Review Adherence to and Retention in Medications for Opioid Use Disorder Among Adolescents and Young Adults

The volatile opioid epidemic is associated with higher levels of opioid use disorder (OUD) and negative health outcomes in adolescents and young adults. Medications for opioid use disorder (MOUD) demonstrate the best evidence for treating OUD. Adherence to and retention in MOUD, defined as continuous engagement in treatment, among adolescents and young adults, however, is incompletely understood. We examined the state of the literature regarding the association of age with adherence to and retention in MOUD using methadone, buprenorphine, or naltrexone among persons aged 10 – 24 years, along with related facilitators and barriers. All studies of MOUD were searched for that examined adherence, retention, or related concepts as an outcome variable and included adolescents or young adults. Search criteria generated 10,229 records; after removing duplicates and screening titles and abstracts, 587 studies were identified for full-text review. Ultimately, 52 articles met inclusion criteria for abstraction and 17 were selected for qualitative coding and analysis. Younger age was consistently associated with shorter retention, although the overall quality of included studies was low. Several factors at the individual, interpersonal, and institutional levels, such as concurrent substance use, MOUD adherence, family conflict, and MOUD dosage and flexibility, appeared to have roles in MOUD retention among adolescents and young adults. Ways MOUD providers can tailor treatment to increase retention


INTRODUCTION
Opioid misuse has quickly become a major public health concern globally. Most high-income countries already are experiencing epidemic rates of opioid misuse, and rates are increasing in low-and middle-income countries (LMIC) (1). According to the most recent Global Burden of Disease Study (2), opioid dependence is estimated to be the most common substance use disorder worldwide after alcohol use disorder and poses a considerable burden in the form of disability-adjusted life-years, responsible for 86,200 deaths and 3,656,100 years of life lost.
Opioid misuse is increasing among young people between the ages of 10 and 24 years, especially in high-income countries, mainly due to increasing availability and nonmedical use of prescription opioids (3,4). This population is especially vulnerable to negative consequences of opioid misuse, developing dependence faster than adults as a result of being in a life period critical for cognitive, emotional, and social development (5).
Medications for opioid use disorder (MOUD) are the most effective treatment for chronic opioid use disorder (OUD) and includes pharmacotherapies like methadone, buprenorphine, and naltrexone, all of which work on the μ opioid receptor (6). Both methadone and buprenorphine act as opioid agonists, activating the μ opioid receptor and alleviating opioid withdrawal. Unlike methadone, which is a full agonist, buprenorphine is a partial agonist, which means it has a ceiling effect, limiting it from producing the same effects as a full agonist and resulting in a lower likelihood of producing euphoria and of causing opioid overdose and other negative effects associated with opioid consumption. Naltrexone, on the other hand, is an opioid antagonist, which means it blocks the μ opioid receptor, preventing opioids from taking effect if used. As such, rather than alleviate opioid withdrawal, naltrexone functions to prevent relapse (7).
In industrialized and developing countries, MOUD consistently curtail the negative health, social, and legal consequences of OUD (8,9). More specifically, MOUD reduce opioid use and subsequent risk of HIV and hepatitis C transmission, while improving mental health and reducing drug-related and all-cause mortality (3,(9)(10)(11)(12)(13)(14). Despite their remarkable success, MOUD remain either absent or inadequately scaled to need in most countries (1,15,16). Despite recommendations of optimal MOUD coverage of 40%, with minimal effective coverage at 20%, as few as an estimated 16 persons per 100 persons who inject drugs receive MOUD globally, with high variability observed (1,16).
The evidence base demonstrating the effectiveness of MOUD among adolescents and young adults is limited and mostly focuses on buprenorphine (17); that said, multiple international agencies, professional groups, and other institutions, such as the World Health Organization and the American Academy of Pediatrics (18,19), recommend the use of MOUD for individuals in this age group. Despite these recommendations, many young people with OUD still lack access to MOUD (20)(21)(22). Even when they are able to access MOUD, young people still experience challenges with adherence to and retention in MOUD (23,24). As such, there is an urgent need to identify the factors causing young people to prematurely drop out of MOUD therapy.
In previous systematic reviews, authors have explored the factors associated with MOUD initiation and outcome among young persons (25,26) or have explored barriers to retention in MOUD-defined as continuous engagement in treatment-among the general population (27,28). None to our knowledge have examined specifically which factors are responsible for adherence to and retention in MOUD among young persons. In addition, a global perspective on this issue, which could highlight different approaches and their effectiveness depending on the social context, development status, and policies and procedures related to MOUD, is also lacking. Here, we review the research on the influence that age has on adherence to and retention in MOUD among adolescents and young adults globally.

METHODS
We conducted a systematic review of all studies of MOUD including evidence of adherence, retention, or related concepts (e.g., compliance, dropout, attrition) as an outcome variable and in which individuals between the ages of 10 and 24 years were part of the study sample. This age range was chosen because it is consistent with the definition of young people used by the World Health Organization and other international agencies (4). Adherence was defined primarily by continued consumption of MOUD. Adherence (also called compliance) is sometimes measured through attendance, especially in methadone programs where consumption is observed. It also may be confirmed through urinalysis, which can test for metabolites of MOUD. Retention was defined by uninterrupted engagement in treatment, operationalized in a variety of ways, such as by attendance at program appointments or renewal of prescriptions. According to this definition, retention could be defined as specific time points (e.g., at 3 months or 1 year) or as the time from treatment initiation to dropout. As may be evident from the latter measurement, retention in MOUD is also sometimes defined by its opposites: treatment attrition or dropout. These definitions were used to broadly capture the full scope of studies examining these concepts as treatment outcomes. Standard definitions for these terms in the context of MOUD were not found in other systematic reviews on the topic.
The methods used for this review are largely summarized in the International Prospective Register of Systematic Reviews protocol (29). A search strategy was tested against a set of validation articles, then used in 7 bibliographic databases: CINAHL, Criminal Justice Abstracts, Embase, Medline via Ovid, PsycINFO, Scopus, and Web of Science Core Collection. Similar search strategies were used in each database, adjusted for the subject headings available in each database's controlled vocabulary schema. The searches took place at between December 20, 2018, and January 3, 2019.
We searched for articles using title-abstract keywords and, where available, controlled vocabulary for 4 concepts: the population of adolescents and young adults, OUD, MOUD, and adherence or retention (Web Appendix) (available at https://academic.oup.com/epirev). In databases with subject indexing, animal studies that did not involve humans were excluded. No language or date restriction was applied.
After removing as many duplicate articles as possible, title/abstract review was conducted to identify potentially relevant studies. Inclusion criteria required articles to focus on MOUD; include data on adherence, retention, or related concepts as an outcome among individuals aged 10 to 24 years (or people identified by the authors as adolescents or young adults); be an original, peer-reviewed research article; provide original data (i.e., commentaries, editorials, letters, opinions, and reviews were excluded); and have been published in a peer-reviewed journal. No publication date restriction was applied and only studies published in Bosnian, Croatian, English, Italian, Montenegrin, Russian, Serbian, or Spanish were included in the final review. These languages were selected on the basis of the fluency of the members of the research team, which viewed including articles in as many languages as possible so important research studies could be included from non-English-speaking countries, given the international focus of this review. Covidence (Melbourne, Victoria, Australia) was used for both title and abstract and full-text review. Disagreements around inclusion of articles in the title and abstract or full-text review was resolved by a third member of the research team.
The research team systematically searched the gray literature, such as relevant government websites (e.g., National Institute on Drug Abuse, Substance Abuse Mental Health Services Administration, United Nations Office of Drugs and Crime), and provider organizations (e.g., American Academy of Pediatrics, American Society of Addiction Medicine) for any documents or data they may provide. Google Scholar was used to identify validation articles to ensure quality of the review.
From all relevant sources, the research team extracted data related to adherence to or retention in MOUD among young persons, as defined at the beginning of Methods, or data on the association between age and these outcomes. Information about the research study and covariates was also collected.
Two members of the research team were required to agree on study inclusion during both title and abstract review and full-text review. Members of the research team individually extracted data, which were double-checked by another member of the research team to ensure quality of data extraction. Two members of the research team conducted qualitative coding with a subset of the final articles that focused solely on adolescents and young adults; this coding was reviewed and analyzed by a third member of the research team, who developed themes on the basis of an adapted version of the Social Ecological Model (30), a framework for organizing into nested levels factors that affect behavior. This framework was selected because of the way it represents the relationships among the individual, interpersonal, institutional, and structural factors that may influence adherence to and retention in MOUD among adolescents and young adults. Quality assessment was conducted independently by a pair of researchers using the Grading of Recommen- dations Assessment, Development, and Evaluation system and guidelines (31,32). Citation analysis was conducted on April 8, 2019, on this final subset of 17 articles using Web of Science Core Collection to identify articles that cited those included articles. After deduplication of that set, 181 citations remained. Those citations were processed through title and abstract review and full-text screening. None of the studies identified met inclusion criteria and so no additional articles were added to the synthesis.

RESULTS
The initial search yielded 10,229 articles, of which 5,896 were immediately identified as duplicates. The title and abstracts of a total of 4,333 articles were screened against inclusion criteria, leaving 683 articles for full-text review. A total of 52 articles were included in qualitative synthesis after full-text review because the studies they reported included adolescents or young adults in the sample and examined the relationship between age and retention either continuously or in a categorical way in which the results specific to this age group could be examined. Seventeen studies were focused on adolescents and/or young adults specifically; the results from these studies were coded and analyzed qualitatively. Figure 1 presents the flow of studies through the review according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (33).
More than half of the included studies had been conducted in the United States (n = 30). The majority of the other studies were conducted in other high-income countries, including Australia (n = 4), Ireland (n = 3), Canada (n = 2), France (n = 2), Israel (n = 1), Italy (n = 1), Norway (n = 1), and Switzerland (n = 1). Only 7 studies were conducted in low-and middle-income countries (LMIC), including 3 studies in India and 1 each in Afghanistan, Tanzania, Turkey, and Ukraine.
More than a third of the included studies assessed adherence to or retention in buprenorphine only (n = 21) or methadone only (n = 16). Four studies assessed retention in treatment with naltrexone only, 2 with oral naltrexone, and 2 with injectable naltrexone. Some studies assessed both methadone and buprenorphine (n = 8), buprenorphine and injectable naltrexone (n = 1), methadone and oral naltrexone (n = 1), or all 3 together (n = 1).
No other studies were identified through additional gray literature searches. Information about the 52 articles included in qualitative synthesis is reported in Table 1.

Relationship between age and retention
Most of the 35 studies (n = 20; 57.1%) that examined age as a predictor of retention found that adolescents and young adults are at increased risk of dropout from MOUD treatment. Chaudhry et al. (34) did initially find younger age was associated with increased retention in MOUD; however, multivariate analysis determined this association was due to confounding variables. Cousins et al. (35) found younger age was associated with longer retention in MOUD even after adjusting for other variables, with individuals aged 40-65 years having 25% greater odds of discontinuing MOUD compared with individuals aged 16-19 years. Almost onethird of studies (n = 11, 31.4%) found age was not a statistically significant predictor of adherence or retention in MOUD among adolescents and young adults (34,(36)(37)(38)(39)(40)(41)(42)(43)(44)(45). All these studies either were conducted in LMIC or had a small sample size (largest sample size outside of studies done in LMIC, n = 981) (36).

Quality assessment
The quality of the articles exploring the association between age and retention in MOUD was low according to the guidelines outlined in the Grading of Recommendations Assessment, Development, and Evaluation system (32). This was because few of the studies in included articles were randomized controlled trials. Overall, the researchers who conducted quality assessment agreed that the studies, as a group, demonstrated low risk of overall bias and no evidence of publication bias. Indirectness was not seen as an issue in these studies. A moderate degree of inconsistency was observed across study findings, though it is notable that the studies in which no evidence was found of association between age and retention in MOUD were of lower quality and exhibited a greater degree of imprecision than the studies that observed a positive association between age and retention in MOUD, due largely to differences in sample sizes between the 2 sets of articles. Studies in which older age was associated with longer retention (n = 20) averaged a sample size of almost 5,000 participants, whereas the studies in which no significant association was observed between age and retention (n = 11) had an average sample size more than 7 times smaller, at 709 participants. As such, it is possible that some of the studies in which no association was found between age and retention did not have sufficient sample size to identify significant results.
In only 1 study was lower age correlated with increasing retention (35). Of the remaining 3 articles, 1 was focused on adherence as an outcome (46), 1 reported mixed results by sex (47), and the last reported mixed results on the basis of the time frame, with a positive association identified between age and retention at 6 months that was no longer significant at 12 months (48). None of these studies were included in either of the subgroups in quality assessment. The results of quality assessment are presented in Table 2.

Relationship between age and MOUD retention by MOUD type
To explore the relationship between age and retention in MOUD, we stratified studies by MOUD type (i.e., methadone, buprenorphine, naltrexone). After excluding the studies with mixed findings, described previously, we found a total of 20 studies had examined this relationship for methadone, 15 for buprenorphine, and 2 for naltrexone. The sum of the studies listed here is greater than the total number of studies examining the relationship between age     Abbreviations: Max., maximum; MOUD, medications for opioid use disorder; RCT, randomized controlled trial. a The time frame for analysis of the relationship between age and retention when multiple time frames are given for the length of follow-up. In cases where multiple time frames are included but none is highlighted with an asterisk, then survival analysis (usually Cox proportional hazards regression analysis) was conducted using the length of time from treatment initiation to dropout for each study participant. and retention in this subgroup analysis (n = 32) because 6 of these studies included more than 1 MOUD type, though D'Ippoliti et al. (49) only explored the relationship between age and retention for those patients treated with methadone and not for those receiving naltrexone.
Of the 20 studies in which the authors looked at the relationship between age and retention for individuals treated with methadone, 75.0% reported a positive association, with older age corresponding to a greater likelihood of retention in MOUD, though 1 study found a negative association (5.0%). Of the 15 studies exploring this relationship among individuals on buprenorphine, two-thirds (66.7%) found a positive association between age and retention. Finally, neither of the studies examining the relationship between age and retention in individuals treated with naltrexone found a significant association.

Relationship between age and MOUD retention by measurement of retention
The relationship between age and retention in MOUD was explored relative to the length of retention measured in the study to see if studies measuring retention over longer times found different results related to the association between age and retention. Of the 32 studies that did not present mixed results, one-third (34.4%; n = 11) measured retention using a time frame shorter than 12 months and one-quarter (25.0%; n = 8) measured retention using a time frame longer than 12 months; the remaining studies (40.6%; n = 13) measured retention using a time frame of exactly 12 months or its equivalent (e.g., 52 weeks).
Similar percentages of studies that measured retention using a time frame shorter than 12 months or of exactly 12 months found a positive association between age and MOUD retention (63.6% and 61.5%, respectively); the remaining studies found no significant association between age and retention in MOUD. A larger percentage of studies that measured retention using a time frame longer than 12 months found a positive association between age and MOUD retention (75.0%); the only study to find a negative association between age and MOUD retention also used a time frame longer than 12 months (35).

Relationship between age and retention by treatment of age
The relationship between age and retention was further explored to see if this relationship varied on the basis of whether age was treated as a continuous or categorical variable. Of the 28 studies included in this analysis, 16 (57.1%) treated age as a continuous variable whereas the remainder treated age categorically. Four studies were excluded because it was not clear from the authors' discussion of the analysis how age was treated; all but 1 of these studies found no significant association between age and retention.
Among those studies in which age was treated continuously, nearly two-thirds (62.5%) found a positive association between age and retention in MOUD. A larger percentage of studies that treated age categorically (83.3%) found a positive association between age and MOUD retention. The 1 study that found a negative association between age and MOUD retention also treated age categorically.

Qualitative coding and analysis
Qualitative coding and analysis were conducted with a subset of 17 studies that had data segregated by age for adolescents and young adults, either as defined by the author or based on the age range of 10-24 years. All these studies explored retention in MOUD among adolescents and young adults, so the coding focused on factors related to retention (to the exclusion of adherence). After coding, the findings from these studies were thematically organized using Bronfenbrenner's Social Ecological Model (30). A total of 16 codes were used, which were grouped into themes according to 3 levels: individual, interpersonal, or institutional; no factors at the community or societal levels were observed. Individual-level themes included age, adherence to MOUD (as a predictor of retention), demographics other than age (i.e., gender, race/ethnicity), drug use (past and current), physical health, psychiatric comorbidity, and treatment history for substance use disorders. Interpersonal themes encompassed relationships with others, including family members and intimate partners. Institutional themes focused on aspects of the MOUD program itself and included treatment modality, dose, structure, and timeliness (Figure 2). At the individual level, age was not consistently associated with retention when looking only at adolescents and young adults. Two studies found age was unrelated to retention (50, 51), 1 reported younger age and retention were negatively related (52), and another study found younger age and retention were positively related (53). These findings contrast with the earlier trend seen in studies with samples that included older adults, which consistently observed that age was positively related to retention. This may be due to greater homogeneity within the adolescent and youngadult age group or a nonlinear relationship between age and retention, though it is important to note that Vo et al. (51) compared participants younger than 24 years with a separate sample of individuals aged 24 years or older and found no significant association between age and retention.
Study results were similarly inconsistent when exploring the associations among race, gender, and retention. One study found no association between either race or gender and retention (53) and 2 other studies (51,54) presented opposite results with respect to the association between gender and retention. It is possible that these results are attributable to differences in the age ranges of participants, with participants in the study by Vo et al. (51) older than those in the other 2 studies; however, the type of medication used might have played more of a role in the difference in the reported results.
The research showed a consistent relationship between drug use and retention. In 3 studies (53-55), authors found that opioid use during treatment was negatively associated with retention; 1 of these studies (55) also found that hallucinogen use during treatment was negatively associated with retention. Two studies (54,55) found adherence to MOUD was positively related to retention. The research exploring the associations of physical health issues (50,55) and treatment history (50,56) with retention demonstrated mixed results, with 1 study finding these variables positively associated with retention and a second study finding a negative association between the 2. Finally, in 1 study (50), psychiatric comorbidity was positively associated with retention.
At the interpersonal level, stressful family situations, such as drug use among immediate family members (57), being raised in a single-parent household, or having a child (58), were all negatively associated with retention. On the other hand, being in an intimate relationship with someone with a history of heroin use was positively associated with retention (58).
At the institutional level, 3 studies found the type of medication used was unrelated to retention (51,58,59); however, this finding is in contrast to the results of 2 studies (20, 60) in which greater retention was reported among individuals in methadone-based treatment compared with individuals in buprenorphine-based treatment. Results were not consistent when looking at the relationship between treatment dose and retention. One study (50) found a higher dose of buprenorphine was associated with longer retention, whereas another (57) found it was negatively associated with retention. Finally, timely provision of treatment was also positively associated with retention (20).

DISCUSSION
This systematic review was conducted to assess the state of the literature on the factors associated with adherence and retention to MOUD among adolescents and young adults experiencing OUD, a population often identified as at heightened risk for dropout from such treatment. Many studies examining adherence or retention in MOUD that included adolescents and young adults did not account for the possible association with age. Of the 631 articles that were excluded after full-text review, one-third (33.1%) did not focus explicitly on adherence or retention among young persons, nor did they account for associations with age in their analysis while including adolescents or young adults in the total sample. Many of these studies represented age categorically in analyses, using age groups that were too broad for meaningful analysis of the relationship between adolescent or young adult status and retention in MOUD and too inconsistent for comparison of associations across studies. In other studies, the possible association with age was controlled for using statistical comparison of mean ages across groups. This approach, however, does not fully control for the possible association between age and retention, especially if the age range is skewed. Only in the study by Guillou Landreat et al. (46) did researchers look at adherence as an outcome.
In this review, we found that adolescents and young adults with OUD demonstrated heightened risk for dropout from MOUD in more than half of included articles. Although one-third of included studies did not find an association between younger age and increased risk of dropout from MOUD, the overall quality of these studies was lower. The studies in this group with larger sample sizes tended to be conducted in LMIC (37,42,44). This may point to the presence of potentially confounding contextual or structural factors specific to LMIC settings, such as cultural differences in attitudes toward substance use in younger age groups or different levels of acceptability or policy standards with respect to the use of MOUD among adolescents and/or young adults. Differences in the operationalization of retention in MOUD may also play a role in this discrepancy; the study by Armstrong et al. (42), for example, looked at departure from MOUD due specifically to relapse as an outcome rather than retention or attrition from MOUD more generally. It is also possible that these studies indicate there are other confounding factors responsible for the observed relationship between older age and increasing retention; for example, del Rio et al. (43) found no association between age and retention after controlling for duration of opioid use. Finally, Perreault et al. (40), who did not find a significant association between age and retention, hypothesized that this may be due to "differences in sampling." Only 2 of the included studies found that adolescents and young adults had longer retention in MOUD than older adults. Of these, the study by Chaudhry et al. (34), although initially finding an association between younger age and longer retention, determined that this association was due to the presence of confounding variables. The authors corrected for this in multivariate analysis and their final results were consistent with those of most of the other studies included in our sample. In the other study, by Cousins et al. (35), important confounders, such as duration of drug use history, may not have been controlled for.
Stratifying by MOUD type (i.e., methadone, buprenorphine, and naltrexone) revealed that the evidence for a positive association between age and MOUD retention is fairly consistent for both methadone and buprenorphine, with a majority of studies finding such a relationship for both medications (75.0% and 66.7%, respectively). Neither of the studies examining the relationship between age and retention for naltrexone found a significant association, which points to the need for more studies examining retention in naltrexone maintenance to explore the association with age and to include adolescents and young adults in the sample.
Exploring the relationship between age and retention in MOUD stratified by how retention was measured yielded similar results. Most of the studies that measured retention using a time frame shorter than 12 months or exactly 12 months found a positive association between age and retention in MOUD (63.6% and 61.5%, respectively). A larger percentage of studies that measured retention using a time frame longer than 12 months found a positive association between age and retention in MOUD (75.0%); however, only 8 studies (25%) included in the subgroup analysis used a time frame longer than 12 months. This highlights the need for more studies examining retention in MOUD beyond 12 months among adolescents and young adults.
Similar results were observed when looking at if the relationship between age and retention varied by whether age was treated continuously and categorically. Although 62.5% of studies that treated age continuously in analysis found a positive association between age and retention, a higher percentage (83.3%) of those that treated age categorically in analysis found a positive association. These findings point to the possibility of a nonlinear relationship between age and MOUD retention.
Few studies (n = 17; 6%) exploring retention in MOUD as an outcome focused on adolescents or young adults specifically, allowing for an analysis of the facilitators of and barriers to adherence and retention in MOUD in this population. The results of the qualitative analysis of this group of articles demonstrated that the relationship between age and retention did not stay constant when assessing the association between age and retention within adolescents and young adults, suggesting some level of homogeneity within this age group or a nonlinear relationship between age and retention. Qualitative analysis revealed a number of factors at different levels that are associated with retention in MOUD among adolescents and young adults, but it is also important to note the factors that were absent from analysis and were not seen in any of the included research studies, such as individual behaviors, socioeconomic status, the role of social networks, or structural factors related to MOUD accessibility and availability. Additional research exploring the possible role of these factors in adherence to and retention in MOUD among adolescents and young adults would help inform strategies to better engage this group in treatment.
The most consistently observed association in this review was that younger age is linked to lower rates of retention. This association was also seen in other systematic reviews on the topic of retention in MOUD (28). Of the final included articles in the Brorson et al. (28) 2013 systematic review of risk factors for dropout from MOUD, younger age was the most frequently cited risk factor for MOUD dropout. Unlike in the present review's final sample, Brorson et al. (28) did not find any studies that contradicted this relationship; indeed, of all risk factors, younger age was the only factor that consistently predicted dropout. Similarly, the observation from qualitative synthesis that youth receiving methadone treatment seem to experience better retention than youth receiving buprenorphine or naltrexone is consistent with the findings in the reviews by Hopfer et al. (26), Minozzi et al. (25), and Timko et al. (27). Our review adds to the findings of these previous reviews by exploring the factors, especially at the interpersonal and institutional levels, that are possibly responsible for the shorter retention in MOUD seen among adolescents and young adults. Although it is not possible to draw any definitive statements from the quantitative synthesis in this review, our findings do point to factors that merit more research. Future research, especially qualitative and mixed methods research, should explore possible factors that explain the relationship between younger age and increased dropout from MOUD. Higher-quality research studies exploring adherence to and retention in MOUD and that stratify populations by age groups to compare adolescents and young adults with older adults would also add substantially to the literature. These findings also point to the need for standardization in the approach to agebased analyses in quantitative research on MOUD adherence and retention where the sample includes adolescents and young adults. Finally, these findings indicate that providers of MOUD should consider tailored approaches to retain adolescents and young adults in treatment. At the individual and interpersonal levels, this would include screening and referral services for factors that might compromise retention, such as concurrent drug use, physical health, psychiatric comorbidity, or family stress and conflict. At the institutional level, MOUD providers should ensure that adolescents and young adults receive adequate dosing of medication and should adjust policies and procedures so younger program participants have opportunities for more flexible dosing and pick-up schedules.
This review faces the same limitation addressed by Brorson et al. (28) in that differences in the operationalization of retention and in reason for dropout may act as a confounder of our observations. LMIC are underrepresented in our final sample of articles, whereas the United States is overrepresented. Studies conducted in LMIC that included adolescents and young adults and would otherwise meet our inclusion criteria did not segregate their data by age group, nor did they assess the associations between age and adherence or retention in MOUD; instead, these studies tended to focus on clinical process improvement (e.g., the study by Hasečić et al. (61)), demonstrating different research priorities compared with studies done in highincome countries. Few qualitative studies explore adherence and retention in MOUD in this age group. As such, the existing body of research shows an association between age and adherence and retention without exploring the specific factors responsible for lower rates of adherence and retention among adolescents and young adults.
The quality of the articles included for quantitative analysis and synthesis in this review was, overall, low, according to Grading of Recommendations Assessment, Development, and Evaluation standards. More high-level research studies, especially randomized controlled trials, are needed to boost the level of evidence of the association between age and retention in MOUD.
A total of 22 articles were excluded from this review because they were in languages not spoken by any of the reviewers (specifically, Chinese, French, German, Norwegian, Persian, and Swedish). Given the international focus of this review, we worked to ensure we could review articles in as many languages as possible, but we were unable to cover the languages of all potentially eligible articles. In addition, Embase introduced indexing for adolescents late and subsequent indexing in that database is not as sensitive as the Medline indexing for this concept, so it is possible that some studies in which adolescents were part of the sample were overlooked in this review because of this limitation.

CONCLUSION
This review upholds the association seen in previous reviews that adolescents and young adults experience shorter retention in MOUD than individuals in older age groups. Studies that explore retention in MOUD as an outcome and that include adolescents or young adults in their sample should adequately stratify their findings or otherwise account for the possible difference in retention in this age group. Much of the challenge in drawing definitive conclusions in this review was due to the lack of a standardized definition of concepts such as "adolescent," "young adult," "adherence," and "retention"; as such, the research community should develop and use standardized definitions for different age groups of interest as well as for the concepts of adherence and retention, to facilitate comparison across large groups of studies. More studies are needed, especially randomized controlled trials and qualitative and mixed methods studies, that explicitly look at what mechanisms are at play that lead adolescents and young adults to drop out of MOUD earlier. Individual, interpersonal, and institutional factors, such as concurrent drug use, family conflict, and MOUD dosage and flexibility, play key roles in moderating or mediating this relationship and suggest possible avenues for better tailoring MOUD to meet the needs of this age group. It is clear that adolescents and young adults experience unique needs when it comes to continued engagement in MOUD, and this review highlights the main factors that should be addressed.