 years, showing no clear linear trend (F = 0.58, df = 1, p = 0.45). For the sub-sample, the mean number of years between injection debut and NSP entry was 6.3 for IDU < 18 and 2.0 for IDU ≥ 18 (t = 31.01, df = 1172, p < 0.01), giving an additional 4.3 years before accessing NSP services.

Comparison of static/historical variables
Bivariate analysis of the static/historical variables found statistically significant differences between IDU < 18 and IDU ≥ 18 in minimum education level (p < 0.01), debut drug (p < 0.01), and debut injection drug (p < 0.01), but not sex ( Table 1).
The largest differences in minimum education levels concerned statutory schooling (up to the age of 15/grade 9) and sixth-form college (gymnasium in Sweden, or high school for US readers). In particular, the IDU < 18 group who did not complete basic statutory schooling was twice the size of the IDU ≥ 18 group (p < 0.01). Concerning debut drug, a higher proportion (+7 percentage units) of the IDU < 18 group began with amphetamine (p < 0.01). There were no other statistically significant differences concerning debut drug between the two groups in terms of the static/historical variables. Amphetamine was also the debut injection drug of choice for the IDU < 18 group, with a higher proportion of the IDU ≥ 18 group beginning with heroin (p < 0.01).

Comparison of dynamic presenting treatment needs
Based on the sub-sample of those age 25 or under at entry, the dynamic or current presenting needs differed between IDU < 18 and IDU ≥ 18 regarding the presence of children at home, self-reported HCV positive, past month injection frequency and use of sterile equipment. Note that self-reported HCV status is not necessarily the same as actual status (Kåberg et al., 2017). Accommodation type, marital status, self-reported HIV status, and past year primary drug showed no differences (Table 2).
Whilst the presence of children is uncommon in both groups, the IDU < 18 group were more likely to have children (p < 0.001). In both groups, almost all parents were not living with their children. The prevalence of self-reported HCV among IDU < 18 was more than double the rate among IDU ≥ 18, while self-reported HIV status was very low/  non-existent in both groups. Interestingly, when running the analysis on the total sample, 1.9% of IDU ≥ 18 report being HIV positive, compared to 3.9% of IDU < 18 (χ2 = 36.8, df = 2, p < 0.01), though this may be capturing an older cohort who began injecting in the 1980s, before NSP were available and when HIV awareness was lower. The IDU < 18 group were more likely to inject several times a day and less likely to have used sterile injection equipment in the past month.

Comparison of contact with control and support systems
The IDU < 18 group had greater levels of contact with all the forms of statutory control surveyed, compared to the IDU ≥ 18 group (Table 3). Over half had been in secure residential children's homes, which is twice the proportion of the IDU ≥ 18 group (p < 0.01). Nearly half the IDU < 18 group had been arrested, nearly a third had experienced compulsory (secure) drug treatment, a quarter had been in prison, and nearly a fifth had been sectioned under mental health legislation.
Regarding support services, the IDU < 18 group had greater levels of contact with ADHD services and child services (p < 0.05), though actual proportions using these services were low (10% and 4%, respectively). There were no significant differences regarding other support services, except that the IDU < 18 group were less likely to be in contact with psychiatric out-patient services (p < 0.01).

Discussion
Our findings, based on a national dataset of over 8,000 individuals covering a 9-year period (2013-2021), indicated a mean age of injection debut of 22.4 years, approximately 1-2 years higher than reported by previous studies that are either local or based on more limited data (Britton et al., 2009;Stenström, 2008), and 3 years above an estimate by the Public Health Agency of Sweden (PHAS, 2015). Moreover, we have for the first time shown a linear increase in the mean age of injection debut over time. On the face of it this appears to be a positive development in that earlier injecting is associated with more risky practices. An alternative nb. percentage differences greater than 5% are shown in bold. Some percentages may not add up to 100 due to rounding. *p < 0.05. **p < 0.01. a although the expected cell count for this level is below 5, we retain all three levels for demonstrative purposes. We also ran the same analysis on the total sample where low expected cell count was not experienced and found similar proportions for this level of the variable, also with a significant result (χ2 = 51.7, df = 2, p < 0.01). nb. percentage differences greater than 5% are shown in bold. Some percentages may not add up to 100 due to rounding. *p < 0.05. **p < 0.01. explanation is that age at treatment entry is confounding. As there were very few NSPs available pre-2013, making access virtually impossible in many regions, there may have been an initial increase in numbers when NSP implementation increased nationally after 2013 comprising older cohorts who had been 'waiting' for NSP access. These older cohorts may have had their injection debut at older ages and thus artificially increased the initial mean age of injection found at the start of the study period. However, our analysis by birth year cohorts found that the older age cohorts of people born in the 1940s-60s had lower or equivalent mean ages of injection debut, meaning that any initial influx of older people in 2013/14 is unlikely to have inflated the mean age of injection debut for the first part of the study period. Further research is needed to explore the reasons for this apparent increase in mean age of injection debut.
Earlier research using samples local to Stockholm found that 37% of people who inject reported beginning before the age of 18 (Britton et al., 2009), with a fifth beginning before or at the age of 16 (Kåberg et al., 2020). On a national level, the present study shows that 29% reported injection debut before the age of 18. The same percentage was found for the entire sample and the younger sub-sample, suggesting that this prevalence rate is applicable to a more recent cohort despite the increase in the mean age of injection debut since 2013. Thus, the lower prevalence rate in the present study is more likely to be due to the fact that our dataset was not limited to one city, rather because of using different time-periods. Further, our data show that among legal minors who inject, the mean age of injection debut is just over 15. At this age, they are 'children' for the purposes of Swedish law and policy (Family Code 1949:381;Social Services Act, 2001:453).
We sought to understand if there were differences in socio-demographic factors and presenting treatment needs between the IDU < 18 and IDU ≥ 18 groups. We hypothesized that IDU < 18 would have different socio-demographics reflecting different life circumstances at the point of injection initiation, and that IDU < 18s would have greater treatment needs, compared to IDU ≥ 18s. These hypotheses were at a general level not supported for all the socio-demographic and treatment variables. However, statistically significant differences were found that may be divided into three interconnected themes: social circumstances, health, and service consumption.
With regard to social circumstances, both parenthood and schooling showed significant differences. Very few parents within the sub-sample lived with their children, and in this regard there was no difference between IDU < 18 and IDU ≥ 18. However, IDU < 18 were twice as likely to have had children (15% v 8%). Previous research involving young injectors has called attention to risky sexual practices among this group Frajzyngier et al., 2007;Miller et al., 2006;Roy et al., 2007;Thurstone et al., 2013). This could explain earlier parenthood in this sub-population, but further research is needed in the Swedish context. Moreover, from the present data we do not know why children were not living with their parents. Children growing up in foster care tend, however, to face many struggles and challenges, such as grief, anger, guilt and shame (Schofield et al., 2011). This draws attention to the potential intergenerational effects of early injection debut, which may require further study.
Educational attainment is an important social determinant of health, affecting outcomes during the lifecourse. IDU < 18, however, are less likely to have completed either compulsory schooling (i.e. leaving school before 15) or to have moved on to 'gymnasium' (high school/sixth form college). Early educational drop-out has been found in previous studies with young injectors (Abelson et al., 2006). We cannot know the reasons for leaving school prematurely from registry data, but it should be stressed that initiation into injecting drug use at such a young age is likely to be caused by a complex range of psychosocial factors. Early school drop-out is potentially another symptom, alongside injecting drugs, of these complex issues, rather than drug using behaviors themselves being the proximate cause of leaving school.
In terms of health, a striking finding was that self-reported HCV prevalence was far higher among IDU < 18 in the sub-sample. This supports findings from other countries with higher HCV rates among younger injectors (Dalgard et al., 2009;Day et al., 2005;Miller et al., 2006). A plausible explanation for this can be derived from the remaining data, combined with legal and policy issues. IDU < 18 have, by definition, a longer injecting career (see also Miller et al., 2006). Among our sub-sample, IDU < 18 experienced over four additional years between injection debut and registry at NSP. The period prior to NSP entry is likely to entail riskier injecting drug use, e.g., where the risk of HCV infection is high. In addition, IDU < 18 reported more frequent daily injection and a higher likelihood of sharing injecting equipment, which is associated with higher HCV infection risk (Bryant, 2014;Fortier et al., 2021). In sum, these four additional years before first attending an NSP are likely to be a more intense and riskier injecting period for young injectors than for older counterparts, which highlights the need for harm reduction policies and practices for legal minors who inject.
The above has also important policy implications concerning access to and consumption of services. While longer exposure to riskier injecting behaviors can explain the higher HCV rates, this should not be seen in a vacuum from the legal and policy setting. Rather, policies concerning access to services are part of the risk environment within which drug-related harm occurs (Rhodes, 2002). That environment is affected by the fact that IDU < 18 are legal minors. Pharmacy sales of needles and syringes for other than medical conditions are not permitted in Sweden, with additional proofs required if a customer is under 20 (SFS 2012:595 4 §). Prior to March 2017, those under the age of 20 were not permitted to access NSP in Sweden. Since then, an age restriction of 18 has been in place (Barrett et al., 2022). Removing an age restriction does not guarantee younger people will attend low threshold services, as we see from low uptake of those aged 18-20 since the age restriction was lowered. But the existence of an age restriction guarantees they legally cannot. Moreover, anonymous access to NSP is not permitted in Sweden, as attendees must always verify themselves and their age. Under 18s attempting to attend an NSP would place a legal obligation upon staff to report to social services. This situation raises legal, practical and ethical challenges relating to the low threshold nature of NSP services, and the 'best interests' of under 18s who inject drugs (Barrett et al., 2022).
Conversely, the IDU < 18 group had higher levels of contact with all forms of statutory control, from arrest and prison and probation contacts, to compulsory psychiatric interventions, compulsory drug treatment and secure children's homes. All such control measures were prior to NSP entry. Our data show that later in life there is no significant difference in their access to drug treatment or social services. As these are NSP register data, by default IDU < 18 have continued to inject despite statutory control interventions. Their health outcomes were either no different to IDU ≥ 18 or worse (e.g. the higher prevalence of HCV), calling into question the longer-term efficacy and efficiency of these prior control measures. For those who begin injecting before 18, there may be opportunities for intervention that focus on preventing drug-related harms that are currently being missed. Further research is required into harm reduction and treatment and care pathways for IDU < 18.
The legal and policy context is again important in this regard. Those who are discovered to be injecting prior to 18 are very likely to be reported to social services for this behavior, with compulsory institutional care a strong possibility. In Swedish law, the best interests of the child are 'decisive' in decisions regarding placement in secure residential homes, where care is provided under the terms of the Care of Young Persons (Special Provisions) Act (LVU) (Lag 1990:52), and the detention of children should be used 'only as a measure of last resort' according to child rights legislation (Lag 2018:1197, Art 37(b)). Our results show that over half of IDU < 18 had been in secure children's homes, compared to a quarter of IDU ≥ 18. Longer-term outcomes for children in care, such as drop out-rates from school and healthcare, are far worse compared to the general population in Sweden (Randsalu & Laurel, 2018). Further research may explore assumptions regarding the best interests of the child when placing minors in compulsory care due to their drug use, related discretions among social workers, and the availability and appropriateness of alternatives.
From a prevention perspective, there is much to be gained from focusing on IDU < 18 as early as possible, both to prevent increased frequency of injecting and very risky injecting practices, as well as to encourage transition away from injecting as a method of intake. However, IDU < 18 are currently absent from policy frameworks in Sweden, except to exclude them from NSP (Lag 2006:323 6 §). Voluntary drug services for legal minors typically cater to the needs of socially established drug users, often cannabis users, and tend to exclude IDU < 18s due to their injecting behavior. Instead, they become the responsibility of child protection services, where the default intervention appears to be compulsory control measures, which the present study confirms. Injecting drug use at a very early age, however, is likely connected to wider psychosocial factors. Using a 'drug, set and setting' framework for the etiology of drug use (Zinberg, 1986), attention needs to be paid to not just to the drug itself, but also mental health (set) and social and physical environment factors (setting). If there are mental health and social factors driving initiation into injecting at a very early age, then these initiating factors may persist if left untreated, continuously affecting the developmental trajectory in negative ways. The lack of harm reduction resourcesor worse, the deleterious effect of punitive control methods-may exacerbate the developmental trajectory pushing the individual into worsening marginalization.
Some limitations with the present study should be noted. The analysis makes use of existing registry data and is thus limited to the questions posed in the NSP entry questionnaire. While the questionnaire covers a range of relevant areas, there may be other differences between the IDU < 18 and IDU ≥ 18 groups that are not covered. Future research may benefit from theory-informed prospective designs and the current results should be seen as an important part of the emerging picture. The use of registry data also excludes IDU who never contacted an NSP in the study period. This sub-group of IDU may have different health and social factors surrounding initiation and continuation of injecting drug use to those in the present study. Thus, caution should be applied in generalizing the results to the wider population of IDU. The data is also self-report and reliant on memory, which may mean some minor loss of detail particularly concerning older clients presenting at the NSP some years later. The use of the sub-sample of those presenting at the NSP age 25 or under helps minimize this issue.

Conclusions
The present study has shown that almost one third (29%) of people attending NSP in Sweden begin injecting before the age of 18. This raises important questions for both child protection services and harm reduction approaches for legal minors who inject, who still qualify as 'children' in a legal and policy sense. Confirming earlier work from other countries, our study has shown that there are important differences in terms of health and social outcomes between those who inject prior to 18 and those who begin injecting as adults. At present, however, there is no harm reduction approach or plan, at least in a Swedish context, relating to injection debut prior to 18 that can take into account these outcomes later in the life course, and the acute risks associated with early injecting. These factors together speak to the need for targeted interventions at the intersection of childhood and harm reduction (Barrett et al., 2022). As Degenhardt et al. (2008) noted, "Harm reduction strategies must adapt continually to target successive cohorts of younger IDU [injecting drug users], in ways that are meaningful and relevant to these groups" (p. 360).

Note
1. Note that those born in the 2000s (n = 57) excluded from this analysis to minimise confounding by age.

Declaration of interest
The authors report there are no competing interests to declare.