Introduction

Mental health disorders (MHDs) among runaway and homeless youth (RHY) have become a major public health issue worldwide [1]. RHY refers to individuals aged 12–24 years who either leave their homes without parental or legal guardian consent [2], lack a permanent dwelling, and reside in public spaces, shelters, with unfamiliar individuals, on the streets, with friends, in transitional housing, or other non-domicile settings [3].

The prevalence of RHY has increased in recent years [4]. Moreover, RHY are more likely to report co-occurring disorders, such as MHDs [5,6,7], suicidal behaviors [8] and/or substance abuse [9, 10]. They also are more likely to report risky sexual behaviors [11] and experience trauma disorders [1]. While these findings highlight RHY as one of the most vulnerable populations globally, RHY culture can serve as a substitute for absent parental support, creating significant challenges for healthcare and human services professionals working with this group [1]. Indeed, RHY often exhibit resistance to traditional methods of assistance, including substance use counseling, HIV prevention programs, and psychotherapy [1].

Youth experience homelessness or run away for various reasons. Some are compelled to leave their homes or voluntarily choose to depart due to family conflicts or dysfunction [12]. Homelessness can also result from inadequate discharge planning and a deficiency of support services for youth transitioning from child welfare or juvenile justice systems [13, 14]. Moreover, familial residential instability, familial poverty, and financial hardships significantly impact the lives of many RHY [15, 16].

Two previous meta-analyses have reported the prevalence of MHDs among homeless children (10–26%) [17], or children and adolescents in the child welfare system (4–27%) [18], but neither focused on RHY. As well as being almost a decade old, these studies: (i) only included homeless children (< 18 years) [17] or children and adolescents in the child welfare system (7–17 years) [18]; (ii) only reported the pooled prevalence rate of some specific MHDs (i.e., they did not report the pooled prevalence rate of psychological distress, major depressive disorders, bipolar disorders, personality disorders, adjustment disorders, or schizophrenia); (iii) did not report the lifetime and current pooled prevalence rate of MHDs; (iv) did not compare two groups of RHY (adolescent minors vs. young adults); and (v) did not conduct any subgroup analyses and sensitivity analyses, or meta-regression to detect sources of heterogeneity [17].

To the best of the present authors’ knowledge, there are no previous meta-analyses examining the lifetime and current pooled prevalence rates of a wide range of MHDs among RHY. Consequently, the present study is novel in its aim of estimating the pooled prevalence of MHDs among youth, encompassing both adolescents and young adults. The findings could enhance researchers’ understanding of the pooled prevalence rates of various MHDs, assisting practitioners and policymakers in targeting RHY with suitable interventions, developing adapted psychiatric services, providing professional training, and planning further research.

Methods

Registration and protocol

The present systematic review and meta-analysis was conducted based on Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [19]. The review protocol was registered on PROSPERO (Ref: CRD42023476261).

Search strategy

A thorough search of English-language published papers and abstracts from December 1, 1985, to October 1, 2023, was systematically conducted using the Scopus, PubMed, Web of Science, and Cochrane Library databases. Additionally, a search on Google Scholar was performed to locate any additional relevant studies. The search strategy, employing crucial Boolean operators (AND/OR), was developed and adjusted for diverse databases, utilizing the following initial keywords: “(mental disorders), (psychotic disorders), (mentally ill persons), (homeless youth), (homeless persons), (runaway youth), (adolescent), (young adult), (adult children)”. Furthermore, the bibliographies of the published studies included in the meta-analysis were examined to locate the presence of any additional relevant studies. If multiple studies presented findings on the same sample of RHY, the data providing the most comprehensive details regarding the prevalence of MHDs were used in the analysis. The specifics of the search strategy, encompassing the amalgamation of keywords employed across various electronic databases, are outlined in Table 1.

Table 1 Search strategy
Full size table

Study eligibility and exclusion criteria

The eligibility criteria used for inclusion were: (a) RHY aged 12–24 years [2, 3]; (b) reporting original prevalence data on RHYs’ MHDs, and life-time MHDs including ever having a MHD (at least one time) and current MHDs (defined as having MHDs within the past 30 days); and (c) any type of quantitative empirical study (e.g., cross-sectional, cohort, case–control, mixed-methods, and interventions with baseline data). The study excluded qualitative papers, secondary analyses without primary data, systematic reviews, meta-analyses, and unpublished theses (i.e., those not peer-reviewed, such as PhD theses and Master’s theses).

Data extraction process

EndNote X7 software was utilized to eliminate duplicate papers. Subsequently, two authors (BA and JH) independently assessed the titles and abstracts in accordance with the study’s inclusion and exclusion criteria. In instances of disagreement between the two reviewers, resolution was sought from a third author (RM). In the subsequent step, the full texts of the studies were examined based on the criteria for eligibility in the study. Two authors (BA and JH) independently conducted the extraction of data for the studies selected for inclusion in the meta-analysis. The extracted information encompassed details such as the study authors, participants’ age, publication year, country where the data were collected, study design, sample size, population specifics, quality assessment of studies, and criteria for assessing MHDs. Where necessary, the authors of the selected studies were contacted to obtain additional information. The agreement between the two authors was assessed using Cohen’s Kappa statistic. The degree of agreement was categorized into levels such as poor, slight, fair, moderate, substantial, and almost perfect. Corresponding numerical values were assigned as follows: 0, 0.01–0.02, 0.021–0.04, 0.041–0.06, 0.061–0.08, and 0.081–1.00, respectively [20]. Discrepancies between the two authors (comprising less than 10% of the total) were addressed through the intervention of a third author.

Risk of bias of studies

The Newcastle–Ottawa Scale (NOS) [21] was used to evaluate the quality of studies, encompassing three criteria: (i) the selection domain, which includes the representativeness of the exposed group, selection of the non-exposed group, and ascertainment of exposure (three items for cross-sectional studies and four items for cohort studies); (ii) the comparability domain, involving group comparability based on the study design or analysis (one item each for both cross-sectional and cohort studies); and (iii) the exposure/outcome domain, incorporating the assessment of outcome (one item for cross-sectional studies and three items for cohort studies) (Table 2). The studies were classified into four categories: unsatisfactory, satisfactory, good, or very good, with a potential maximum score of 8 for cohort and case–control studies. In assigning scores, studies with a total score of 0–2 were deemed “unsatisfactory,” those with scores of 3–4 were labeled “satisfactory,” 5–6 were considered “good,” and 7–8 were categorized as “very good.” In total, 23 studies received a high-quality rating, 41 were rated as good quality, and 37 were rated as satisfactory quality.

Table 2 Risk of bias assessment using the Newcastle–Ottawa Scale
Full size table

Study selection process

Initially, 11,266 papers were found through the four database searches (Fig. 1). After paper duplicates were excluded (n = 6358), the titles and abstracts of 4,908 papers were screened. Of these, 845 were found to be related to the study’s aim. After a full text review, 744 studies were excluded. The main reasons for exclusion were as follows: 39 studies did not meet the quality appraisal score (5%), and 705 studies utilized a non-quantitative methodology or did not report parametric measurements such as lifetime prevalence of suicidal behaviors, coefficients or odd ratios of relative risks of determinants of study outcomes (95%). Following these exclusions, 101 studies remained for meta-analysis [6, 22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120].

Fig. 1
figure 1

PRISMA flow diagram

Full size image

Data synthesis and statistical analysis

The analysis considered the lifetime or current prevalence of conduct disorders, psychological distress, depression, major depressive disorders, post-traumatic stress disorder (PTSD), personality disorders, attention-deficit/hyperactivity disorder (ADHD), bipolar disorders, anxiety, oppositional defiant disorders, anorexia, adjustment disorders, dysthymia, schizophrenia, obsessive–compulsive disorders, and gambling disorder as MHDs. Any reports of overall prevalence without mentioning the specific time period in the studies were considered as lifetime use for the purposes of the meta-analysis. The pooled-prevalence estimates were obtained using a robust random-effects model [121], which considered the various sampling methods employed in the studies. Additionally, the sources of heterogeneity among studies were evaluated through the application of Cochran’s Q and I2 tests. Subgroup analyses were performed to pinpoint the sources of heterogeneity, considering factors such as participants’ age, year of publication, geographical location, quality assessment of studies, diagnosis criteria for MHDs, and sample size. For each subgroup analysis, a minimum of two studies reporting data on the variable of interest was necessary. A sensitivity analysis was performed using Baujat plots to assess the impact of the most significant study on overall heterogeneity and to exclude it during the evaluation of each specific study’s effect on the overall estimate. Ultimately, a multivariate meta-regression analysis was undertaken to investigate the primary source of heterogeneity. Statistical significance was defined as a p-value < 0.05, and the meta-analysis was performed using R 3.5.1 with the “meta” package.

Results

Study characteristics

Of 101 studies selected, 90 were from the America region (n = 123,197 participants), one from the African region (n = 227 participants), five from the European region (n = 541 participants), one from the Western pacific region (n = 187 participants) and four from the South-East Asia region (n = 326 participants). The country with the highest number of included studies was the USA, with 82 studies (n = 122,037) (Table 3). Considering country income level, 96 studies were conducted in high-income countries (n = 123,925), four studies were conducted in a lower-middle-income country (n = 532), and one was conducted in an upper-middle income country (n = 21). The study sample size had a mean of 1232 participants, with 21 being the lowest sample size [56] and 76,596 being the largest sample size [71]. Response rates between the studies varied from 35% to 100%. Participants had a mean age of 18.43 years and were more likely to be male (mean 56%), varying from 0% to 100%. Moreover, 47% had adolescent minors as participants (12–17 years old) and 53% had young adults as participants (18–25 years). Almost all studies were cross-sectional (93%). Finally, 57 studies were published between 2010 and 2023 (56%), 73 studies used a standard scale/questionnaire to determine MHDs (72%), with the Diagnostic and Statistical Manual of Mental Disorders (DSM) being the most prevalent (22%), followed by the Mini International Neuropsychiatry Interview MINI and the Center for Epidemiological Studies-Depression Scale (CES-D) (both 19%).

Table 3 Characteristics of the 101 studies identified for review
Full size table

Pooled prevalence of life-time and current MHDs among RHY

The findings showed that lifetime MHDs most frequently reported by RHY were conduct disorders (47%), psychological distress (47%), depression (43%), major depressive disorders (34%), post-traumatic stress disorder (PTSD) (33%), personality disorders (27%), ADHD (25%), bipolar disorders (23%), anxiety (22%), oppositional defiant disorders (21%), anorexia (15%), adjustment disorders (15%), dysthymia (14%), schizophrenia (11%), obsessive–compulsive disorders (9%), and gambling disorder (8%) (Table 4 and Supplementary Files 1–26). In addition, the data showed that the current MHDs most frequently reported by RHY were depression (31%), major depressive disorder (23%), anxiety (23%), PTSD (21%), ADHD (16%), bipolar disorder (15%), personality disorders (13%), oppositional defiant disorders (13%), schizophrenia (8%), and obsessive–compulsive disorders (6%).

Table 4 Pooled prevalence of life-time and current prevalence of mental health disorders, subgroup analyses of pooled prevalence of life-time and current mental health disorders by age of participants, and time of study publication among runaway and homeless youth
Full size table

Subgroup analysis

Several subgroup analyses were conducted to determine the primary factor causing heterogeneity in the pooled odds of MHDs (Supplementary Files 27–105). The factors considered included participants’ age, year of study publication, geographical location, quality assessment of studies, diagnostic criteria for MHDs, and sample size. Subgroup analyses detected some source of heterogeneity in some specific MHDs (see next two sections).

Subgroup analyses of pooled prevalence of life-time and current MHDs based on age of participants among RHY

The participants were divided into two groups based on their age, and a subgroup analysis was conducted with those aged: (i) 12–17 years (adolescent minors) and (ii) 18–24 years (young adults) (Table 4 and Supplementary Files 27–41). Results showed that as age increased, the (i) lifetime and current prevalence of depression, bipolar disorder, anxiety and PTSD increased, (ii) lifetime prevalence of schizophrenia and major depressive disorders increased, (iii) lifetime and current prevalence of personality disorders decreased, (iv) lifetime prevalence of ADHD, conduct disorders, and obsessive–compulsive disorders decreased, and (v) current major depressive disorders decreased. The lifetime prevalence of adjustment disorders at these specific ages was stable. The findings also show that externalizing disorders were predominantly prevalent among adolescent minors (except for current major depressive disorders), while internalizing disorders were more prevalent among young adults.

Subgroup analyses of pooled prevalence of life-time and current MHDs based on time of study publication among RHY

The studies were divided into two groups based on the year of publication for each study and a subgroup analysis was conducted by classifying the studies into two different time periods: (i) ≤ 2010 and (ii) > 2010 (Table 4 and Supplementary Files 42–58). This analysis showed several trends: (i) an increase in the lifetime and current prevalence of depression, major depressive disorders, PTSD, and schizophrenia; (ii) an increase in the lifetime prevalence of anxiety, ADHD, personality disorders, conduct disorders, and psychological distress; (iii) an increase in the lifetime prevalence of bipolar disorders; (iv) a decrease in the lifetime prevalence of bipolar disorders, oppositional defiant disorders, and obsessive–compulsive disorders; and (v) no difference in the lifetime prevalence of adjustment disorders over time.

Sensitivity analysis

Sensitivity analysis utilizing Baujat plots was conducted to evaluate influential effects. Effects on the right-hand side of the plots indicate studies with higher levels of heterogeneity (Supplementary Files 106–147). The sensitivity analysis was able to decrease the heterogeneity between studies for lifetime prevalence of anorexia (Cutuli (2018) [83] made the most significant contribution to heterogeneity), dysthymia (Feitel et al. (1992) [88] made the most significant contribution to heterogeneity), gambling disorders (Taylor et al. (2006) [117] made the most significant contribution to heterogeneity), and psychological distress (Narendorf et al. (2020 and 2023) [41, 113] made the most significant contributions to heterogeneity). The test did not detect heterogeneity between studies for other lifetime prevalence of MHDs. The sensitivity analysis reduced the heterogeneity between studies for current prevalence of anxiety and schizophrenia (Middleton et al. (2018] [103] made the most significant contributions to heterogeneity), but was unable to decrease the heterogeneity between studies for other current prevalence of MHDs.

Meta-regression

Multivariate meta-regression analysis was conducted to further investigate the sources of heterogeneity (Table 5). The meta-regression results indicated that the age of participants may contribute to the heterogeneity between the included studies in terms of the lifetime prevalence of depression, conduct disorders, and obsessive–compulsive disorders. Additionally, it was found that the year of study publication may contribute to the heterogeneity between the included studies for (i) the current prevalence of depression and (ii) the lifetime prevalence of anxiety and ADHD. Moreover, diagnostic criteria for MHDs may contribute to the heterogeneity among the included studies on lifetime bipolar and obsessive–compulsive disorders. Finally, the quality assessment of studies may contribute to the heterogeneity among the included studies on the lifetime prevalence of conduct disorders.

Table 5 Multivariate meta-regression of the lifetime and current pooled prevalence of mental disorders among runaway and homeless youth (RHY) by age, sample size, publication year, quality assessment of studies, geographic location, and diagnosis criteria
Full size table

Moderator analysis

Subgroup analyses confirmed that (i) age was a statistically significant moderator for current depression, current major depressive disorders, lifetime adjustment disorders, and lifetime obsessive–compulsive disorders; (ii) year of study publication was a statistically significant moderator for current depression, lifetime adjustment disorders, and current schizophrenia; (iii) geographical location was a statistically significant moderator for current anxiety, current bipolar, lifetime bipolar, and lifetime obsessive–compulsive disorders; (iv) the quality assessment of studies was a statistically significant moderator for current major depressive disorders, lifetime obsessive–compulsive disorders, lifetime personality disorders, current PTSD, and lifetime psychological distress; (v) diagnostic criteria for MHDs were statistically significant moderators for lifetime depression, lifetime adjustment disorders, lifetime schizophrenia, lifetime obsessive–compulsive disorders, and lifetime personality disorders; and (vi) sample size was as a statistically significant moderator for current depression and lifetime obsessive–compulsive disorders.

Multivariate meta-regression analysis found that the year of study publication was a statistically significant moderator for the current pooled prevalence of depression. Additionally, the (i) age of participants was a statistically significant moderator for the lifetime pooled prevalence of depression, conduct disorders, and obsessive–compulsive disorders; (ii) year of study publication was a statistically significant moderator for the lifetime prevalence of anxiety and ADHD; (iii) diagnostic criteria for MHDs were statistically significant moderators for the lifetime prevalence of bipolar and obsessive–compulsive disorders; and (iv) quality assessment of studies was a statistically significant moderator for the lifetime prevalence of conduct disorders. Older age (younger adults) was associated with a higher prevalence of depression, while younger age (adolescent minors) was associated with a higher prevalence of conduct and obsessive–compulsive disorders (p < 0.05). Publishing a study after 2010 was associated with a higher prevalence of current depression and lifetime prevalence of anxiety and ADHD (p < 0.05). Diagnostic criteria for MHDs other than DSM and MINI were associated with a higher prevalence of lifetime bipolar and obsessive–compulsive disorders (p < 0.05). Studies of lower quality were associated with a higher prevalence of conduct disorders (p < 0.05).

Discussion

The present meta-analysis estimated the pooled prevalence rates of MHDs among RHY. As far as the present authors are aware, no previous meta-analyses have ever been conducted estimating the pooled prevalence of MHDs among RHY. Findings from the present study indicated that the lifetime pooled prevalence of MHDs among RHY ranged from 8% to 47%, while the current pooled prevalence ranged from 6% to 31%. It was expected that the lifetime pooled prevalence of MHDs among RHY would be higher than the current pooled prevalence rate and this was the case. The prevalence in the present study was higher than the pooled prevalence rate of MHDs in previous meta-analyses among the general population of children and adolescents (13%) [122], homeless children (10–26%) [17], and children and adolescents in the child welfare system (4–27%) [18]. A possible reason for this may be that RHY face several challenges and have multiple health issues [123, 124], leading to a higher rate of MHDs. Additionally, adverse experiences such as maltreatment and serious neglect among RHY [125,126,127] may contribute to an increased prevalence of MHDs among RHY.

Regarding the most common MHDs among RHY, psychological distress and conduct disorders had the highest lifetime pooled prevalence rates (both 47%), while depression had the highest current pooled prevalence rate (31%). These rates are higher than those reported in previous meta-analyses among children and adolescents in the child welfare system or general population of children and adolescents for conduct disorders (6%-20%) [18, 122] and depression (3%–18%) [18, 122]. To the best of the present authors’ knowledge, no previous study has reported the pooled prevalence rate of psychological distress among RHY.

The elevated prevalence of psychological distress among RHY can be attributed to various factors. Firstly, as indicated by prior studies, psychological distress is associated with several adverse behaviors, including substance abuse, conduct problems, and engaging in sexual risk behavior [6, 7, 9,10,11], behaviors that may all be present among RHY. Secondly, this heightened psychological distress might stem from a lack of self-determination within the RHY population [128]. Lastly, the intricate dynamics of loneliness and social support may contribute to this phenomenon [129]. While ‘close’ relationships can offer some degree of support, they simultaneously expose individuals to potential victimization and challenging interactions with other distressed RHY [130].

According to previous studies, RHY are more likely to have conduct disorders than peers who reside in stable housing [25, 77]. Research also indicates that conduct disorder diagnoses are typically feasible until the age of 18 years, with a noticeable decline in prevalence among young adults aged 18 to 25 years [131, 132]. This could be due to the fact that aggression and impulsivity can have negative effects on RHYs’ abilities to reside in housing and shelter systems, reducing access to mental health support and potentially increasing the odds that they will experience homelessness [133]. Another explanation could be the fact that the presence of conduct disorder heightens the likelihood of initiating substance use by the age of 15 years, particularly illicit substances, with this risk persisting until 18 years of age [132]. Moreover, the probability of initiating cocaine, amphetamines, inhalants, and club drugs remains notably elevated up to the age of 21 years [134].

Also, RHY may experience a high level of neighborhood violence, have violent and criminal peers, and be involved with gangs [4]. For everyone, whether housed or not, adolescence may be a challenging period characterized by self-doubt and low self-esteem [123]. RHY may experience various forms of physical and sexual abuse, leading to diminished self-worth [135]. Additionally, facing abuse in the streets, from passers-by or the police, contributes to a reduction in their self-image, potentially leading to increased depression [135].

Examining other MHDs as reported in previous meta-analyses, the prevalence rates were notably higher among RHY. More specifically, 22% had anxiety, 25% had ADHD, 21% exhibited oppositional defiant disorders, and 33% had PTSD, each of these much higher the rates observed among the general population of children and adolescents (6.5% for anxiety, 3.4% for ADHD) [122]. Additionally, the rates in the present study were higher than those found among children and adolescents within the child welfare system (18% for anxiety, 12% for oppositional defiant disorders, 11% for ADHD, and 4% for PTSD) [18].

Another important finding was the increasing prevalence of MHDs over time. However, there is no previous study reporting the prevalence of MHDs among RHY over time. According to a previous meta-analysis, there had been a minor increase of MHDs between the 1980s and the 2000s among the general adolescent population [136]. There have been noticeable indications of parallel trends among the general adolescent population, including a rise in symptoms related to MHDs [137]. Furthermore, there has been an increase in the utilization of health services for the diagnosis and treatment of both psychosomatic health complaints and MHDs in high-income countries [138]. These trends broadly align with the findings from the present review. However, it is important to note that not all findings were equally consistent.

Another novel finding of the present study was that internalizing disorders (e.g., anxiety disorders, depression, PTSD) were more prevalent among young adults, while externalizing disorders (e.g., conduct disorders, ADHD) were more prevalent among adolescent minors. This finding is in line with previous studies with US and Swedish adolescents [139, 140] which reported that there is a trend of increasing internalizing symptoms and decreasing externalizing symptoms among general adolescents. One possible explanation for this finding could be that young adults with MHDs may engage in substance use concurrently [141], experience family conflicts [142] and increased violence [143] and/or have limited access to therapeutic and supportive care in the community [144], leading to an increase in internalizing symptoms among this cohort. Moreover, internalizing symptoms, linked to adverse social and health outcomes in young adults [145], hold clinical significance. Given that these symptoms can progress to psychiatric disorders among a subset of young adults [146], clinicians should be vigilant about the rising prevalence of internalizing symptoms among young adults. Finally, the trend of the rise in internalizing symptoms and the decline in externalizing symptoms suggests a shift more closely associated with the natural evolution of psychiatric disorders rather than homelessness.

Finally, the current meta-analysis identified several sources of heterogeneity between studies through meta-regression and moderator analysis. Similar levels of heterogeneity have been observed in previous meta-analyses conducted within the general adolescent population [122]. Heightened heterogeneity can be influenced by several factors, including country variations, mean age, sample size, year of publication, diagnostic criteria, and the quality assessment of studies. These aspects may contribute to methodological challenges and should be considered in future studies. In the present study, there was a lower estimate of the lifetime pooled prevalence of major depressive disorders and obsessive–compulsive disorders with increasing age. This finding is unexpected, as an inverse relationship was anticipated. This should be examined in future studies. It was also found that studies published after 2010 were associated with a higher prevalence of current depression, as well as lifetime prevalence of anxiety and ADHD. This association could be due to increased numbers of studies carried out post-2010, leading to more comprehensive assessments of MHDs among RHY. There were also variations in the diagnostic criteria which could have led to discrepancies in the prevalence rates, aligning with findings from prior meta-analyses [18, 122]. In an older study examining two prominent nosological systems, the DSM-IV-TR consistently categorized a higher number of children and adolescents as having an anxiety disorder compared to the ICD-10 [147]. Moreover, it was observed that studies of lower quality exhibited a higher prevalence of conduct disorders. This highlights the importance of acknowledging that studies with lower quality may yield elevated prevalence rates compared to those with higher quality data. This consideration should be taken into account in future epidemiological studies.

Limitations

The present meta-analysis has several limitations that should be noted. First, nine-tenths of the studies were carried out in the US and Canada (89%), and the distribution of MHDs was unequally dispersed across the 101 included studies. Therefore, the authors were unable to create a distribution map for each MHD in each country or continent. Second, almost all studies (95%) were conducted in high-income countries, and the findings may not be generalizable to other countries. Third, there was inconsistency in defining MHDs across various studies. For instance, some studies relied on self-reporting rather than adhering to standardized criteria. Fourth, some studies did not specify the duration (whether it was lifetime or current) of MHDs. Fifth, in some studies, the type of MHD was not specified (i.e., they reported MHDs without specifying the particular type of MHD. Therefore, these studies were excluded from the analysis. Sixth, several important pieces of data were not available from studies (e.g., age at the first episode of homelessness or runaway, co-occurring disorders such as substance use, and receiving treatment). Therefore, these trends could not be investigated with these crucial covariates. For example, according to a study from 2005 to 2018, only one-fifth of adolescents received MHD treatment [148], and receiving treatment might affect MHD trends. Finally, excluded from consideration were gray literature sources such as dissertations, research and committee reports, government reports, conference papers, ongoing research, manuscripts, and unpublished studies. The decision to exclude these sources was based on the research team’s inability to adequately assess their quality and the fact that they had not been peer-reviewed.

Conclusion

The present study is the first meta-analysis to estimate the lifetime and current pooled prevalence rate of MHDs among RHY. Several innovative subgroup analyses, such as trends of MHDs over time and comparisons of MHDs among adolescent minors vs. young adults, were conducted. Additionally, to the best of the authors’ knowledge, the present study is the first to perform meta-regression and moderator analysis to detect potential contributing factors to the heterogeneity of studies on the lifetime and current pooled prevalence rate of MHDs among RHY. The findings suggest that RHY had significantly higher prevalence of MHDs compared to other high-risk populations reviewed. Regular screening and the implementation of evidence-based treatments, and the promotion of integration and coordination between mental health services for adolescent minors and young adults with other service systems are recommended.

Based on the present results, externalizing symptoms were observed among adolescent minors, while internalizing symptoms were evident among young adults. Tailoring mental health services based on the type of symptoms and age is crucial, because adolescent minors and young adults may require distinct interventions for internalizing and/or externalizing symptoms. Future research should investigate the causal factors driving the observed difference between internalizing and externalizing symptom trends among adolescent minors and young adults. Moreover, it is imperative to formulate and implement effective brief and low-intensity psychological interventions [149, 150], including cognitive-behavioral therapy [151], for mitigating internalizing disorders among young adults. Additionally, it is recommended to incorporate selective parent training, family support, and school-based programs [152] for addressing and reducing externalizing disorders among adolescent minors.