Methadone Maintenance Treatment Discontinuation Among Young People who use Opioids in Vancouver,Canada

Abstract

French

Objective

Retaining adolescents and young adults (AYA) in medications for opioid use disorder (MOUD), like methadone maintenance treatment (MMT), is critical to reducing toxic drug fatalities. This analysis sought to identify factors associated with MMT discontinuation among AYA.

Method

Data were derived from the At-Risk Youth Study, a prospective cohort study of street-involved AYA in Vancouver, Canada, between December 2005 and June 2018. Multivariable extended Cox regression identified factors associated with time to MMT discontinuation among AYA who recently initiated MMT. In subanalysis, multivariable extended Cox regression analysis identified factors associated with time to “actionable” MMT discontinuation, which could be addressed through policy changes.

Results

A total of 308 participants reported recent MMT during the study period. Participants were excluded if they reported MMT in the past 6 months at baseline and were retained in MMT (n = 94, 30.5%); were missing MMT status data (n = 43, 14.0%); or completed an MMT taper (n = 11, 3.6%). Of the remaining 160 participants who initiated MMT over the study period, 102 (63.8%) discontinued MMT accounting for 119 unique discontinuation events. In multivariable extended Cox regression, MMT discontinuation was positively associated with recent weekly crystal methamphetamine use (adjusted hazard ratio [AHR] = 1.67, 95% confidence interval [CI]: 1.19 to 2.35), but negatively associated with age of first “hard” drug use (per year older) (AHR = 0.95, 95% CI: 0.90 to 1.00) and female sex (AHR = 0.66, 95% CI: 0.44 to 0.99). In subanalysis, recent weekly crystal methamphetamine use (AHR = 4.61, 95% CI: 1.78 to 11.9) and weekly heroin or fentanyl use (AHR = 3.37, 95% CI: 1.21 to 9.38) were positively associated with “actionable” MMT discontinuation, while older age (AHR = 0.87, 95% CI: 0.76 to 0.99) was negatively associated.

Conclusions

Efforts to revise MMT programming; provide access to a range of MOUD, harm reduction, and treatments; and explore coprescribing stimulants to AYA with concurrent stimulant use may improve treatment retention and reduce toxic drug fatalities.

Keywords

Introduction

The toxic drug crisis has led to thousands of preventable deaths across North America.¹ This is particularly true for people with opioid use disorder (OUD), who are at a heightened risk of experiencing toxic drug poisoning.^2–4 Methadone maintenance treatment (MMT) has proven effective in treating OUD and reducing associated harms.^5,6 Hence, existing clinical guidelines in North America and in British Columbia (BC), where this study is situated, recommend the prescription of MMT to people who report the regular use of illicit opioids, including adolescents and young adults (AYA).^7,8

Existing evidence indicates that being retained in MMT is central to realizing positive treatment outcomes^9,10; however, examinations of MMT discontinuation are primarily restricted to adults,¹¹ while research among AYA primarily focuses on buprenorphine-naloxone retention. For example, one study demonstrated that AYA are less likely to be retained in buprenorphine-naloxone in comparison to adult populations.^12,13 Additionally, a retrospective chart review of AYA at outpatient pediatric medications for OUD (MOUD) clinic in the Northeast USA found that female sex, and a negative urine screen for tetrahydrocannabinol and opioid use, and positive urine screen for buprenorphine-naloxone after initiating buprenorphine-naloxone, were associated with being retained on buprenorphine-naloxone at follow-up.¹⁰ A randomized controlled trial was also conducted among AYA in the USA, where buprenorphine-naloxone discontinuation was found to be positively associated with baseline cocaine use, not being in an intimate relationship, having children, and having a single parent.¹⁴

To date, evidence among adult populations suggests people receiving MMT were more likely to be retained in treatment in comparison to other MOUD, while alluding to potential factors associated with MMT discontinuation. For example, findings from a 24-week randomized controlled trial among adults found that 74% of participants receiving MMT and 46% of participants receiving buprenorphine-naloxone were retained in the study by the end of the 24-week study period.¹⁵ Similarly, a randomized controlled trial among pregnant people between 6 and 30 weeks of gestation with OUD in the USA, Australia, and Canada that examined MOUD retention at the time of delivery reported higher retention in MMT (82%) when compared to buprenorphine (67%).¹⁶ Additionally, a study examining the association between OUD treatment goals and factors associated with their attainment among AYA in Canada found that those receiving MMT were more likely to decrease their dose 3 months after treatment initiation (odds ratio = 4.42, 95% confidence interval [CI]: 1.40 to 14.0) in comparison to those receiving buprenorphine.¹⁷ This finding is supported by evidence of a higher preference for MMT among AYA who are interested in initiating opioid agonist treatment (OAT), which may be explained by perceptions that it is more effective at addressing mental illness and physical pain.¹⁸

Within the study setting, Vancouver, BC, research among adults found self-reported Indigenous identity, recent incarceration, sex work, and injection heroin use were negatively associated with MMT use, while female sex, HIV-positive status, and crack cocaine smoking were positively associated.¹⁹ Additionally, one study examined the impacts of regulatory changes to the MMT formulation on substance use and other health outcomes among HIV-positive people, finding that a switch to a more concentrated Methadose™ formulation in January 2014 was associated with significant increases in heroin injection.²⁰ One study from Vietnam found that HIV-positivity, higher education, challenges managing life skills, and disclosing one's health conditions to their partner were negatively associated with methadone adherence, while older age, living with pain, and mental illness were positively associated with MMT retention.²¹ Lastly, one qualitative study examining how AYA navigate MOUD treatment in Vancouver, Canada identified health policy barriers to MMT retention, which included burdensome, required daily witnessed dispensation; stringent missed dose and reinitiation policies; limited or inaccessible take-home dosing; and recommended dosage levels that were insufficient in the context of the potent illicit drug supply.²²

Of the evidence examining MMT among AYA, there has been one retrospective chart review of AYA accessing MOUD in Sydney, Australia. This study found that participants who received MMT were retained for an average of 354 days versus 58 days in buprenorphine-naloxone,²³ suggesting that some AYA are better retained in MMT compared to buprenorphine-naloxone. In the context of the toxic drug crisis,²⁴ this knowledge gap highlights the need for additional research into ways to prevent MMT discontinuation among AYA.

Given existing guidelines support the use of MMT for AYA in the study setting⁷ our research sought to understand ways to improve MMT retention among AYA. This involved identifying reasons for and factors associated with time to MMT discontinuation among AYA who reported initiating MMT over the study period. Additionally, this study sought to examine factors associated with potentially “actionable” MMT discontinuation, which are reasons for MMT discontinuation that can be addressed through policy or guideline changes.

Materials & Methods

Data Source & Study Sample

Our study drew on data from the At-Risk Youth Study (ARYS), an open prospective cohort study of “street-involved” AYA between ages 14 and 26 who were homeless or accessing services intended for homeless AYA at the time of recruitment, which has been described previously.²⁵ The study commenced in October 2005, and street-based outreach and snowball sampling approaches were used by ARYS staff to recruit AYA from street-based settings and youth-serving agencies, and through peer networks. To be eligible individuals are required to report illicit drug use in the past 30 days (other than or in addition to cannabis use) and provide informed consent to participate. Upon recruitment, participants complete an interviewer-administered questionnaire and a follow-up interview at 6-month intervals. Participants also see a study nurse and complete a nurse-administered questionnaire, and have access to wound care, infectious disease testing, and nursing support. All participants who provided written informed consent were reimbursed 40 CAD for each study visit. This study received ethical approval from the University of British Columbia/Providence Health Care Research Ethics Board.

For our study, the analysis was restricted to the period between December 2005 and June 2018, which is the period in which data was available, and the sample was restricted to participants who reported any weekly or daily illicit opioid use in the last 6 months at any point over the study period. Existing clinical guidelines recommend the use of MOUD, including MMT, for all AYA who use opioids regularly given the increased presence of toxic, synthetic opioids within the illicit drug supply.⁷ Participants who responded affirmatively to the question “in the last 6 months, have you been in any kind of alcohol or drug treatment (including methadone/Methadose™ or Suboxone^®)?” and subsequently reported accessing MMT were eligible for inclusion in the analytic sample. Participants who, at their baseline interview, reported MMT enrolment in the last 6 months were excluded from the analysis due to the inability to deduce when these participants commenced MMT; however, participants were reincluded in the sample if they reinitiated MMT at any point over the study period. Additionally, participants were right-censored if they reported an MMT discontinuation event, but they were reincluded in the analysis if they subsequently reinitiated MMT and they could therefore contribute more than one initiation event. Participants were also right-censored if they were lost to follow-up or if they reported recent MMT at the end of the study period, hence we expected high censoring rates in this study. More specifically, we observed an 83.4% censoring rate in the primary analysis, which was positively associated with the female sex but negatively associated with older age, and a 93.6% censoring rate in the subanalysis, though the sample size was small, and no statistically significant differences were observed. Given the intentional use of censoring and that we do not suspect the study conditions impact participants’ decisions to initiate MMT, we assessed that censoring is likely to be noninformative.

Participants who provided an affirmative response to receiving MMT in the past 6 months at some point over the study follow-up period were asked the follow-up question of “are you currently on MMT?” Participants who responded affirmatively were considered retained, while participants who responded negatively were deemed to have discontinued MMT. Participants who reported that they were not currently on MMT were then asked to provide one or more reasons for ceasing MMT from a list or could provide an open-ended response; however, participants who discontinued MMT between study periods were not asked for their reasons for MMT discontinuation and therefore did not have the opportunity to provide a reason. Participants who reported that they had discontinued MMT due to treatment completion (e.g., physician-initiated tapering, switching to another MOUD) were excluded from the analysis, and the remaining reasons for MMT discontinuation were tabulated.

Study Variables

The primary outcome of interest was time to any MMT discontinuation while the secondary outcome was time to MMT discontinuation that could possibly be addressed through policy or guideline changes, which included: missing a dose or being taken off by a healthcare provider; having difficulty accessing a pharmacy; and experiencing cost or affordability barriers. Dependent variable selection was guided by the Risk Environment²⁶ and Patient-Centered Accessibility²⁷ conceptual frameworks and also drew from prior literature examining MMT discontinuation among AYA and adult populations. A number of variables hypothesized to be associated with the outcomes of interest were considered, and include: age (per year older); age of first “hard” drug use (per year older); sex (male vs. female); self-reported Indigenous identity, defined as First Nations,²⁸ Inuit,²⁹ or Métis peoples³⁰ (Indigenous vs. White), which may act as a proxy for the impacts of anti-Indigenous racism, socioeconomic status, and intergenerational trauma; other self-reported racialized or ethnic identities, defined as an affirmative response to being Black, South Asian, Chinese, Other Asian, Latinx, Middle Eastern, or other (Other vs. White); moderate to severe depression (yes vs. no), measured using the Centre for Epidemiological Studies Depression scale³¹; child welfare involvement (yes vs. no); and moderate to severe levels of the following five categories of childhood adverse events—sexual abuse (yes vs. no), physical abuse (yes vs. no), emotional abuse (yes vs. no), physical neglect (yes vs. no), and emotional neglect (yes vs. no)—measured using the Childhood Trauma Questionnaire.³² Other variables include MMT initiation period, defined as having initiated MMT before or after the introduction of Methadose™ in the study setting (≥2014 vs. ≤2013)³³; recently living in the Downtown Eastside (DTES) (in the last 6 months) (yes vs. no), which is an inner-city neighbourhood of Vancouver with high levels of poverty and illicit drug use; any injection drug use (yes vs. no); any and recent injection drug use (yes vs. no); nonfatal overdose, defined as an overdose or acute reaction following drug use (yes vs. no) in the last 6 months³⁴; recent daily illicit opioid use (yes vs. no), a combined variable that includes daily injection and noninjection heroin, fentanyl, or nonmedical prescription opioid use (NMPOU); recent weekly heroin or fentanyl use (yes vs. no); recent weekly NMPOU (yes vs. no); recent weekly cocaine use (yes vs. no); recent weekly crack cocaine use (yes vs. no); and recent weekly crystal methamphetamine use (yes vs. no). Additional factors included: recent employment (yes vs. no), defined as legal self-employment, regular or temporary work; recent homelessness, defined as sleeping on the street, having no fixed address, staying with friends or staying in a shelter or hostel (yes vs. no); recent incarceration, defined as being in detention, prison or jail overnight or longer in the past 6 months (yes vs. no); recent non-pharmacological addiction treatment access, defined as accessing a treatment centre, recovery house, counsellor, Narcotics Anonymous, Cocaine Anonymous, Alcoholics Anonymous, Self-Management and Recovery Training, or any other non-pharmacological treatments (yes vs. no); recent difficulty accessing addiction treatment services (yes vs. no); and, recent difficulty accessing health and social services (yes vs. no).

Statistical Analyses

A baseline analysis comparing participants who were retained in MMT to those who discontinued MMT was conducted. Pearson's χ² test was employed to compare binary variables and the Mann–Whitney U-test was used to make comparisons for continuous variables. A multivariable extended Cox regression model with time-dependent variables was used among participants who reported initiating MMT at any point over the study period. Given that participants could report multiple MMT initiation and reinitiation events, an extended Cox model allows the analysis to proceed without requiring that the proportional hazard assumption be met.³⁵

An initial examination of the bivariate associations between the explanatory variables and our outcome of interest was conducted, and then a full model was built that consisted of all variables that were significant at P < 0.10. A backwards model selection approach was used whereby explanatory variables with the largest P-values were removed until a reduced model with the lowest Akaike Information Criterion (AIC) was identified. The variance inflation factor was calculated to determine collinearity between variables, which was not present. Previous ARYS studies have described and utilized this approach.^36–38

A multivariable extended Cox regression subanalysis was also conducted to identify factors associated with time to potentially “actionable” MMT discontinuation events. This was defined as MMT discontinuation that could be addressed through policy or clinical guideline changes, and the subanalysis was restricted to participants who reported discontinuing MMT due to missing a dose or being taken off by a healthcare provider, having difficulty accessing a pharmacy, or experiencing cost or affordability barriers. The same model-building approach was employed as previously described and involved a backwards model selection approach until a reduced model with the lowest AIC was identified. All significance tests were two-sided at a significance level of P < 0.05, and statistical analyses were conducted using SAS software version 9.4 (SAS, Cary, NC). Study data is not publicly available; however, study data can be made available by the corresponding author upon reasonable request.

Results

Among 696 participants who reported any weekly opioid use within the past 6 months, 308 (44.3%) participants reported recent MMT during the study period. Participants who had initiated MMT prior to their baseline interview (n = 94, 30.5%), were missing data on their MMT status (n = 43, 14.0%), or who discontinued MMT due to treatment completion (i.e., tapering) (n = 11, 3.6%) were excluded from the analysis. Of the remaining 160 participants who initiated MMT at some point over the study period, 102 (63.8%) reported discontinuing MMT and contributed 119 unique MMT discontinuation events. The baseline descriptive statistics comparing participants who were retained in MMT to participants who discontinued MMT are provided in Table 1.

Table 1.

Baseline descriptive statistics of adolescents and young adults (AYA) who reported initiating methadone or Methadose maintenance treatment (MMT) stratified by ever reporting MMT discontinuation between December 2005 and June 2018 (n=160).

Characteristic	Total (%) (n = 160)	Discontinued MMT		p - value
Characteristic	Total (%) (n = 160)	Yes (%) (n = 102)	No (%) (n = 58)	p - value
Age (med, Q₁–Q₃)	24 (22 – 27)	24 (22 – 26)	25 (22 – 27)	0.448^j
Age first drug use (med, IQR)	14 (12 – 15)	13 (12 – 15)	14 (12 – 15)	0.208^j
MMT Initiation Period (≥2014 vs. ≤2013)	67 (41.9)	33 (32.4)	34 (58.6)	0.001
Sex (female vs. male)	58 (36.3)	32 (31.4)	26 (44.8)	0.089
Indigenous identity (vs. white)	38 (23.8)	26 (25.5)	12 (20.7)	0.410
Other race/ethnicity (vs. white)^a	11 (6.9)	8 (7.8)	3 (5.2)	0.596^k
Highschool education (vs. <highschool education)	55 (34.4)	29 (28.4)	26 (44.8)	0.036
Depression^b (yes vs. no)	65 (40.6)	37 (36.3)	28 (48.3)	0.137
Child welfare involvement (yes vs. no)	84 (52.5)	60 (58.8)	24 (41.4)	0.034
Sexual abuse^c (yes vs. no)	33 (20.6)	21 (20.6)	12 (20.7)	0.914
Physical abuse^c (yes vs. no)	47 (29.4)	33 (32.4)	14 (24.1)	0.337
Emotional abuse^c (yes vs. no)	74 (46.3)	51 (50.0)	23 (39.7)	0.239
Physical neglect^c (yes vs. no)	51 (31.9)	34 (33.3)	17 (29.3)	0.772
Emotional neglect^c (yes vs. no)	79 (49.4)	50 (49.0)	29 (50.0)	0.765
Living in the DTES^{d, e} (yes vs. no)	76 (47.5)	54 (52.9)	22 (37.9)	0.068
Injection drug use (yes vs. no)	142 (88.8)	88 (86.3)	54 (93.1)	0.297
Injection drug use^d (yes vs. no)	120 (75.0)	80 (78.4)	40 (69.0)	0.184
Non-fatal overdose^d (yes vs. no)	22 (13.8)	13 (12.7)	9 (15.5)	0.624
Daily illicit opioid use^d (yes vs. no)	92 (57.5)	62 (60.8)	30 (51.7)	0.265
Weekly heroin or fentanyl use^d (yes vs. no)	100 (62.5)	63 (61.8)	37 (63.8)	0.799
Weekly NMPO use^{d, f} (yes vs. no)	20 (12.5)	14 (13.7)	6 (10.3)	0.534
Weekly cocaine use^d (yes vs. no)	16 (10.0)	13 (12.7)	3 (5.2)	0.172^k
Weekly crack use^d (yes vs. no)	40 (25.0)	29 (28.4)	11 (19.0)	0.184
Weekly CM use^{d, g} (yes vs. no)	58 (36.3)	41 (40.2)	17 (29.3)	0.169
Employment^d (yes vs. no)	44 (27.5)	22 (21.6)	22 (37.9)	0.026
Homelessness^d (yes vs. no)	74 (46.3)	48 (47.1)	26 (44.8)	0.786
Incarceration^d (yes vs. no)	29 (18.1)	19 (18.6)	10 (17.2)	0.827
Non-pharmacological treatment^{d, h} (yes vs. no)	55 (34.4)	36 (35.3)	19 (32.8)	0.745
Difficulty accessing treatment^{d, h} (yes vs. no)	16 (10.0)	13 (12.7)	3 (5.2)	0.174^k
Difficulty accessing services^{d, i} (yes vs. no)	34 (21.3)	22 (21.6)	12 (20.7)	0.959

Other self-reported racialized or ethnic identities include Black, Latinx, and Middle Eastern

Refers to moderate or severe depression as measured by the Centre for Epidemiological Studies Depression Scale

Refers to child welfare involvement, moderate or severe childhood sexual abuse, physical abuse or neglect, and emotional abuse or neglect as measured by the subscales of the Childhood Trauma Questionnaire

Refers to activities in the last six months

Denotes the Downtown Eastside

Denotes non-medical prescription opioids

Denotes crystal methamphetamine

Refers to treatments for substance use disorders

ⁱ

Refers to health and social services

P-value is generated using Mann-Whitney U-test

P-value is generated using Fisher’s Exact Test because of small cell count

Of the MMT discontinuation events, a total of n = 33 (27.7%) included an assessment of the reason for discontinuation, while n = 86 (72.3%) events occurred between study follow-up periods and participants’ reasons for MMT discontinuation were not captured. The primary reasons for MMT discontinuation were missing doses or being taken off by a healthcare provider (n = 9), not wanting to take it anymore (n = 8) and having difficulty accessing the pharmacy (n = 5), with additional reasons listed in Table 2.

Table 2.

Reasons for MMT Discontinuation among a Prospective Cohort of Street-Involved AYA who Initiated and Discontinued MMT at Some Point Between December 2005 and June 2018 in Vancouver, Canada (n = 102, 119 Unique Discontinuation Events).

Reasons for MMT discontinuation (n = 119)^a	Total (n)
Missed doses or taken off by healthcare provider	9
Did not want to take it anymore	8
Difficulty accessing pharmacy	5
Side effects	5
Was not effective or could not find a preferred dose	4
Too restrictive	1
Cost or affordability	1

Note. AYA = adolescents and young adults; MMT = methadose maintenance treatment.

A total of 86 unique MMT discontinuation events occurred between study follow-up visits where participants were not asked to provide a reason for MMT discontinuation.

Multivariable extended Cox regression was then employed among participants who reported initiating MMT at some point over the study period and who conducted a subsequent follow-up interview (n = 137). As demonstrated in Table 3, results indicate that time to MMT discontinuation was positively associated with recent weekly crystal methamphetamine use (adjusted hazard ratio [AHR]: 1.67, 95% CI: 1.19 to 2.35). Conversely, older age of first “hard” drug use (AHR: 0.95, 95% CI: 0.90 to 1.00) and female sex (AHR = 0.66, 95% CI: 1.01 to 2.28) were negatively associated with MMT discontinuation, after adjusting for age (per year older) (AHR: 0.95, 95% CI: 0.90 to 1.00), MMT initiation period (≥2014 vs. ≤2013) (AHR = 1.10, 95% CI: 0.74 to 1.64), self-reported Indigenous identity (Indigenous vs. White) (AHR = 1.29, 95% CI: 0.88 to 1.89).

Table 3.

Extended Bivariate and Multivariable Cox Proportional Hazard Model of Factors Associated with Time to MMT Discontinuation among Street-Involved AYA who Report Initiating MMT Between December 2005 and June 2018 (n = 137).

	Unadjusted	Adjusted
Characteristic	Odds ratio (95% CI)	P-value	Odds Ratio (95% CI)	P-value
Age (per year older)	0.96 (0.91 to 1.01)	0.095	0.95 (0.90 to 1.00)	0.057
Age first drug use (per year older)	0.94 (0.90 to 0.99)	0.015	0.95 (0.90 to 1.00)	0.044
MMT Initiation Period (≥2014 vs. ≤2013)	1.13 (0.77 to 1.67)	0.095	1.10 (0.74 to 1.64)	0.626
Sex (female)	0.72 (0.60 to 1.06)	0.095	0.66 (0.44 to 0.99)	0.047
Indigenous identity (vs. White)	1.33 (0.92 to 1.93)	0.135	1.29 (0.88 to 1.89)	0.199
Other race/ethnicity (vs. White)^a	1.88 (1.04 to 3.39)	0.036	1.80 (1.00 to 3.24)	0.051
Highschool education	0.70 (0.48 to 1.04)	0.075
Depression^b	0.67 (0.48 to 0.94)	0.021
Child welfare involvement	1.27 (0.91 to 1.77)	0.158
Sexual abuse^c	1.04 (0.65 to 1.65)	0.883
Physical abuse^c	1.28 (0.90 to 1.82)	0.168
Emotional abuse^c	0.90 (0.64 to 1.27)	0.539
Physical neglect^c	1.40 (0.96 to 2.05)	0.078
Emotional neglect^c	0.89 (0.63 to 1.26)	0.511
Living in the DTES^d,e,f	1.18 (0.83 to 1.69)	0.348
Injection drug use	0.79 (0.56 to 1.10)	0.162
Injection drug use^d,e	1.01 (0.73 to 1.40)	0.939
Nonfatal overdose^d,e	1.17 (0.71 to 1.93)	0.545
Daily illicit opioid use^d,e	1.08 (0.76 to 1.55)	0.665
Weekly heroin or fentanyl use^d,e	1.19 (0.83 to 1.69)	0.342
Weekly NMPOU^d,e,g	0.67 (0.34 to 1.34)	0.259
Weekly cocaine use^d,e	0.99 (0.61 to 1.59)	0.958
Weekly crack use^d,e	0.82 (0.54 to 1.25)	0.355
Weekly CM use^d,e,h	1.47 (1.04 to 2.07)	0.031	1.67 (1.19 to 2.35)	0.003
Employment^d,e	0.99 (0.63 to 1.54)	0.959
Homelessness^d,e	0.83 (0.57 to 1.21)	0.324
Incarceration^d,e	1.26 (0.78 to 2.03)	0.341
Nonpharmacological treatment^d,e,i	1.03 (0.71 to 1.50)	0.872
Difficulty accessing treatment^d,e,i	1.02 (0.52 to 2.00)	0.946
Difficulty accessing services^d,e,j	1.04 (0.67 to 1.60)	0.871

Note. AYA = adolescents and young adults; MMT = methadose maintenance treatment; NMPOU = nonmedical prescription opioid use; DTES = Downtown Eastside; CM = crystal methamphetamine.

Other self-reported racialized or ethnic identities include Black, Latinx, and Middle Eastern.

Refers to moderate or severe depression as measured by the Centre for Epidemiological Studies Depression Scale.

Refers to activities in the last 6 months.

Refers to variables that were lagged to the previous study visit.

Denotes the DTES.

Denotes nonmedical prescription opioids.

Denotes CM.

ⁱ

Refers to treatments for substance use disorders.

Refers to health and social services.

Results from the subanalysis of factors associated with time to “actionable” MMT discontinuation are presented in Table 4. Findings indicate that time to “actionable” MMT discontinuation was positively associated with recent regular crystal methamphetamine use (AHR: 3.79, 95% CI: 1.46 to 9.81) and recent regular heroin or fentanyl use (AHR = 3.37, 95% CI: 1.21 to 9.38). However, older age (AHR = 0.87, 95% CI: 0.76 to 0.99) was negatively associated with time to “actionable” MMT discontinuation, after adjusting for MMT initiation period (≥2014 vs. ≤2013) (AHR = 1.84, 95% CI: 0.63 to 5.35), non-pharmacological treatment access (AHR = 1.81, 95% CI: 0.72 to 4.56), self-reported racialized or ethnic identity (Other vs. White) (AHR = 1.77, 95% CI: 0.54 to 5.72), self-reported Indigenous identity (Indigenous vs. White) (AHR = 0.93, 95% CI: 0.31 to 2.76), and ever any injection drug use (AHR = 0.39, 95% CI: 0.11 to 1.38).

Table 4.

Bivariate and Multivariable Extended Cox Proportional Hazard Model of Factors Associated with Time to MMT Discontinuation That Could be Addressed Through Policy Change among Street-Involved AYA who Initiated MMT Between December 2005 and June 2018 (n = 62).

	Unadjusted		Adjusted
Characteristic	Odds ratio (95% CI)	P-value	Odds ratio (95% CI)	P-value
Age (per year older)	0.89 (0.80 to 0.99)	0.038	0.87 (0.76 to 0.99)	0.031
Age first drug use (per year older)	0.82 (0.71 to 0.95)	0.010
MMT Initiation Period (≥2014 vs. ≤2013)	2.31 (0.88 to 6.09)	0.090	1.84 (0.63 to 5.35)	0.262
Sex (female)	0.73 (0.30 to 1.79)	0.490	0.57 (0.22 to 1.59)	0.241
Indigenous identity (vs. White)	1.30 (0.56 to 3.32)	0.548	0.93 (0.31 to 2.76)	0.897
Other race/ethnicity (vs. White)^a	2.21 (0.76 to 6.40)	0.145	1.77 (0.54 to 5.72)	0.343
Highschool education	0.78 (0.33 to 1.82)	0.567
Depression^b	0.86 (0.38 to 1.97)	0.721
Child welfare involvement	1.65 (0.74 to 3.67)	0.219
Sexual abuse^c	0.67 (0.21 to 2.11)	0.494
Physical abuse^c	1.87 (0.87 to 4.02)	0.109
Emotional abuse^c	0.94 (0.39 to 2.27)	0.898
Physical neglect^c	2.57 (1.15 to 5.77)	0.022*
Emotional neglect^c	1.33 (0.58 to 3.08)	0.499
Living in the DTES^d,e,f	0.83 (0.38 to 1.82)	0.636
Injection drug use	0.37 (0.15 to 0.91)	0.031	0.39 (0.11 to 1.38)	0.145
Injection drug use^d,e	1.86 (0.69 to 5.04)	0.220
Nonfatal overdose^d,e	1.80 (0.68 to 4.72)	0.234
Daily illicit opioid use^d,e	2.11 (0.87 to 5.11)	0.097
Weekly heroin or fentanyl use^d,e	4.14 (1.56 to 11.0)	0.004	3.37 (1.21 to 9.38)	0.020
Weekly NMPOU^d,e,g	1.80 (0.49 to 6.57)	0.376
Weekly cocaine use^d,e	0.52 (0.07 to 3.84)	0.522
Weekly crack use^d,e	0.21 (0.03 to 1.48)	0.117
Weekly CM use^d,e,h	3.47 (1.50 to 8.05)	0.004	4.61 (1.78 to 11.9)	0.002
Employment^d	0.93 (0.37 to 2.32)	0.871
Homelessness^d	1.89 (0.87 to 4.14)	0.110
Incarceration^d	2.15 (0.76 to 6.06)	0.148
Nonpharmacological treatment^d,i	1.99 (0.91 to 4.36)	0.085	1.81 (0.72 to 4.56)	0.205
Difficulty accessing treatment^d,i	0.71 (0.09 to 5.76)	0.752
Difficulty accessing services^a,j	0.87 (0.32 to 2.38)	0.781

Note. AYA = adolescents and young adults; MMT = methadose maintenance treatment; NMPOU = nonmedical prescription opioid use; DTES = Downtown Eastside; CM = crystal methamphetamine.

Other self-reported racialized or ethnic groups include Black, Latinx, and Middle Eastern.

Refers to moderate or severe depression as measured by the Centre for Epidemiological Studies Depression Scale.

Refers to activities in the last 6 months.

Refers to variables that were lagged in the previous study visit.

Denotes the DTES.

Denotes NMPOU.

Denotes CM.

ⁱ

Refers to treatments for substance use disorders.

Refers to health and social services.

Discussion

Our findings demonstrate that over two-thirds of AYA in the study sample who initiated MMT subsequently discontinued treatment. Given evidence that AYA are retained in MMT for longer durations than other MOUD,^9,23 the low levels of MMT retention observed in the study setting are concerning. This may be explained by AYAs’ views on OAT as a short-term intervention to support their immediate transition off of illicit opioids, as reported in recent qualitative research in the study setting.¹⁸

Although a majority of participants were not asked for their reason for discontinuing MMT, the reasons for MMT discontinuation described by a subsample of participants provide some insight into ways that treatment services may be improved to better meet the needs of AYA. For example, the most common reasons for MMT discontinuation—such as missing doses, being taken off by a care provider, or having difficulty accessing a pharmacy—highlight potential policy and clinical changes that have potential to prevent discontinuation. In the study setting, clinical guidelines stipulate that missing three consecutive doses of MMT requires patients to reinitiate MMT at a lower dose.³⁹ Although this is appropriate for individuals who have been abstaining or reducing their overall consumption of opioids, for AYA who missed MMT doses because they were consuming illicit street opioids—which are characterized in the study setting by highly potent, synthetic opioids—reinitiating MMT at a lower dose is likely insufficient to meet their physical dependence and may contribute to MMT discontinuation.

The high levels of treatment discontinuation also point to the need for research into ways to improve MMT retention, including through improvements to existing clinical guidelines. Our finding that missing a dose or being removed by a healthcare provider was the most commonly reported reason for MMT discontinuation reinforces previous findings that AYA view the required daily witnessed dispensation of MMT as overburdening and constraining.¹⁸ To address this, there have been suggestions that take-home dosing be expanded for AYA when able to be done safely¹⁸; however, existing clinical guidelines limit take-home dosing. Given that existing guidelines around missed doses and reinduction were designed at a time when highly potent synthetic opioids were not widespread within the illicit drug supply, it is important that these guidelines are reviewed in the current context—and in light of interim Risk Mitigation Guidelines that improved access to take-home dosing and daily OAT delivery services to promote social distancing in the study setting^40,41—to improve MMT retention among AYA.

In addition to the reasons for MMT discontinuation provided, the analyses of factors associated with MMT discontinuation highlight additional ways to improve retention among AYA. Firstly, our finding that female AYA were less likely to discontinue MMT in comparison to male AYA suggests sex- and gender-based differences in treatment retention. This is complemented by evidence that adult women, and especially pregnant people,⁴² are retained in MMT for longer durations than men.⁴³ Given trauma-informed perinatal programming, which prescribes MMT, is accessible within the study setting,^44–47 the negative association between female sex and MMT discontinuation may reflect the availability of these supports. Alternatively, evidence has found that peer networks significantly influence male AYA substance use treatment decision-making compared to female AYA,⁴⁸ reiterating the importance of engaging with peer networks to identify ways to improve MMT retention.^49–52 Nevertheless, an AYA-specific gender-based analysis on MMT retention may highlight ways to address gender-based gaps in MMT retention among AYA.

Another important finding was the association between recent crystal methamphetamine use and time to “actionable” MMT discontinuation. This finding may be due to the increased use of stimulants in order to counteract the negative side effects of MMT (i.e., lethargy, tiredeness, etc.), due to participants’ seeking pleasure, or because of the high levels of polysubstance use reported among AYA in comparison to adult populations.⁵³ Study findings support exploration of novel treatment approaches to address simultaneous opioid and stimulant use among AYA. A potential intervention that has demonstrated some promising results among adults who report polysubstance use is the dual prescription of MOUD alongside prescription stimulants. To date, preliminary research has found that when adequately dosed, the provision of stimulants alongside MOUD reduces both illicit opioid and stimulant use.^54–57 Given the potential benefits of prescribing both prescription opioids and stimulants to AYA, future research investigating the use of both types of medications and their impact on the health of AYA may be beneficial.

The positive association between recent weekly fentanyl or heroin use and MMT discontinuation also provides some insight into ways to potentially improve the retention of AYA in treatment. Firstly, this may be due to AYA receiving an insufficient MMT dose and so increasing their dosage may support reductions in illicit opioid use. Alternatively, AYA who continue to use illicit opioids alongside MMT may benefit from being offered alternative types of OAT, such as slow-release oral morphine or injectable OAT, to reduce their reliance on the illicit drug supply. This is an important step to effectively address physical dependency and reduce overdose risk,^58–60 especially in the context of an increasingly toxic drug supply.

Lastly, the non-statistically significant, positive association between nonpharmacological treatment access and “actionable” MMT discontinuation is notable and raises concerns that some AYA may be discouraged from continuing MMT when accessing other treatment services. Adding to this, our findings suggest that younger participants are more likely to report “actionable” MMT discontinuation. This similarly raises concerns of a specific bias or stigma against prescribing MOUD to adolescents versus young adults, which is reinforced by evidence that adolescent-tailored treatment facilities are significantly less likely to provide MOUD to AYA in comparison to adult facilities.^61–63 Study results signal the importance of further research into stigma towards MMT and other MOUD within nonpharmacological treatment settings, especially among adolescent populations. Additionally, there may be benefits to implementing protections to ensure access to MOUD for all AYA to address stigma-related barriers to treatment access and continuation.

There are several limitations with the ARYS cohort that have been described previously, including generalizability, unmeasured confounding, and social desirability bias.^64–66 This analysis is additionally limited by the study questionnaire, which did not capture participants’ reasons for MMT discontinuation if they discontinued MMT between study follow-up periods. Hence, certain reasons for MMT discontinuation, such as being imprisoned, are expected to be underrepresented in the reasons provided and in the statistical analyses. Conversely, participants could have discontinued MMT between study follow-up periods due to treatment completion (i.e., tapered off of MMT), which could have led to the inclusion of “treatment completers” within the analytic sample and minimized or conflated the observed associations or they could have discontinued and reinitiated MMT in the previous 6 months but still have been considered retained in MMT as the questionnaire did not discern this. Another limitation is that some participants may have received both methadone and Methadose™ due to changes MMT provision during the study period and we were therefore unable to account for dosage in this analysis. The subanalysis is also limited by the small sample size and findings should be interpreted with caution, but this analysis meets sample size requirements^67–69 and provides preliminary insight into ways to address “actionable” MMT discontinuation. Lastly, the study period for these analyses was restricted to the period before the adoption of clinical guidelines for the treatment of OUD among AYA in the study setting, and so future research investigating the impacts of those clinical guidelines on MMT retention would be valuable.

Conclusion

Consistent with previous research among adults and AYA, this study demonstrated high levels of MMT discontinuation among the study sample, suggesting MMT may be a limited response in the context of a toxic drug crisis. While study findings point to potential opportunities to optimize MMT programming through measures such as providing access to a broader range of MOUD and treatment services and exploring the potential benefits of providing prescription stimulants to AYA who report polysubstance use, a key implication is that MMT is not sufficient for many AYA. Innovative interventions and ones that center harm reduction remain critical to protecting the health of AYA.

Footnotes

Acknowledgements

This research was conducted on the traditional territories of the xʷməθkʷəy̓əm, Sḵwxwú7mesh, and Səl̓ílwətaʔ Nations. The authors would like to thank ARYS participants for their involvement in this research, as well as all previous and current research staff supporting this study. We would also like to thank Lizzy Ambler for supporting this research.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Canadian Institutes of Health Research, National Institutes of Health, University of British Columbia, Michael Smith Foundation for Health Research (grant number MOP-286532, New Investigator Award, Postdoctoral Fellowship Award, U01-DA021525, Four-Year Doctoral Fellowship, Career Scholar Award).

ORCID iDs

Andreas Pilarinos

Kora DeBeck PhD

References

Donroe

Socias

Marshall

. The deepening opioid crisis in North America: historical context and current solutions. Curr Addict Rep. 2018;5(4):454-463.

Lyons

Yule

Schiff

Bagley

Wilens

. Risk factors for drug overdose in young people: a systematic review of the literature. J Child Adolesc Psychopharmacol. 2019;29(7):487-497.

Park

Lin

Hosanagar

Kogowski

Paige

Bohnert

. Understanding risk factors for opioid overdose in clinical populations to inform treatment and policy. J Addict Med. 2016;10(6):369-381.

Armoon

SoleimanvandiAzar

Rostami

, et al. Drug type and risk behaviors associated with non-fatal overdose among people who use drugs: a systematic review and meta-analysis. J Addict Dis. 2022;40(1):114-125.

Mariolis

Bosse

Martin

Wilson

Chiodo

. A systematic review of the effectiveness of buprenorphine for opioid use disorder compared to other treatments: implications for research and practice. J Addict Res Ther. 2019;10(379):2.

Hadland

Yule

Levy

Hallett

Silverstein

Bagley

. Evidence-based treatment of young adults with substance use disorders. Pediatrics. 2021;147(Supplement 2):S204-S214.

British Columbia Centre on Substance Use. A guideline for the clinical management of opioid use disorder: Youth supplement. June 31, 2018 ed. 2018.

Committee On Substance Use Prevention. Medication-assisted treatment of adolescents with opioid use disorders. Pediatrics. 2016;138(3):9.

Hadland

Bagley

Rodean

, et al. Receipt of timely addiction treatment and association of early medication treatment with retention in care among youths with opioid use disorder. JAMA Pediatr. 2018;172(11):1029-1037.

10.

Matson

Hobson

Abdel-Rasoul

Bonny

. A retrospective study of retention of opioid-dependent adolescents and young adults in an outpatient buprenorphine/naloxone clinic. J Addict Med. 2014;8(3):176-182.

11.

Timko

Schultz

Cucciare

Vittorio

Garrison-Diehn

. Retention in medication-assisted treatment for opiate dependence: a systematic review. J Addict Dis. 2016;35(1):22-35.

12.

Schuman-Olivier

Weiss

Hoeppner

Borodovsky

Albanese

. Emerging adult age status predicts poor buprenorphine treatment retention. J Subst Abuse Treat. 2014;47(3):202-212.

13.

Dreifuss

Griffin

Frost

, et al. Patient characteristics associated with buprenorphine/naloxone treatment outcome for prescription opioid dependence: results from a multisite study. Drug Alcohol Depend. 2013;131(1–2):112-118.

14.

Warden

Subramaniam

Carmody

, et al. Predictors of attrition with buprenorphine/naloxone treatment in opioid dependent youth. Addict Behav. 2012;37(9):1046-1053.

15.

Woody

Bruce

Korthuis

, et al. Hiv risk reduction with buprenorphine-naloxone or methadone: findings from a randomized trial. J Acquir Immune Defic Syndr. 2014;66(3):288.

16.

Jones

Kaltenbach

Heil

, et al. Neonatal abstinence syndrome after methadone or buprenorphine exposure. N Engl J Med. 2010;363(24):2320-2331.

17.

Chai

Rosic

Panesar

, et al. Patient-reported goals of youths in Canada receiving medication-assisted treatment for opioid use disorder. JAMA Netw Open. 2021;4(8):e2119600-e2119600.

18.

Giang

Thulien

McNeil

Sedgemore

Anderson

Fast

. Opioid agonist therapy trajectories among street entrenched youth in the context of a public health crisis. SSM Popul Health. 2020;11:100609.

19.

Kerr

Marsh

Montaner

Wood

. Factors associated with methadone maintenance therapy use among a cohort of polysubstance using injection drug users in Vancouver. Drug Alcohol Depend. 2005;80(3):329-335.

20.

Socías

Wood

McNeil

, et al. Unintended impacts of regulatory changes to British Columbia methadone maintenance program on addiction and HIV-related outcomes: an interrupted time series analysis. Int J Drug Policy. 2017;45:1-8.

21.

Tran

Nguyen

Tran

Latkin

. Social and structural barriers for adherence to methadone maintenance treatment among Vietnamese opioid dependence patients. PLoS One. 2018;13(1):e0190941.

22.

Pilarinos

Kwa

Joe

, et al. Navigating opioid agonist therapy among young people who use illicit opioids in Vancouver, Canada. Int J Drug Policy. 2022;107:103773.

23.

Bell

Mutch

. Treatment retention in adolescent patients treated with methadone or buprenorphine for opioid dependence: a file review. Drug Alcohol Rev. 2006;25(2):167-171.

24.

Woody

Poole

Subramaniam

, et al. Extended vs short-term buprenorphine-naloxone for treatment of opioid-addicted youth: a randomized trial. JAMA. 2008;300(17):2003-2011.

25.

Wood

Stoltz

J-A

Montaner

Kerr

. Evaluating methamphetamine use and risks of injection initiation among street youth: the ARYS study. Harm Reduct J. 2006;3(1):1.

26.

Rhodes

. The ‘risk environment’: a framework for understanding and reducing drug-related harm. Int J Drug Policy. 2002;13(2):85-94.

27.

Levesque

J-F

Harris

Russell

. Patient-centred access to health care: conceptualising access at the interface of health systems and populations. Int J Equity Health. 2013;12(1):1-9.

28.

Gadacz

Parrott

Gallant

. First nations. The Canadian Encyclopedia. Historica Canada. 2022.

29.

Freeman

Pedersen

Parrot

Gallant

. Inuit. The Canadian Encyclopedia. Historica Canada. 2020.

30.

Gaudry

Welch

. Métis. The Canadian Encyclopedia. Historica Canada. 2019.

31.

Radloff

. The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Meas. 1977;1(3):385-401.

32.

Bernstein

Fink

Handelsman

Foote

. Childhood trauma questionnaire. In Assessment of family violence: A handbook for researchers and practitioners. Washington, DC: American Psychological Association; 1998.

33.

Krebs

Hongdilokkul

Dale

, et al. The effect of a methadone reformulation on opioid agonist treatment outcomes: a population-based study in British Columbia, Canada, 2013–14. J Subst Abuse Treat. 2021:138:108714.

34.

Brugal

Barrio

Regidor

Royuela

Suelves

. Factors associated with non–fatal heroin overdose: assessing the effect of frequency and route of heroin administration. Addiction. 2002;97(3):319-327.

35.

Kleinbaum

Klein

. Survival analysis: A self-learning text. New York: Springer; 2012.

36.

Mittal

Jain

Sun

, et al. Opioid agonist treatment and the process of injection drug use initiation. Drug Alcohol Depend. 2019;197:354-360.

37.

Barker

Sedgemore

Tourangeau

, et al. Intergenerational trauma: the relationship between residential schools and the child welfare system among young people who use drugs in Vancouver, Canada. J Adolesc Health. 2019;65(2):248-254.

38.

Barker

Shoveller

Grant

Kerr

DeBeck

. Long-term benefits of providing transitional services to youth aging-out of the child welfare system: evidence from a cohort of young people who use drugs in Vancouver, Canada. Int J Drug Policy. 2020;85:102912.

39.

British Columbia Centre on Substance Use; British Columbia Ministry of Health. A guideline for the clinical management of opioid use disorder. 2017.

40.

British Columbia Centre on Substance Use, BC Ministry of Health, Ministry of Mental Health and Addictions. Risk mitigation in the context of dual health emergencies: interim clinical guidance: Update. 2022.

41.

Vancouver Coastal Health. Prescriber guidelines for risk mitigation in the context of dual public health emergencies. 2020.

42.

Khachikian

Amaro

Guerrero

Kong

Marsh

. Disparities in opioid treatment access and retention among women based on pregnancy status from 2006 to 2017. Drug Alcohol Depend Rep. 2022;2:100030.

43.

Marsh

Cao

D’Aunno

. Gender differences in the impact of comprehensive services in substance abuse treatment. J Subst Abuse Treat. 2004;27(4):289-300.

44.

Abrahams

Kelly

Payne

Thiessen

Mackintosh

Janssen

. Rooming-in compared with standard care for newborns of mothers using methadone or heroin. Can Fam Physician. 2007;53(10):1722-1730.

45.

Abrahams

MacKay-Dunn

Nevmerjitskaia

MacRae

Payne

Hodgson

. An evaluation of rooming-in among substance-exposed newborns in British Columbia. J Obstet Gynaecol Can. 2010;32(9):866-871.

46.

Garm

. The Sheway project [project for pregnant substance-using women]. Can Nurse. 1999;95(10):22.

47.

Benoit

Carroll

Chaudhry

. In search of a healing place: aboriginal women in Vancouver’s downtown eastside. Soc Sci Med. 2003;56(4):821-833.

48.

Brands

Leslie

Catz-Biro

. Heroin use and barriers to treatment in street-involved youth. Addict Res Theory. 2005;13(5):477-487.

49.

Shen

Assanangkornchai

Liu

, et al. Influence of social network on drug use among clients of methadone maintenance treatment centers in Kunming, China. PLoS One. 2018;13(7):e0200105.

50.

Kumar

Oles

Howell

, et al. The role of social network support in treatment outcomes for medication for opioid use disorder: a systematic review. J Subst Abuse Treat. 2021:108367.

51.

Sharp

Carlson

Howell

Moore

Schuman-Olivier

. Letting the sun shine on patient voices: perspectives about medications for opioid use disorder in florida. J Subst Abuse Treat. 2021;123:108247.

52.

Russell

Neufeld

Sabioni

, et al. Assessing service and treatment needs and barriers of youth who use illicit and non-medical prescription drugs in northern Ontario, Canada. PLoS One. 2019;14(12):e0225548.

53.

Corace

Willows

Schubert

Overington

Howell

. Youth require tailored treatment for opioid use and mental health problems: a comparison with adults. Can J Addict. 2018;9(4):15-24.

54.

Siefried

Acheson

Lintzeris

Ezard

. Pharmacological treatment of methamphetamine/amphetamine dependence: a systematic review. CNS Drugs. 2020;34(4):337-365.

55.

Nuijten

Blanken

van de Wetering

Nuijen

van den Brink

Hendriks

. Sustained-release dexamfetamine in the treatment of chronic cocaine-dependent patients on heroin-assisted treatment: a randomised, double-blind, placebo-controlled trial. The Lancet. 2016;387(10034):2226-2234.

56.

Palis

Harrison

MacDonald

Marsh

Schechter

Oviedo–Joekes

. Self–managing illicit stimulant use: a qualitative study with patients receiving injectable opioid agonist treatment. Drug Alcohol Rev. 2020;39(7):914-923.

57.

Brothers

Leaman

Bonn

, et al. Evaluation of an emergency safe supply drug and managed alcohol program in covid-19 isolation hotel shelters for people experiencing homelessness. medRxiv. 2022.

58.

Strang

Groshkova

Uchtenhagen

, et al. Heroin on trial: systematic review and meta-analysis of randomised trials of diamorphine-prescribing as treatment for refractory heroin addiction. Br J Psychiatry. 2015;207(1):5-14.

59.

Oviedo-Joekes

Guh

Brissette

, et al. Hydromorphone compared with diacetylmorphine for long-term opioid dependence: a randomized clinical trial. JAMA Psychiatry. 2016;73(5):447-455.

60.

Jegu

Gallini

Soler

Montastruc

Lapeyre–Mestre

. Slow–release oral morphine for opioid maintenance treatment: a systematic review. Br J Clin Pharmacol. 2011;71(6):832-843.

61.

Alinsky

Hadland

Matson

Cerda

Saloner

. Adolescent-serving addiction treatment facilities in the United States and the availability of medications for opioid use disorders. J Adolesc Health. 2020;66(2 Supplement):S12.

62.

Hadland

Jent

Alinsky

Marshall

BDL

Mauro

Cerda

. Opioid use disorder treatment facilities with programs for special populations. Am J Prev Med. 2020;59(3):e125-e133.

63.

Pilarinos

Bromberg

Karamouzian

. Access to medications for opioid use disorder and associated factors among adolescents and young adults: a systematic review. JAMA Pediatr. 2022;176(3):304-311.

64.

Pilarinos

Barker

Nosova

, et al. Coercion into addiction treatment and subsequent substance use patterns among people who use illicit drugs in Vancouver, Canada. Addiction. 2020;115(1):97-106.

65.

Cheng

Small

Dong

Nosova

Hayashi

DeBeck

. An age-based analysis of nonmedical prescription opioid use among people who use illegal drugs in Vancouver, Canada. Subst Abuse Treat Prev Policy. 2018;13:Article number 41.

66.

DeBeck

Kerr

Nolan

Dong

Montaner

Wood

. Inability to access addiction treatment predicts injection initiation among street-involved youth in a Canadian setting. Subst Abuse Treat Prev Policy. 2016;11(1):1-5.

67.

Concato

Peduzzi

Holford

Feinstein

. Importance of events per independent variable in proportional hazards analysis I. Background, goals, and general strategy. J Clin Epidemiol. 1995;48(12):1495-1501.

68.

Peduzzi

Concato

Feinstein

Holford

. Importance of events per independent variable in proportional hazards regression analysis II. Accuracy and precision of regression estimates. J Clin Epidemiol. 1995;48(12):1503-1510.

69.

Vittinghoff

McCulloch

. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol. 2007;165(6):710-718.