Front. Adolesc. Med. Frontiers in Adolescent Medicine Front. Adolesc. Med. 2813-8589 Frontiers Media S.A. 10.3389/fradm.2025.1522678 Adolescent Medicine Systematic Review A systematic review of recovery high schools and collegiate recovery programs for building recovery capital among adolescents and emerging adults HennessyEmily A. 1 2 * GeorgeShane S. 1 KleinMorgan R. 1 O’ConnorJenny B. 1 EddieDavid 1 2 KellyJohn F. 1 2 Recovery Research Institute, Center for Addiction Medicine, Massachusetts General Hospital, Boston, MA, United States Department of Psychiatry, Harvard Medical School, Boston, MA, United States

Edited by: Anita Cservenka, Oregon State University, United States

Reviewed by: Bo Cleveland, The Pennsylvania State University (PSU), United States

Hannah Apsley, The Pennsylvania State University (PSU), United States

George Comiskey, Texas Tech University, United States

 Correspondence: Emily A. Hennessy ehennessy@mgh.harvard.edu 17022025 2025 31522678 04112024 23012025 © 2025 Hennessy, George, Klein, O'Connor, Eddie and Kelly. 2025Hennessy, George, Klein, O'Connor, Eddie and Kelly

This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

 Introduction

Educational settings represent a prime opportunity for reaching adolescents and young adults (ages 12–27) experiencing substance use disorders (SUD). Recovery high schools (RHSs) were established to help adolescents in recovery finish their education while maintaining alcohol and other drug (AOD) abstinence and collegiate recovery programs (CRPs) were implemented for college and university students. With the continued proliferation of these programs, this review synthesizes the empirical literature on the impact of both types of educational recovery supports for youth.

Method

This review's methodology was based on a previous Campbell review on the same topic from 2017. We searched five public databases (May 2024), used PRISMA 2020 reporting guidelines, and followed Campbell Collaboration guidelines including Synthesis Without Meta-Analysis. Eligible studies focused on adolescent/emerging adult participants in RHSs or CRPs. Any outcome broadly related to SUD recovery was eligible. We included quantitative or mixed methods studies and allowed for cost-benefit/cost-effectiveness designs. Data from eligible studies were extracted in duplicate and assessed for quality (Quality Assessment Tool for Quantitative Studies) by four review team members. Two cost-benefit/cost-effectiveness studies included in the CRP review were rated for quality using a separate tool appropriate for these designs by a single reviewer.

Results

We identified 37 manuscripts representing 25 unique studies focused on RHS (K = 7: N = 14,209) or CRPs (K = 18: N = 2,795). In the RHS studies, participants were predominantly White (41%) and females represented 29% of the sample. No studies met the criteria for low risk of bias. In the CRP studies, participants were predominately white (89%) and had slightly less female than male participants (45%). 11 of the 16 studies that did not use cost-benefit analysis were rated as high risk of bias. The quality rating of the two studies using cost-benefit designs indicated that both studies had fairly rigorous approaches.

Discussion

This research base suggests that students who participate in RHSs and CRPs may demonstrate reductions in AOD use, and improvements in social and academic outcomes, although given the existing research designs, statements about the incremental public health utility of investing in these programs relative to other approaches of equal intensity or duration cannot be made with confidence.

 collegiate recovery communities collegiate recovery programs recovery capital for adolescents model recovery high school systematic review 1H79TI085588K01AA028536, K23AA027577, K24AA022136-10Substance Abuse and Mental Health Services Administration and Opioid Response NetworkNational Institute on Alcohol Abuse and Alcoholism section-at-acceptanceAddiction in Adolescents Introduction

In 2019, over 70,000 youth enrolled in publicly funded substance use disorder (SUD) treatment in the United States, and an additional 1.9 million needed treatment but did not receive it (1, 2). More effective engagement of youth in seeking treatment for alcohol and other drug (AOD) use and sustained recovery support would have major public health benefits. Educational settings represent a prime opportunity for reaching adolescents and young adults experiencing heavy or problematic substance use including those with an SUD, given they spend the majority of their time in such settings. Although these educational environments present an opportunity to identify and support youth in need of treatment and recovery services, they also pose barriers due to the availability of substances on campuses and the social influence of peer groups who use substances (3, 4). In response to this need, recovery high schools (RHSs) were established to help adolescents in or seeking SUD recovery finish their high school education while maintaining alcohol and other drug abstinence (5) and collegiate recovery programs (CRPs) were implemented for college and university students (6, 7). It has now been a decade since federal offices formally recognized RHSs and CRPs as important supports for youth recovery (8).

These education-based recovery support services emerged in the 1970s to support students in their recovery while also helping them achieve their academic goals (7). Recovery high schools vary in size and structure, with enrollment typically ranging from 2 to 115 students (9), and exist as both independent schools and programs embedded within another school (10). They often offer much of what a traditional high school environment offers: classes, coursework, and a high school diploma. Yet, they also offer recovery-focused programming such as recovery group check-ins, individual counseling or links to mental health supports, opportunities for volunteer community services, and ensure regular connection/check-ins with parents or guardians (7). Collegiate recovery programs (CRP) also range in size and structure, with typical student enrollment ranging from 10 to 50 students (4, 11). They do not directly offer coursework or a college diploma but instead operate on college campuses and support students in recovery pursuing university degrees. They offer recovery supports such as sober housing/dormitories, recovery specific meeting spaces/congregational areas, AOD-free social events, group recovery meetings, and scholarships and/or work-study opportunities (12, 13). Some CRPs are membership driven where a period of abstinence is required to become an official member, while others have a more open membership policy. Whereas recovery high schools are professionally led, CRPs can be peer-driven, with a limited professional staff (10, 11, 14, 15). The underlying philosophy of both CRPs and RHSs is that by surrounding young people with others seeking recovery or those farther along in recovery (e.g., advanced students, staff) they will form new, pro-recovery social network connections that will reduce engagement with pro-substance use peers and serve to reinforce substance use disorder recovery (5).

Recovery capital refers to the resources necessary to initiate and sustain recovery (16–18). This construct provides an ecological framework that outlines different types of resources at different levels: human, financial, social, and community recovery capital. Both RHSs and CRPs are positioned as sources of community recovery capital, or as resources and supports located in the local community which are recovery supportive settings and can link their participants to an assortment of recovery supportive activities and peers. They are posited to support recovery through several different mechanisms and to address a multitude of related outcomes. A previously published recovery capital-oriented logic model of CRPs (13) described how CRPs could address personal, social, and community-level barriers of students in collegiate environments to support accrual of recovery capital through a variety of programming options. RHSs can also offer a similar way of building recovery capital. For example, in an RHS, a student can work towards completing their coursework for a high school diploma, building their skillset and motivation (human or personal recovery capital). Identifying with similar age peers, making pro-recovery peer connections, and seeing recovery modeled in others can help to build new relationships and help satisfy recovery-specific developmental needs related to esteem and belonging (19, 20), as well as boost their social recovery capital. Being in an environment with trusted adults who understand the unique needs of students in recovery, and may be in recovery themselves, can offer another facet of social recovery capital. RHSs may also serve as a bridge to broader community services, for example, to help link students to services that address basic needs (e.g., food pantries, stable housing).

Though no single model for RHSs or CRPs exists, education-based recovery support services have continued to grow in recent years, with over 40 recovery high schools currently in operation (21), and over 100 CRPs in development or operation in the United States (22). This review aims to synthesize the existing empirical literature on the impact of both types of educational recovery supports for youth and provide areas for future research direction.

Aim

A previous Campbell systematic review demonstrated a growth in the empirical literature on RHSs and CRPs within recent years and indicated potential for programming to result in a variety of positive outcomes for their participants (6). Yet, that review had a narrow focus on comparative research study designs and was published in 2018 (with a search from 2017); given these programs have been increasingly implemented across the United States and even globally, there was a need to return to the literature and synthesize any updates since that review.

 Methods

This review was based on the methods used in a previously published Campbell review conducted by the first author [Hennessy et al. (23)], but given time constraints, we were unable to preregister a new protocol. We used the PRISMA reporting guidelines (24) to report all aspects of the review, including the study inclusion/exclusion process and followed Campbell Collaboration best practice guidelines for systematic reviews (25, 26). Throughout the manuscript and tables, we use standard review convention (27) to refer to the number of studies (K) and to the number of participants (N).

Inclusion criteria

Participants, Intervention, and Comparator. Eligible studies focused on adolescent and emerging adult participants engaged in recovery high schools (RHSs) for high school students in AOD use recovery or collegiate recovery programs/collegiate recovery communities (CRCs/CRPs) for college students in AOD use recovery. Studies needed to directly identify the program as an RHS or CRP for students in recovery. There were no eligibility restrictions for the comparison groups, provided one was present.

Inclusion criteria: outcomes

If a study focused on and reported quantifiable participant outcomes related to SUD recovery, it was deemed eligible. Studies that only focused on the organizational functioning of RHS/CRPs were not eligible. Any outcome broadly related to AOD use recovery was eligible for inclusion. We anticipated for example, outcomes related to substance use (e.g., abstinence, quantity, frequency), academics (e.g., grades/grade point average), criminal history/behavior (e.g., interaction with the criminal justice system), and broader indicators of quality of life and recovery capital.

 Inclusion criteria: study design, setting, and timing

We included quantitative or mixed methods studies and allowed for cost benefit or cost effectiveness study designs. Given the difficulty of randomizing participants to these programs, we anticipated primarily quasi-experimental designs and few randomized controlled trials. Single group designed studies were also eligible. Studies that were solely qualitative in nature were not eligible. There were no restrictions on the study setting or timing (e.g., cross-sectional and longitudinal studies were both included as were studies from any timepoint).

 Search

A systematic search of the literature (from inception until 5/9/2023), using the search terms “collegiate recovery”, “recovery school”, “recovery high school”, “recovery hous*”, “university-based recovery center”, or “university based recovery center” in combination with substance use terms (see Supplementary Material), identified 786 records across five publicly available databases (i.e., PubMed, EMBASE, CINAHL, CENTRAL, and PsycInfo). A title/abstract screen removed 482 duplicate records and 260 ineligible reports, leaving 44 for full-text review. 11 additional articles were identified through reference list searching of previous reviews and eligible articles and assessed for inclusion. In May of 2024, we completed an updated search, which resulted in 82 records, 36 of which were duplicates. The remaining 46 titles and abstracts were screened and 33 were excluded. The full-text review of 10 articles resulted in 3 more articles being excluded, for a final addition of 7 new articles to the original search. See Figure 1 for the PRISMA flow diagram created using the Shiny app developed by Haddaway and colleagues (28). In the original search, only a single screener completed the screening due to resource constraints, but in the updated search, two screeners independently completed both the title/abstract and full-text screening through the Covidence platform and compared discrepancies. Screeners had 100% agreement for including/excluding studies: the only areas of disagreement were for the reason for excluding a study as only a single item could be selected as a reason for exclusion.

PRISMA study flow diagram (28).

 Data extraction

Data from eligible studies were extracted in duplicate by four members of the review team (EH, JO, MK, SG) using an instrument developed for this purpose in the Research Electronic Data Capture (REDCap) electronic data capture tools hosted at Massachusetts General Hospital (29). The coding form included categories for characteristics at the study level (study design, retention rate, demographic characteristics of participants), group level (intervention and comparator characteristics), and outcome level (type of outcome, measurement of outcome, author-reported statistical results of outcome measure). Coding discrepancies were resolved by discussion. Studies where the first author was an author on this review were coded by alternate review team members.

 Quality assessment

As suggested by previous reviewers to address quality assessment of non-RCTs (30, 31), we used the Quality Assessment Tool for Quantitative Studies by the Effective Public Health Practice Project [EPHPP (32)], for all studies aside from the cost-effectiveness/cost-benefit studies (32). Four reviewers (EH, JO, MK, SG) assessed for risk of bias independently and in duplicate and came to a consensus on the component ratings before a global assessment (Strong, Moderate, Weak) was decided. The Risk-of-bias VISualization (ROBVIS) online tool was used to visualize and present the quality assessment results from the EPHPP tool (33).

For the two cost-effectiveness/cost-benefit studies included in the CRP review, a single reviewer (EH) used a separate tool appropriate for these designs to rate their quality (34), which has 16 items and uses a weighted grading scheme, resulting in a score for each study of up to 100. In this tool, the question about sub-group analysis (weight of 1) was deemed not applicable for either study so the resulting scale range is up to 99.

 Synthesis

We used Synthesis Without Meta-analysis (SWiM) to guide our reporting of the outcomes (35). Given the likely diversity of the included studies and the small evidence base, the review team decided a priori that a descriptive synthesis would be undertaken, highlighting areas for future primary studies and meta-analyses to examine. All extracted data were tabularized and summarized in tables to aid in the descriptive synthesis of findings. We grouped the outcomes into the following seven broad categories of outcomes we established a priori and reported all identified outcomes in these categories in the text and/or tables: (1) Academic; (2) Crime/Criminal Involvement; (3) Employment; (4) Recovery Capital (validated measures); (5) Social network change; (6) Substance use; (7) Other psychosocial outcome. After coding the included studies, we returned to the “Other psychosocial outcome” grouping for each intervention (RHS and CRP) and created a further sub-coding of categories based on emerging categories. Drug craving/urges and length of time in recovery were both recategorized from “other psychosocial outcome” to the (6) Substance use category for ease of interpretation. We added new categories of (8) Mental or Physical Health (9), Life Satisfaction/Flourishing (10), Cost benefit, and (11) Metrics of Participation (CRPs only) to replace the broader “other” outcomes category. We considered metrics of participation important to gather as these can be considered an intermediate or enabling outcome. By attending programming, participants are positioned to gain the resources and social support necessary to achieve longer-term outcomes. Although the ultimate goals of CRPs include long-term impacts like remission and graduation, participation metrics recognize that the process of engaging the community is itself an important achievement. This aligns with theories of change that value process-oriented outcomes alongside impact-oriented ones (36).

Given the diversity of types of outcomes and ways these were measured, we did not standardize the coded quantitative outcomes into any given metric but instead reported these outcomes as they had been reported in the primary studies using the text and/or table entries provided by the authors.

 Results

We identified 37 manuscripts representing 25 unique studies that focus on RHS (K = 7: N = 14,209 including a national sample from a pooled dataset of adolescents who have undergone SUD treatment, or N = 1,242 without that sample included) or CRPs (K = 18: N = 2,795). An overall summary of key characteristics of these studies is in Table 1 (RHS) and Table 2 (CRP). Notably, there were no RCTs of either type of educational recovery support, so the strongest study designs identified are comparative studies that attempt to equate groups through matching or cost-benefit analysis studies.

Characteristics of recovery high school program studies (K = 7).

Study Citation N Sample Study design Exp. N Comp. N Follow-Ups Retention Sex Age Race/ethnicity Primary substance Primary substance of focus
1 Hensel & Wilburn (37) 6 Adolescents in substance use recovery attending a recovery high school in Indiana (from larger sample of 40) Single-group prospective; longitudinal Daily (Jan.-Feb. 2022) 81.2% (147/180 EMAs) 57.3% Female43.7% Male NR White = 71% NR NR
2 Giles (38) 375 Students (13 to 21 years) enrolled at the William J. Ostiguy Recovery High School in Boston, Northshore Recovery High school in Beverly, and Independence Academy in Brockton between September 2011 and January 2022 Single-group retrospective; cross-sectional 37.87% Female62.13% Male M = 16.98 White = 83.47%Black = 7.2%Asian = 0.53%Hispanic = 6.4% Alcohol, alcohol and other drugs, cannabis, crack cocaine, cocaine, heroin, opioids, multi-substance Alcohol (measured at intake; N = 126, 33.60% of sample); Marijuana (N = 109, 29.07%); Heroin, Prescription Opiate Misuse, Prescription Opiate Misuse (N = 77, 20.53%); Cocaine, Crack (N = 27, 7.20%); Benzodiazepines (N = 20, 5.33%); Other (N = 16, 4.27%)
3 Knotts (39) 57 RHS: Students enrolled at Hope Academy (IN) between Fall 2010 and Spring 2015. Comparator: Matched comparison group of peers not in SUD recovery from the NWEA norms data QED; longitudinal 43 43 49.1% Female50.9% Male NR White = 82.5%Black = 3.5%Hispanic = 1.8% Any NR
4 Finch et al. (40) 194 High-school adolescent with a history of alcohol and substance use; recently discharged from an intensive treatment program for substance use disorder and enrolled in high school. Recruited upon discharge from 10 substance use treatment facilities in MN, WI, or TX. Compared by RHS (28 + days) vs. non-RHS status. QED; longitudinal 134 60 3-mo., 6-mo., 12- mo. 78% (6-mo.) 49.49% Female50.51% Male M = 16.41 (SD = 1.01) White = 85.57%Black = 7.22%Other = 7.21% Any
Tanner-Smith et al. (41) 194  QED; longitudinal 134 60 3-month, 6-month, 12- month 78% 49.49% Female50.51% Male M = 16.41 (SD = 1.01) White = 85.57%Black = 7.22% Any At baseline: 19 days of alcohol use, 55 days of marijuana use, and 30 days of other substance use in the past 90 days
Tanner-Smith et al. (42) 260  QED; longitudinal 143 117 3-month, 6-month, 12- month 81% (6 m) and 71% (12 m) NR M = 16.07 (SD = 1.03) White = 82.69%Black = 10.38%Hispanic = 11.5% Any NR
Glaude et al. (43) 246 Cross-sectional 121 125 48.37% Female51.63% Male M = 16.34 (SD = 1.06) White = 74%Black = 3.7%Asian = 2.0%Native American  = 1.6%Hispanic = 6.1% Any NR
Tanner-Smith et al. (44) 13,261 (1) and (2): High-school adolescent with a history of alcohol and substance use; recently discharged from an intensive treatment program for substance use disorder and enrolled in high school. Recruited upon discharge from 10 substance use treatment facilities in MN, WI, or TX. (3) a national comparison sample of U.S. adolescents receiving SUD treatment  QED; longitudinal 171 (2) 123(3) 12,967 (1) 51.46% male(2) 59.84% male(3) 73.23% male Age 13–14 (1) 2.92% (2) 9.84% (3) 15.28%Age 15–16 (1) 42.69% (2) 48.36% (3) 58.22%Age 17–18 (1) 54.39% (2) 41.80% (3) 26.49% (1)White = 85.96% African American = 4.09% Hispanic = 5.26%Other = 4.68%(2)White = 77.24%African American = 4.88%Hispanic = 12.2%Other = 5.69%(3)White = 37% African American = 15.07% Hispanic = 29.95%Other = 18.48% Any
Weimer et al. (45) 260 Cost effectiveness 143 117 3-month, 6-month, 12- month 80.6% (6-month); 71.4% (12-month) 45% Female55% Male M = 16.3 (SD = 1.09) White = 82% Any NR
5 Lanham & Tirado (46) 72 179 students who graduated from the Serenity High School, a public school located in the McKinney Independent School District in Texas, between 2000 and 2010. Single-group retrospective; cross-sectional 58.3% Female41.7% Male M = 21.4 (SD = 2.5) White = 90.3% Any NR
6 Moberg & Finch (47) 321 Students in 17 recovery high schools [California (3), Colorado (1), Minnesota (8), Pennsylvania (1), Tennessee (1), Texas (3)] that fit ARS guidelines and had been operating for at least two years. Single-group retrospective; cross-sectional 46% Female54% Male Not Reported White = 78%Black = 3%Asian = 3%Native American = 4%Hispanic = 7% Any NR
7 Hennessy et al. (48) 28 Youth ages 12-19 attending a RHS or other treatment/recovery support service for AOD, or having seen a provider for AOD. Compared RHS vs. non-RHS student status. Mixed methods; cross-sectional 20 8 28.57% Female71.43% Male M = 17.32 (SD = 1.33) White = 82.15% Alcohol and other drugs NR
Sum 14209 3 Longitudinal; 4 Cross-sectional 29% Female 17.3 years (k = 4) 41% White (K = 7)28% Hispanic (K = 4)14% African American (K = 4)3% Native American (K = 2) <1% Asian (K = 3)
Adj. Suma 1,242 48% Female 78% White7% Hispanic <1% African American

Without national comparison sample. K = Number of studies.

Characteristics of collegiate recovery program studies (K = 18).

Study Citation N Sample Study design Follow-ups/retention Sex Age Race/ethnicity Primary substance of focus
1 Ashford et al. (49) 54 Undergraduate and graduate students participating in the UNT Psychology Sona System who were also in recovery. Comparison between those “actively” engaged in CRP (n = 14) vs. not engaged (n = 40). Single-group retrospective; cross-sectional NA 64.3% Female28.6% Male M = 27 NR NR
2 Brown et al. (50) 88 Self-reported alumni of CRPs at institutions of higher education in the United States Single-group retrospective; cross-sectional  NA 44.3% Female54.5% Male M = 30.13 (SD = 6.89) White = 96.6%Black = 1.1%Asian = 1.1%Native American = 1.1%Hispanic = 3.4% Alcohol (21.6%); Opiates (23.9%); Cocaine (4.5%); Amphetamines (2.3%); Marijuana (5.7%); Multiple (33%); Other (9.1%)
3 Castedo de Martell et al. (51) Secondary data from 2 national surveys of CRPs at the programmatic level (mean budget: 54 CRPs and mean membership size: 20 CRPs). Average relapse rates were drawn from a 2012 survey of 29 CRPs across 19 US states. 20 schools provided sufficient information to estimate annual CRP budget. Mean membership size estimated from 2017 CRP survey. Cost effectiveness  NA NA NR NR NA
4 Gerber et al. (52) Costs derived from previous work by De Martell (2019). Utilizing her findings of the average CRP budget, the costs were $191,389. The mean room and board cost of $9,488 was used. National average alumni giving was determined for the US: an average of $3,896 was given by alumni to their alma maters. Nationally, 32% of alumni give to their respective alma maters, but individuals volunteering in college gave closer to 79%. Cost effectiveness  NA NA NR NR NA
5 Hennessy et al. (48) 861 Students enrolled in CRPs; secondary data analysis from 5 CRPs who provided data operating across the US within four-year large, public schools Single-group retrospective; longitudinal Varied 44% Female56% Male M = 27 (SD = 8) White = 89.0%Black = 2.66%Asian = 3.39%Hispanic = 0.48% Varied
Nichols et al. (53) 861 Single-group retrospective; cross-sectional 44.2% Female56% Male M = 26.99 (SD = 7.6) White = 88.9%Black = 3.0%Asian = 0.5%Native American = 0.3%Hispanic = 3.8% Only reported in 188/861 cases: Alcohol (47.3%) Marijuana (11.2%) Opiates (23.9%) Hallucinogens (1.6%) Stimulants (11.2%) Other (4.8%)
6 Smith et al. (54) 185 Students from 28 CRPs nationwide Single-group prospective; longitudinal 45.4% Female54.6% Male M = 29.09 (SD = 9.42) NR Alcohol as their primary substance of choice (38.4%)
Smith et al. (55) 334 CRP students at 4-year universities and community colleges in the US and Ontario, Canada. Single-group prospective; longitudinal 59% 51% Female31% Male M = 29.48 (SD = 9.62) White = 84.0%Black = 6.0%Asian = 2.7%Native American = 1.5%Hispanic = 5.1% Nearly all with polysubstance use history: alcohol (93%) and cannabis (93%) most commonly usedMajority identified alcohol (40%), opioids (22%), or cannabis (14%) as primary substance of choice
Vest et al. (56) 435 Students from 51 CRPs across the US and Ontario, Canada from 4 cohorts (Fall 2020, Spring 2021, Fall 2021, and Spring 2022). Part of the National Longitudinal Collegiate Recovery Study. Single-group prospective; cross-sectional 53.6% Female28.4% Male M = 27.65 (SD = 6.61) White = 81.7% NR
7 Smith (57) 16 Current members of the KSU CRP (CYAAR) atlarge Georgia university Single-group retrospective; cross-sectional NR NR NR NR
8 Botzet et al. (58) 83 Current students (n = 46) and alumni (n = 37) of StepUP recovery program: provided (a) drug and alcohol-free living, (b) weekly 1-on-1 and group meetings to discuss recovery and school-related issues, (c) individual abstinence contracts, and (d) drug-free social activities. Other; longitudinal 7 months (NR) 35% Female65% Male M = 22.5 White = 97.6% NR
9 Cleveland et al. (4) 82 Student members of the TTU CRC between March 2004 and February 2005 Single-group retrospective; cross-sectional 38% Female62% Male M = 23.2 (SD = 6.2) White = 95.0% Non-Alcohol Drug dependence (n = 47; Score: M = 1.44, SD = 0.43).Alcohol dependence (n = 27; Score: M = 1.00, SD = 0.48).Multiple Substance Use: 4 indicated alcohol as both their 1st and 2nd drug of choice.Eating Disorders: Several participants identified eating disorder as primary addiction.
10 Cleveland & Harris (59) 55 Student members of a CRP supported by a large public university in the southwestern US between 2004 and 2005 (60 recruited – 5 excluded from analysis). Single-group prospective; longitudinal (1) Baseline; (2) End-of-day daily diaries, max 33 days (92%) 29% Female71% Male M = 22.6 (SD = 5.7) White = 98%Black = 0%Asian = 0%Native American = 1.8%Hispanic = 0% Primary Addictions: 18% alcohol only (N = 10), 31% nonalcohol drug only (N = 17), 49% nonalcohol drug with alcohol (N = 27), 5% marijuana (N = 3), 2% food (N = 1)
Wiebe et al. (60) 55 30.77% Female69.23% Male
Knapp et al. (61) 50 50 members of a twelve-step CRC located in a southwestern U.S. public university, ages 18–29 and full-time students. Original sample was 55 but 5 were excluded for age or non-SUD recovery status. Single-group prospective Daily diaries for 3 weeks 29% Female71% Male M = 21.42 White = 98% Alcohol was 1st/2nd drug of choice for 52% (n = 26); marijuana for 50% (n = 25), stimulants for 40% (n = 20), opiates for 26% (n = 13), and club drugs or hallucinogens for 32% (n = 16)
11 Laudet et al. (11) 486 CRP-enrolled students from 29 different CRPs in 2013 Single-group retrospective; cross-sectional 42.8% Female57.2% Male M = 26.2 (SD = 8.19) White = 91.3%Black = 1.9%Asian = 1.0%Native American = 1.0%Hispanic = 5.0% Alcoholism (38.9%), Drug addiction (52.6%), behavioral addictions (7.1%), “other” addictions(1.3%); 41.2% indicated alcohol as primary substance, 0.2% indicated sedatives (e.g., barbiturates), 3.3% indicated tranquilizers or anti-anxiety drugs, 11.1% indicated pain relievers (e.g., Codeine), 6.9% indicated stimulants (e.g., uppers), 10.0% indicated marijuana, 11.7% indicated cocaine or crack, 2.0% indicated hallucinogens (e.g., Ecstasy), 0.2% indicated inhalants or solvents (e.g., amyl nitrate), 11.9% indicated heroin, and 1.5% indicated performance enhancing drugs (e.g., steroids, HGH).
Laudet et al. (62) CRP-enrolled students from 29 different CRPs in 2013 Single-group retrospective; cross-sectional NR NR NR NR
12 Patterson et al. (63) 40 Students involved in CRC, defined as “participating in recovery programming or meetings at least 2/week.” Participants used CRC average of 8.40 h/week (SD = 11.23). Single-group retrospective; cross-sectional 57.5% Female42.5% Male M = 22.45 (SD = 4.22) White = 72.5%Black = 2.5%Asian = 7.5%Native American = 2.5%Hispanic = 15.0% Recovery from alcohol/substance use (52.5%), mental health disorders (30%), sex/pornography (22.5%), self harm (10%)15% identified as in recovery from two addictions.
13 Wattick et al. (64) 13 Students enrolled in CRP at West Virginia University in 2018; focus on culinary and nutrition component Single-group retrospective; cross-sectional 61.5% Female38.5% Male NR White = 76.9% NR
14 Odefemi-Azzan (65) 222 Students from an ethnically diverse university in the southeast who applied to the CRP (2007–2017) Single-group retrospective; cross-sectional 37.1% Female62.9% Male M = 24 White = 96.3%Black = 1.2% NR
15 Bennett et al. (66) 78 Students who received 6 months of treatment at the New Jersey Collegiate Substance Abuse Program Single-group prospective; longitudinal 6 months (70.51) 48.7% Female51.3% Male M = 22.12 (SD = 3.71) NR Alcohol (41.9%); Alcohol/drug combinations (37.8%)
16 Tuliao et al. (67) 162 At least 18 years old, enrolled in CRP for a SUD from 43 colleges universities with a CRP Single-group retrospective; cross-sectional 44% Female56% Male M = 27.01 (SD = 7.49) White = 83.0%Black = 2.0%Asian = 3.0%Hispanic = 1.0% NR
17 Herold (68) 31 18 years or older, a student at a college or university, and actively participating in at least 1 element or service of CRP at least 1/week. CRP must be ARHE member Qualitative/Mixed methods; cross-sectional 64.5% Female35.5% Male NR White = 71.0%Black = 3.2%Asian = 3.2%Native American = 3.2%Hispanic = 19.4% Alcohol (87.1%), Opiates (16.1%), Benzodiazepines (9.7%), cocaine or crack-cocaine (22.7%), methamphetamine (25.8%), other stimulants (9.7%), hallucinogens or psychedelics (25.8%), f synthetic or “club’ drugs (12.9%), ecstasy/MDMA (16.1%), marijuana (54.8%)
18 Hamner (69) 89 Enrolled college students affiliated with ARHE and/or SAFE Project Single-group retrospective; cross-sectional 43.2% Female46.6% Male M = 30.45 (SD = 10.43) White = 78.4%Hispanic = 10.2% NR
Summary 2795 12 cross-sectional;4 longitudinal 45% Female(k = 15) 26.28 years(k = 13) White = 89%(k = 13)Black = 2.73%(k = 10)Asian = 2.68%(k = 9)Native American = 0.09%(k = 7)Hispanic = 5.04% (k = 9)

ARHE, Association of Recovery in Higher Education; NR, not reported; SAFE, stop the addiction fatality epidemic; K, number of studies.

 Description of recovery high school studies Types of research designs

Of the seven RHS studies, there were two comparative studies: (1) a dissertation by Adam Knotts in 2017 (39) and (2) another study from Finch and colleagues in 2017 [Finch and colleagues (40)] was the main publication with a series of additional manuscripts focused on different subsets of the population or research questions, including a publication led by Weimer utilizing cost-benefit analysis (41–45).1 There was also one cross-sectional mixed methods study that explored differences between RHS and non-RHS groups (48), three cross-sectional single-group studies of just RHS students (38, 46, 47), and one longitudinal single-group study of RHS students (37).

 Study characteristics

Study samples ranged in size from just six persons up to 13,261. Tanner-Smith and colleagues published multiple studies comparing their sample of RHS attendees with a previously-collected national comparison sample of US adolescents receiving treatment for AOD use treatment, but who attended mainstream high school (44). This inclusion inflated the sample demographics across several categories to be more nationally representative. Including this sample, participants were predominantly White (41%), with Hispanic (28%), Black (14%), and Native American (3%) participants represented at rates generally consistent with the national United States demographics. Females represented only 29% of the overall participant sample. Without this national sample, participants were predominately White (78%; range, 71%–90%), with only 7% Hispanic and <1% Black participants included. Studies also had about equal female and male participants overall (M = 48% female; M range, 29%–57%). The average age was 17.3 years (M range, 13–21).

 Study quality

Overall, there were three studies of moderate and four of high risk of bias; no studies met the criteria for low risk of bias (See Figures 2, 3). Blinding of participants to the research question and of data collection by the outcome assessor were most likely to have the lowest risk of bias ratings, at primarily moderate. Regarding the potential for selection bias, individuals who were selected to participate were rated as either somewhat or very likely to be representative of the target population, but only a single study reported the percentage of individuals who agreed to participate after selection (46). In terms of appropriately addressing the potential confounders defined by the EPHPP tool, the studies were split between low risk of bias (k = 3) or high risk of bias (k = 4). Those rated as high risk of bias did not measure potential confounders, or if they did, did not include them in their analytic approach to examining outcomes. For example, of the list of eight common confounders (race, sex/gender, marital status/family, age, income/class, education, health status, and pre-intervention score on outcome), only 3 studies controlled for race and/or sex/gender, only 2 studies controlled for age and pre-baseline scores on outcomes (e.g., AOD use), and only a single study controlled for SES, education, or health status of participants.

Risk of bias for each individual study: RHS.

Risk of bias summary across studies: RHS.

 Recovery high school student outcomes Overview

Six of the seven studies examined some aspect of substance use (e.g., frequency, abstinence), with all but one (38) using primarily self-report. Four studies examined academic outcomes, with only two using official school records (38, 39). Two studies examined involvement with the criminal justice system via self-reported delinquency (42) or legal issues (46). Two studies examined social networks, with one focusing on conflict and perceptions of social support (43) and the other focusing on the alcohol and drug use behaviors of social groups within their networks (48). Two studies examined mental (but not physical) health outcomes, and a single study examined life satisfaction/flourishing. A single study examined work status (full vs. part-time) and number of hours worked. No studies examined measures of Recovery Capital.

 Comparative studies

This section details the adolescent outcomes among the studies of RHSs with the most rigorous study design in this sample, two comparative studies of RHS, one of which includes a cost-benefit analysis driven from the findings of the researchers' quasi-experimental study. In addition to having the most rigorous study design, they had the lowest risk of bias (moderate) compared to the majority of other studies. The other study designs and their associated outcomes are reported in full in Table 3.

Recovery high school study outcomes by study quality: academic, cost-benefit, crime/criminal Involvement, Employment, life satisfaction, mental/physical health, social network change, substance Use (K = 7).

A. (academic, cost-benefit, crime/criminal involvement, employment, life satisfaction)
Study Study design N Retention Risk of bias rating Key findings
Academic Cost benefit results Crime/criminal involvement Employment Life satisfaction
Finch et al. (40) QED; longitudinal Moderate
Glaude et al. (43) QED (Cross-sectional outcomes only) 246 RHS: higher life satisfaction [F(1,242) = 7.626, p < 0.001, ƞ2 = .071, η2 = .071]RHS: M = 3.71non-RHS: M = 3.35Males: M = 3.60, SD = 0.68; range = 1.50, 5.0.Females: M = 3.45, SD = 0.64; range = 1.83, 4.67.ANOVA: sex differences were not significant.
Tanner-Smith et al. (44) QED; longitudinal
Tanner-Smith et al. (41) QED; longitudinal 194 78%
Tanner-Smith et al. (42) QED; longitudinal 260 81% (6 m) and 71% (12 m) 6 mo.:No difference between RHS and non-RHS: illegal activity days (b = −0.43, 95% CI [− 0.91, 0.05] and lower intoxicated illegal activity [b = − 1.10, 95% CI (− 1.89,−0.29), IRR = 0.34, d = 0.37]12 mo.:No difference between RHS and non-RHS: illegal activity days (b = −0.17 95% CI [− 0.62, 0.28].and lower intoxicated illegal activity [b = − 0.87, 95% CI (− 1.57,−0.18), IRR = 0.42, d = 0.27].
Weimer et al. (45) Cost effectiveness 260 80.6%(6-mo.); 71.4% (12-mo.) RHS: graduated at higher rates (53.8%) compared to non-RHS (34.2%).Entire analytic sample (n = 260) assuming all students with missing data did not graduate: estimated adjusted impact of RHS = 13.6 percentage points (B = .136; SE = .077).Cases for which graduation status was known and student old enough to graduate (n = 164): estimated impact of RHS = 23.9 percentage points (SE = 0.088).Excluded Texas cases (n = 153): estimated impact of RHS = 19.1 (SE = 0.08).Excludes unknown and Texas cases: estimated RHS impact = 26.8 percent (SE = 0.085). Mean net benefits ranged from $16.1 thousand to $51.9 thousand per participant; benefit-to-cost ratios ranged from 3.0 to 7.2.
Hensel & Wilburn (37) Single-group prospective; longitudinal 6 81.2% (147/180 EMAs) Weak
Giles (38) Single-group retrospective; cross-sectional 375 Weak <10% of students had an unsuccessful academic outcome, 1.07% left school because they moved out of the area.52% transitioned to their sending school in good standing with the MA-RHS, 28.27% graduated HS, 9.87% earned their HiSET. Other students moved, went to treatment, withdrew, or entered the criminal justice system.
Knotts (39) QED; longitudinal 57 Moderate RHS: similar levels of academic growth compared to a nationally-representative matched Virtual Comparison Group (t-stat =  + 0.849; p = 0.397). Average growth of RHS outpaced VCG in reading (+0.201, p = 0.278) and mathematics (+0.019, p = 0.914), but not language (−0.005, p = 0.977).49 cases where a student relapsed: average growth outpaced the VCG in reading (+0.443, p = 0.299) and math (+0.574, p = 0.084), but not language (−0.146, p = 0.623).92 cases where the student did not relapse: average growth outpaced VCG in reading (+0.087, p = 0.634) and language (+0.087, p = 0.631), but not math (−0.214, p = 0.280).
Lanham & Tirado (46) Single-group retrospective; cross-sectional 72 Moderate 65 (90%) had some collegeSix (9%) earned bachelor's or master's degrees, 7 (10%) earned 2-year associate's degree 9 (20%) of nonabstinent respondents: currently facing criminal charges or sentencing4 (9%) were in a controlled environment in last 30 days. Abstinent respondent: 1 facing criminal charges/sentencing (completed survey from prison) ASI composite score: Legal Status (M = 0.119 SD = 0.226). Majority worked at least part time. Number of work days in last 30 days (M = 20.4, SD = 2.4; range = 1, 30)
Moberg & Finch (47) Single-group retrospective; cross-sectional 321 Weak
Hennessy et al. (48) Mixed methods; cross-sectional 28 Weak

Bold indicates a “parent” study of a series of linked studies. HiSet, High School Equivalency Diploma.
B. (mental/physical health, social network change, substance Use)
Study Study design N Retention Risk of bias rating Key findings
Mental/physical health Social network change Substance use
Finch et al. (40) QED; longitudinal Moderate 6-months:RHSs: more likely to be abstinent from all AODs (b = 1.47, 95%CI [0.11, 2.69], OR=4.36, 95% CI [1.19, 15.98]). RHS predicted probability of abstinence (.58) vs. non-RHS (.30)RHS: fewer days of marijuana use (adjusted regression model [b = −14.38, 95% CI (−27.70, −1.06)]:RHS Mean = 8.84, SD = 22.94Non-RHS Mean = 25.82, SD = 34.90RHS fewer days of alcohol use than non-RHS (adjusted model for propensity score, baseline scores, referral sites [b = −2.07, 95% CI (−6.56, 2.41), d = −0.23]:RHS Mean = 2.01, SD = 6.82Non-RHS Mean = 5.43, SD = 12.33RHS less frequent use of other substances (adjusted regression model [b = −6.77, 95% CI (−14.80, 1.27), d = −0.45]):RHS Mean = 3.18, SD = 11.66Non-RHS Mean = 7.08, SD = 20.67
Glaude et al. (43) QED (Cross-sectional outcomes only) 246 No difference in perceptions of social support based on RHS/non-RHS, [ANOVA: F(2,242) = 2.457, p = 0.169].Males: higher perceptions of social support (M = 8.26, SD = 1.09; range = 3.0–9.0). Females: lower perception of social support (M = 8.01, SD = 1.18; range = 3.0, 9.0)
Tanner-Smith et al. (44) QED; longitudinal
Tanner-Smith et al. (41) QED; longitudinal 194 78% 6 mo.: students in both conditions reported less MH symptoms than at baseline. Adjusting for baseline MH and propensity scores: no differences in odds of MH symptoms at follow-up for RHS vs. non-RHS.RHS: lower rates than non-RHS of anti-social personality disorder (9 vs. 18%), hypomanic episodes (8 vs. 10%), suicide risk (14 vs. 19%).RHS: higher levels of symptomatology than non-RHS on major depression (49 vs. 42%), GAD (44 vs. 37%), OCD (7 vs. 3%), panic disorder (25 vs. 23%), and PTSD (26 vs. 10%).
Tanner-Smith et al. (42) QED; longitudinal 260 81% (6 m)71% (12 m) RHS: less AOD use than non-RHS students at 6-mo. [b = − 0.58, 95% CI (− 1.21, 0.04), IRR = .56, d = 0.19] and 12-mo. [b = −0.50, 95% CI (− 0.98,−0.02), IRR = .60, d = 0.21].RHS: AOD rates 0.60 times smaller than non-RHS students.12-mo., predicted number of AOD use days in past 90:RHS: 13.86Non-RHS: 25.67NPSS moderated effect of RHS on AOD use outcomes, with RHSs providing minimal beneficial effects for those students endorsing maladaptive problem solving styles. NSPS styles moderated effect of RHS on AOD use at 6-mo. [b RHS ×NPS = 2.22, 95% CI (0.70, 3.75), IRR = 9.23].
Weimer et al. (45) Cost effectiveness 260 80.6% (6-mo.)71.4% (12-mo.) 12-mo. abstinence:RHS: 54.7% past 3-mo.Non-RHS: 25.6% past 3-mo.12-mo. alcohol use days:RHS: M = 11.5, SD = 28.4Non-RHS: M = 21.8, SD = 39.812-mo. marijuana use days:RHS: M = 43.3, SD = 76.7Non-RHS: M = 119.4, SD = 130.7Urinalysis confirmed THC (cannabis) abstinence for 68% of RHS vs. 37% of non-RHS.12-mo. other drug use days:RHS: M = 16.5, SD = 37Non-RHS: M = 21.3, SD = 54.4
Hensel & Wilburn (37) Single-group prospective; longitudinal 6 81.2% (147/180 EMAs) Weak 6% of urge days involved drug-use (4/48; NS sample size for additional analysis)Greater daily emotional pain increased odds of drug urges (OR=3.75) and associated with higher urge intensity (OR=2.92); Higher positive mood associated with urge to use (OR=1.36), urge frequency (OR=1.17), and urge intensity (OR=1.18); More somatic symptoms associated with greater odds of drug urges (OR=1.26) and greater odds of urge frequency during the day (OR=1.10); Greater daily emotional pain associated with higher within-day urge frequency (OR=2.42);
Giles (38) Single-group retrospective; cross-sectional 375 Weak At discharge: 29.87% not using any AOD. 54.13% reduced use to 1/month. 16% engaged in heavy use or used some substance almost daily.
Knotts (39) QED; longitudinal 57 Moderate Of 141 total testing periods, in 49 of the cases the student relapsed (remaining 92 cases did not)
Lanham & Tirado (46) Single-group retrospective; cross-sectional 72 Moderate 28 (39%) past 30-day abstinence.3 (4%) nonproblem use.Abstinent or consumed alcohol but no illicit drugs in last 30 days: 43 (60%).Among drinkers: 9.2 drinking days in previous 30 (SD = 7.02).Cannabis most used drug (n = 23), followed by sedatives/hypnotics tranquilizers (n = 8), opiates analgesics (n = 6), cocaine (n = 4), amphetamines (n = 3), heroin (n = 2), hallucinogens (n = 2), barbiturates (n = 1)ASI composite scores:Alcohol Use (M = 0.07, SD = 0.094)Drug Use (M = 0.08, SD = 0.115)
Moberg & Finch (47) Single-group retrospective; cross-sectional 321 Weak Retrospective report (n = 291): bothered by any nervous, mental or psychological problem reduced from 69% to 33%.Feel very trapped, lonely, sad, blue, depressed or hopeless about the future (Before:73, Now: 31; p < .001).No energy, losing interest in work, school, friends, sex or other things you cared about? (Before: 60, Now: 20; p < .001).Thought about ending your life or committing suicide (Before: 53, Now: 16; p < .001)Felt very anxious, nervous, tense, fearful, scared, panicked (Before: 68, Now: 44; p < .001)Trembling, heart racing, restless. (Before: 60, Now: 40; p < .001)Very distressed, upset when reminded of the past (Before: 65, Now: 55; p = .003)Hard time expressing feelings, even to people you cared about. (Before: 83, Now: 49; p < .001)Hard time paying attention at school, work or home (Before: 86, Now: 63; p < .001)Unable to stay in a seat or where you were supposed to be (Before: 71, Now: 41; p < .001)Bothered by any nervous, mental or psychological problems? (Before: 69, Now: 33; p < .001)Disturbed by memories of things from the past that you did, saw or happened to you? (Before: 76, Now: 55; p < .001) Retrospective report (n = 291): weekly AOD use decreased from 90% in the 12 months before admission, to 7% since admission (p < .001). Retrospective report (n = 174 students in school at least 90 days): 28.5 days abstinent (SD = 36.8) in the 90 days before entering RHS and 266.1 days abstinent (SD = 258.8) since entering RHS (p < .001). Alcohol days (p < .001):Before: M = 33.6 (SD = 35.4) After: M = 3.5 (SD = 11.2)Drank ≥5 drinks at a time (p < .001):Before: M = 31.3 (SD 34.2)After: M = 3.0 (SD = 11.1)Cannabis days (p < .001):Before: M = 47.2 (SD = 40.0) After: M = 3.1 (SD = 12.3)Other drugs days (p < .001):Before: M = 30.9 (SD = 37.3) After: M = 2.8 (SD = 10.5)Smoked cigarettes, last 30 days (p < .001):91% at follow-up (67% daily)Limited to when they started school (not necessarily past 90 days): Percent abstinent from AOD (p < .001): Before: 20; After: 56Percent abstinent from alcohol (p < .001): Before: 24; After: 62Percent abstinent from cannabis (p < .001): Before: 30; After: 71Percent abstinent from other drugs (p < .001): Before: 40; After: 74Percent days abstinent (p < .001): Before: 32; After: 82
Hennessy et al. (48) Mixed methods; cross-sectional 28 Weak NETWORK CHARACTERISTICSAverage degree of conflictRHSs: 5.26 (SD = 3.72)Non-RHS: 5.50 (SD = 3.78)Total sample: 5.33 (SD = 3.67)Heavy alcohol use ratioRHS: 0.20 (SD = 0.33)Non-RHS: 0.18 (SD = 0.10)Total sample: 0.19 (SD = 0.29)Casual alcohol use ratioRHS: 0.28 (SD = 0.19)Non-RHS: 0.47 (SD = 0.17)Total sample: 0.33 (SD = 0.20)No alcohol use ratioRHS: 0.21 (SD = 0.17)non-RHS: 0.09 (SD = 0.13)Total sample: 0.18 (SD = 0.17)Recovering from alcohol use ratioRHS: 0.19 (SD = 0.18)non-RHS: 0.07 (SD = 0.08)Total sample: 0.16 (SD = 0.16)Heavy drug use ratioRHS: 0.30 (SD = 0.35) non-RHS: 0.24 (SD = 0.16).Total sample: 0.28 (SD = 0.30)Casual drug use ratioRHS: 0.19 (SD = 0.17)non-RHS: 0.23 (SD = 0.14)Total sample: 0.20 (SD = 0.16)No drug use ratioRHS: 0.19 (SD = 0.15)non-RHS: 0.26 (SD = 0.14)Total sample: 0.21 (SD = 0.15)Recovering from drug use ratioRHS: 0.17 (SD = 0.17)non-RHS: 0.03 (SD = 0.07)Total sample: 0.13 (SD = 0.16)

Bold indicates a “parent” study of a series of linked studies.

AOD, alcohol and other drug; Mo., Month; NPSS, negative problem-solving styles.

One dissertation study conducted by Knotts (39) enrolled 57 students from an Illinois RHS, Hope Academy (2010–2017) and matched 43 of these students with a nationally representative virtual comparison group of students from the 2015 Northwest Evaluation Association's Measure of Academic Progress normative data. Matching was conducted to ensure that students were from similar schools (percentage of students receiving free/reduced-price lunches and school location) and that there was a match for each RHS student on gender, ethnicity, subject area, starting score, grade level, and testing timeframe. When examining these academic outcomes using paired t-tests, RHS students displayed similar levels of academic growth when compared to the virtual comparison group (t-stat =  + 0.849; p = 0.397), including on reading (+0.201, p = 0.278), mathematics (+0.019, p = 0.914), and language (−0.005, p = 0.977). Substance use was only assessed among RHS students as the virtual comparison group did not have this assessment. The RHS students were administered the Global Assessment of Individual Needs-Short Screen (GAIN-SS) every eight to twelve weeks. This resulted in 141 instances of the GAIN-SS in the RHS sample: in 49 of these, the student had returned to use, while in the remaining 92 instances, they had not returned to substance use.

In the only other comparative study of RHS students, 294 youth in recovery were recruited from RHS and treatment centers from 2011 to 2016 (40–45). Their school enrollment choice (after treatment, if enrolled from a treatment center) was collected, and they were followed for one year. Matching was conducted to equate the non-RHS and the RHS youth for the analysis. Of the analyzed sample of 260 youth, RHS students reported less substance use than non-RHS students at the 6-month (b = − 0.58, 95% CI [− 1.21, 0.04], Incidence Rate Ratio [IRR] = .56, d = 0.19) and 12-month follow-ups [b = −0.50, 95% CI (− 0.98,−0.02), IRR = .60, d = 0.21]. Students who attended RHS reported substance use rates approximately 0.60 times lower than non-RHS students. At the 12-month follow-up, the number of days predicted by the negative binomial models for use of substances in the past 90 days was 13.86 for RHS students vs. 25.67 for non-RHS students. RHS students reported less frequent delinquent behavior while intoxicated and fewer days of substance use relative to students attending non-RHSs. Urinalysis was used to examine abstinence from THC (cannabis): 68% of RHS students vs. 37% of comparison students were abstinent. In the cost-benefit analysis from 260 students in this sample (45), the mean net benefits ranged from $16.1 thousand to $51.9 thousand per participant, indicating a large cost savings by implementing the RHS for youth struggling with substance use; benefit-to-cost ratios ranged from 3.0 to 7.2. In addition, RHS students' high school graduation rates were 21–25 percentage points higher than comparison students.

 Description of collegiate recovery program studies

Of the 18 CRP studies, there were two cost-benefit (51, 52) and 16 single-group studies, the majority of which were cross-sectional (75%).

Study characteristics

The demographic characteristics for the 17 studies that were not cost-benefit studies indicate a high degree of heterogeneity across the included studies. Study samples ranged in size from 13 to 861 participants, although the largest sample size was from a secondary data analysis that synthesized data across five CRPs (13). Participants were predominately white (89%; range, 71%–98%). Studies had slightly less female than male participants overall (M = 45% female; M range, 29%–66%). The average age was 26.28 years (M range, 22–30).

 Study quality

Overall, the 16 studies that did not use cost-benefit analysis had variable quality (See Figures 4, 5). The majority were rated as high risk of bias (k = 11), primarily due to selection bias, study design, and the way confounders were addressed. All studies were rated as moderate due to their approach to blinding of participants to the research question and of the outcome assessors to the condition of the participants. A subset of five studies were rated as low risk of bias in data collection methods, with a further eight studies at moderate risk of bias as many used validated and reliable scales or a combination of these with measures developed for their study purposes.

Risk of bias for each individual study (K = 16): CRP.

Risk of bias summary across studies (K = 16): CRP.

The quality rating of the two studies using cost-benefit designs indicated that overall both studies had fairly rigorous approaches (See Table 4). The study by Castedo de Martell and colleagues (51) achieved a score of 92. Factors influencing this score were a lack of adjusting for longitudinal outcomes and not discounting benefits over time. The study by Gerber and colleagues (52) achieved a score of 70, primarily due to poor reporting of various aspects of the approach, including a lack of specifying and justifying the modeling approach, using insufficiently or overly optimistic assumptions (e.g., rates of volunteerism) and relying heavily on the earlier cost work by Castedo de Martell (51) without sufficient rationale for several indicators.

Results of the grading assessment of the quality of the CRP cost-effectiveness studies (K = 2).

Criteria Castedo de Martell et al. (51) Gerber et al. (52)
Weight Coding Weighted score Coding Weighted score
1. Was the study objective presented in a clear, specific, and measurable manner? 7 1 7 1 7
2. Were the perspective of the analysis (societal, third-party payer, etc.) and reasons for its selection stated? 4 1 4 1 4
3. Were variable estimates used in the analysis from the best available source (i.e., Randomized Control Trial—Best, Expert Opinion—Worst)? 8 1 8 1 8
4. If estimates came from a subgroup analysis, were the groups prespecified at the beginning of the study? 1 NA NA
5. Was uncertainty handled by: (1) statistical analysis to address random events; (2) sensitivity analysis to cover a range of assumptions? 9 1 9 1 9
6. Was incremental analysis performed between alternatives for resources and costs? 6 1 6 0 0
7. Was the methodology for data abstraction (including value health states and other benefits) stated? 5 1 5 1 5
8. Did the analytic horizon allow time for all relevant and important outcomes? Were benefits and costs that went beyond 1 year discounted (3%–5%) and justification given for the discount rate? 7 0 0 1 7
9. Was the measurement of costs appropriate and the methodology for the estimation of quantities and unit costs clearly described? 8 1 8 0 0
10. Were the primary outcome measure(s) for the economic evaluation clearly stated and were the major short term, long term and negative outcomes included? 6 1 6 1 6
11. Were the health outcomes measures/scales valid and reliable? If previously tested valid and reliable measures were not available, was justification given for the measures/scales used? 7 1 7 1 7
12. Were the economic model (including structure), study methods and analysis, and the components of the numerator and denominator displayed in a clear transparent manner? 8 1 8 0 0
13. Were the choice of economic model, main assumptions and limitations of the study stated and justified? 7 1 7 0 0
14. Did the author(s) explicitly discuss direction and magnitude of potential biases? 6 1 6 1 6
15. Were the conclusions/recommendations of the study justified and based on the study results? 8 1 8 1 8
16. Was there a statement disclosing the source of funding for the study? 3 1 3 1 3
Summary Scores 92 70

1 = Yes. 0 = No. NA = Not applicable.

 Collegiate recovery program student outcomes Overview

Nine studies examined substance use outcomes and all used self-report assessments. Five additional studies did not examine substance use directly but instead inquired about length of time in recovery or substance craving (15, 49, 59, 64, 69). Seven studies examined academic outcomes, with the majority focused on GPA (4, 13, 15, 49, 55, 57, 62, 65). Five studies assessed various metrics related to CRP participation, including duration of participation, perceived helpfulness or program benefits, engagement in recovery-related activities, and whether the CRP was the primary pathway to recovery (50, 54, 55, 58, 62, 67). Five studies examined mental and/or physical health outcomes, with the majority assessing the history of such problems not limited to the resolution of problems during the participant's time in the CRP (11, 53, 58–60, 66). Two studies examined stress among participants related to their current experience in the CRP (54, 58), and another study examined “good health and the absence of depression symptoms” as part of their CRP culinary programming component (64). Four studies (49, 50, 54–56, 65, 69) examined recovery capital using either the 50-item Assessment of Recovery Capital (72) or its 10-item version, the BARC (73). Five studies (49, 50, 54–56, 64, 68) examined life satisfaction by measuring self-reported quality of life (k = 3), personal growth (k = 1), resilience (k = 1), and/or flourishing (k = 1). Four studies examined aspects of social networks: (i) social network change (58), (ii) whether the CRP plays an important role in their social life (55), and (iii) in-group nominations (63). Two additional reports from the same study examined whether and where (CRC setting or not) a participant talked with others about their recovery and how this related to their craving (59), as well as whether stress and negative affect changed who CRC students connected with (family, sponsor, CRC peers) and how frequently (61). Two studies examined employment by collecting whether participants worked during the semester and examining hours worked (4, 56). A single study focused on criminal involvement and examined self-reported “legal severity” (66) and another study analyzed several recovery outcomes by comparing them across student incarceration history (56).

 Cost-benefit studies

This section focuses on the student outcomes among the two studies of CRPs that utilized cost-benefit analyses (51, 52). The other 16 studies, study designs, and their associated outcomes are reported in Table 5.

Study outcomes by study quality: collegiate recovery program engagement and satisfaction, substance use, recovery capital, social connectedness, employment/education, criminal justice involvement, quality of life and well-being (K = 19).

A. (academic, cost-benefit, crime/criminal involvement, employment, recovery capital)
Key findings
Study Study design N Retention Study quality rating risk of bias Academic Cost-benefit Crime/criminal involvement Employment Recovery capital
Ashford et al. (49) Single-group retrospective; cross-sectional 54 Weak All students in recovery (n = 54): GPA = 3.482 (SD = 0.446)Students in recovery and CRP members: GPA = 3.686 (SD = 0.338) Correlations between total ARC scores and CRP scholarship assistance services (rs = .557, p = .039)
Bennett et al. (66) Single-group prospective; longitudinal 78 70.51 Weak ASI Legal SeverityBaseline (M = 0.04, SD = 0.12)6 Months (M = 0.02, SD = 0.06); F = 3.81(n = 55)
Brown et al. (50) Single-group retrospective; cross-sectional 88 Weak Recovery capital M = 45.90 (SD =4.21) Scores not significantly different among gender χ2 (28) = 21.57, p = .801, race χ2 (42) = 34.91, p = .772, ethnicity χ2 (14) = 13.83, p = .462, education level χ2 (14) = 21.33, p = .093, or income level χ2 (42) = 50.33, p = .177.ARC scores positively correlated with human flourishing scale (FS-8 scores) (r = .670, p < .001), and scores from all domains of the WHOQOL BREF (DOM1, r = .371, p < .001; DOM2, r = .741, p < .001; DOM3, r = .606, p < .001; DOM4, r = .470, p < .001).
Botzet et al. (58) Other; longitudinal 83 Weak
Castedo de Martell et al. (51) Cost effectiveness 92 The cost of operating a CRP in the base case is $97,586.24 less than the cost of TAU and adds just over 25 QALYs. The ICER for the societal model in the base case is -$3,872.75, or a cost savings of $3,872.75 per QALY gained when implementing a CRP.
Cleveland et al. (4) Single-group retrospective; cross-sectional 82 Moderate 22.5% GPA higher than 3.7552.5% GPA of 3.25 or better Half of CRC members also work during the semester: 7.4% work >40 h/week, 27.2% work 20 h/week, 14.8% work 10 h per week
Cleveland & Harris (59) Single-group prospective; longitudinal 60 Moderate
Wiebe et al. (60) Single-group prospective; longitudinal 55 92
Knapp et al. (61) Single-group prospective; longitudinal 50 87
Gerber et al. (52) Cost effectiveness 70 Societal CB ratios = 1.94Institutional CB ratio = 1.38Societal IRRs = 249%Institutional IRR = 77%The benefits of a CRP significantly outweigh the costs
Hamner (69) Single-group retrospective; cross-sectional 89 Moderate Recovery capital M = 52.31 (SD = 5.973) Time in Recovery accounted for 9% of the variance explained in Recovery Capital: people with less time in recovery scored lower in than those with more time.Relational Transparency accounted for 8.6% of the variance in Recovery Capital: the higher they scored their mentor on Relational Transparency, the higher they scored their own Recovery Capital (b = 0.293)Those below the median for Time in Recovery had lower Recovery Capital (M = 50.49) than those above the median [M = 54.02; t(87) = -2.90, p = .005].
Herold (68) Qualitative/Mixed methods; cross-sectional 31 Weak
Hennessy et al. (15) Single-group retrospective; longitudinal 861 Varied Weak Site 1 (n = 196) improved GPA across two consecutive college semesters [SMG = 0.54, 95% CI (0.25, 0.83)]. Two separate cohorts of students at site 3 (n = 24 in each) reported improvements in GPA from pre-to post-recovery: SMG = 0.97 [95% CI (0.46, 1.48)] and SMG = 0.31 [95% CI (0.02, 0.60)]. Site 4 (n = 139) reported slight reductions in GPA between entry into the CRP and the follow-up period, but this effect was not significantly different from zero [SMG = −0.04, 95% CI (−0.71, 0.63)].
Nichols et al. (53) Single-group retrospective; cross-sectional 861
Laudet et al. (11) Single-group retrospective; cross-sectional 486 Moderate
Laudet et al. (62) Single-group retrospective; cross-sectional 486 Current GPA M = 3.22 (SD = .0.62).
Odefemi-Azzan (65) Single-group retrospective; cross-sectional 222 Weak 26 students were alumni—no MH92 students were alumni—at least 1 diagnosis12 students were non-active—at least 1 diagnosisOf alumni with no MH, 100% graduated Having an additional mental health disorder is associated with lower graduation rates.100% of female CRP students—graduated 25.4% of male CRP students—graduated Female students had higher graduation rates than male students (p < .001, df = 1).
Patterson et al. (63) Single-group retrospective; cross-sectional 40 Moderate
Smith et al. (55) Single-group prospective; longitudinal 334 59 Weak CRP helps them academically (M = 71.86, SD = 29.10) BARC-10Baseline (M = 52.66, SD = 7.25)Follow-up 1 (M = 52.26, SD = 6.32)Follow-up 2 (M = 51.67, SD = 7.35)Accounting for the effects of age, race/ethnicity, gender, cohort, and time in recovery, pairwise comparisons indicated that recovery capital at baseline was higher than recovery capital at follow-up 1 (Mdiff = 1.72, p < .001, n = 92) and follow-up 2 (Mdiff = 2.21, p = .025, n = 63).
Smith et al. (54) Single-group prospective; longitudinal 185 Recovery capital (M = 53.27, SD = 6.73, range = 29–60).MANOVAs: students who were abstinent from their primary substance had higher recovery capital [Pillai's Trace = 0.37, F(10, 159) = 99.23, p < 0.001, η2 = 0.367] than those who were not abstinent.Number of hours engaged in recovery-related activities was positively associated with recovery capital scores (ß = 0.21, p = 0.006, R2 = 0.04).
Vest et al. (56) Single-group prospective; cross-sectional 435 GPA:Formerly incarcerated students (M = 3.5, SD = 0.52)Non-LS involved students (M = 3.46, SD = 0.48)non-incarcerated LS-involved students (M = 3.4, SD = 0.58).No significant differences in GPA between groups (F = 1.51, p = 0.222). Hours worked:Formerly incarcerated students (M = 25.74, SD = 16.71)Non-incarcerated LS-involved students (M = 16.70, SD = 16.43)non-LS involved students (M = 14.30, SD = 15.27)Significant difference of hours worked between formally incarcerated students and non-LS involved students and between formally incarcerated students and non-incarcerated LS-involved students (F = 7.48, p = .009). Recovery capital:Formerly incarcerated students (M = 54.47, SD = 6.17)non-incarcerated LS-involved students (M = 54.29, SD = 5.67)non-LS involved students (M = 51.06, SD = 6.76)No significant differences in recovery capital between groups (F = 2.58, p = 0.077).
Smith (57) Single-group retrospective; cross-sectional 16 Weak GPA increased from M = 2.40 (SD = 1.01) prior to entering the CYAAR program to 3.48 (SD = .33), a mean GPA growth of 1.08 (SD = 1.06). GPA increased with length of membership.Of participants with only one semester of membership (n = 5, 31.25%), GPA prior to CYAAR M = 2.84, and the current GPA M = 3.22 (13.38% increase).Participants with 5 or more semesters (n = 6, 37.5%), GPA of M = 1.69 upon entry and current GPA of M = 3.62 (114.2% increase).
Tuliao et al. (67) Single-group retrospective; cross-sectional 162 Weak
Wattick et al. (64) Single-group retrospective; cross-sectional 13 Weak

Bold indicates a “parent” study of a series of linked studies.
B. (social networks, substance Use, mental/physical health, CRP involvement)
Key findings
Study Study design N Retention Study quality rating risk of bias Social networks Substance use Mental/physical Health CRP involvement
Ashford et al. (49) Single-group retrospective; cross-sectional 54 Weak All students in recovery (n = 54): 3.93 years in recovery (SD = 2.604)Students in recovery and CRP members: 3.69 years in recovery (SD = 2.869) Correlations between WHOQOL-BREF psychological domain and CRP scholarship assistance services (rs = .612, p = .020).
Bennett et al. (66) Single-group prospective; longitudinal 78 70.51 Weak 74.5% of participants who completed follow-up were abstinent at 6-months (assuming those who did not complete follow-up returned to use, then abstinence rate = 52.6%)In 90 days prior to 6-month assessment, users remained substance-free M = 79.80 days (SD = 18.10 days, n = l0 of 14 relapsers with complete data; range 30–90 days substance free in last 3 months). ASI alcohol severity (n = 55)Baseline (M = 0.29, SD = 0.22)6 Months (M = 0.12, SD = 0.12); F = 44.66, p < 0.001ASI drug severity (n = 55)Baseline (M = 0.08, SD = 0.09)6 Months (M = 0.03, SD = 0.05); F = 16.17, p < 0.001  ASI medical severity (n = 55)Baseline (M = 0.18, SD = 0.25)6 Months (M = 0.21, SD = 0.24); F = 0.74ASI psychiatric severity (n = 55)Baseline (M = 0.36, SD = 0.21)6 Months (M = 0.22, SD = 0.20); F = 26.97, p < .0001Days of psychological problems in past month (n = 55)Baseline (M = 13.47, SD = 9.96)6 Months (M = 7.42, SD = 8.59); F = 15.37, p < .0001BDI scores (n = 45)Baseline (M = 14.60, SD = 9.74)6 Months (M = 10.04, SD = 9.05); F = 11.64, p < .005
Brown et al. (50) Single-group retrospective; cross-sectional 88 Weak 87.5% (n = 86) reported no recurrences of use since graduation-representing a return to use rate of (10.2%).Of those that returned to use (n = 9), 3 participants experienced one episode, 3 participants experienced two to four episodes, and 3 participants experienced five or more episodes. Flourishing scores (FS-8) M = 49.69 (SD = 6.04).Scores not significantly different among gender χ2 (38) = 29.61, p = .833, race χ2 (57) = 31.01, p = .997, ethnicity χ2 (19) = 39.916, p = .05), education level χ2 (19) = 25.08, p = .158, or income level χ2 (57) = 63.81, p = .250. FS-8 scores were significantly positively correlated with ARC scores (r = .670, p < .001) and all domain scores of the WHO-QOL BREF (DOM1, r = .359, p = .001; DOM2, r = .800, p < .001; DOM3, r = .721, p < .001; DOM4, r = .454, p < .001). 68.2% felt CRP directly prepared them for professional environment; 80.7% felt program directly prepared them for post-graduation recovery. 89.8% felt program membership was helpful academically. 36.4% reported they would not have attended their institution if the program had not been there. 81.8% maintained alumni contact with the undergraduate CRP.
Botzet et al. (58) Other; longitudinal 83 Weak Both student groups M = 7.2 and M = 7.8 of ten possible support sources.Alumni faced slightly greater rate of relapse-risk in present social and work environments than the current students, (MANOVA [F (1,82) = 0.72, p > .05 and F (1,82) = 2.32, p > .05, respectively]). 9 (11%) used AOD during prior 6 months, 2 (2%) met DSM-IV criteria for current SUD. Current students: 1 reported using AOD, 0 had a current SUD.Current and past endorsed roughly the same number of SUD symptoms. Current used tobacco more often than alumni (76 days and 41 days, respectively, out past 90 days). There were no significant changes across time substance use and recovery variables (no further information provided). Prospective Sample: Sources of stress increased from first (M = 3.9, SD = 2.5) to second assessment (M = 4.6, SD = 2.6). Severe physical health problems not frequently reported in current students (M = 1.0, SD = 1.1) or alumni (M = 1.0, SD = 1.7) at baseline. Physical health declined from first (M = 1.1, SD = 1.0) to second assessment (M = 0.5, SD = 1.0). Symptoms of prior 6 mo. mental and behavioral health problems showed slight decrease from fall to spring: 1) depression-related problems (mean symptom counts of 2.4 and 1.7, respectively), 2) anxiety problems (1.4 and 1.1, respectively), 4) post-traumatic stress problems (1.5 and 1.1, respectively), and 5) attention-deficit problems (3.5 and 2.9, respectively). Total mental health symptom count decreased (paired sample t = 2.49, p < .05); when corrected for multiple-comparison error [p = .05/# of tests (11) = .005], this difference was no longer significant. Overall sample reported 3.5 (of 10 possible) sources of stress in the past 6 mo.: current students (M = 3.9 SD—2.6) reported more sources of stress than alumni (M = 3.4, SD = 2.5). Symptoms of mental health problems were prevalent: anxiety, depression, post-traumatic stress, and attention-deficit problems the most frequently endorsed MH categories. Current students reported higher symptom counts than alumni students across all mental health problem screens [F (1,82) = 12.70, p < .001]. Rates for individual MH screens were higher for current students only on gambling [F (1,82) = 8.00, p < .01], depression [F (1,82) = 6.46, p < .05], and ADHD [F (1,82) = 21.74, p < .001]. Self-help group attendance, past 90 days:Current = 30 daysAlumni = 17 days.Reports of assets or personal strength were high for both past and current students, with a mean number of seven assets (out of 10).
Castedo de Martell et al. (51) Cost effectiveness 92
Cleveland et al. (4) Single-group retrospective; cross-sectional 82 Moderate Fall 2003 to Spring 2006: CRC helped support M = 64 members/semester; 17 relapses. M = 2.8 students relapsing/semester, community has a within-semester relapse rate of 4.4%.
Cleveland & Harris (59) Single-group prospective; longitudinal 60 Moderate Outside-of-center recovery talks on craving: 0.62 (0.25), t = 2.50**At-center recovery talks on craving: 0.012 (0.017), t = 0.72 Outside-of-center recovery talks on negative affect: 0.100 (0.35), t = 2.87**At-center recovery talks on negative affect: 0.035 (0.023), t = 1.55
Wiebe et al. (60) Single-group prospective; longitudinal 55 92 Step work differed across coping styles. Action steps, but not everyday steps, were associated with day-to-day fluctuations in craving when moderated by coping style: among individuals high in Support Seeking, high in Avoidance, or low in Problem Solving. 12-step meeting attendance and received emotional support (daily level), more strongly predicted use of Everyday Steps than Action Steps.Problem solving coping, avoidance coping, and support seeking coping moderated relationship between craving and use of Action Steps. Individuals above the median in Support Seeking and Avoidance were more likely to work the Action Steps on days in when they had craving. Individuals below the median in Problem Solving, were more likely to work the Action steps on days in when they had cravings. These findings were not seen to predict Everyday steps.
Knapp et al. (61) Single-group prospective; longitudinal 50 Odds of family contact were higher on days when they perceived recovery maintenance to be more difficult than usual (OR=1.27, Est. = 0.24, SE = 0.10, CI 95%[0.05, 0.43], p = 0.02. Odds of having contact with sponsors [OR=1.21, CI (−0.02, 0.40)], romantic partners [OR=0.61, CI (−1.15,0.14)], or CRC peers [OR=0.85, CI(−0.42, 0.10)] were not higher on days with higher-than-usual recovery maintenance difficulty. Days when students were in contact with family: length of time spent in contact was higher on days when they perceived recovery maintenance to be more difficult than usual [b = 0.21, SE = 0.05, p = 0.00, 95% CI (0.12, 0.31)]. Recovery maintenance difficulty was not associated with time that students spent with CRC peers inside [b = 0.03, CI (−0.17, 0.23)] nor outside [b = −0.06, CI (−0.15,0.02)] the drop-in center. Days when students were in contact with sponsors and perceived recovery maintenance to be more difficult than usual: odds of having conversations about recovery were higher relative to not having recovery-focused conversations [OR=1.32, Est. = 0.28, SE = 0.14, p = 0.05, 95% CI (0.001, 0.56)]. Day-level association between recovery maintenance difficulty and recovery-focused conversations with CRC peers: on days when students were in contact with CRC peers and perceived recovery maintenance to be more difficult than usual, the odds of having recovery-focused conversations with peers were higher relative to not having recovery-focused conversations [OR=1.27, Est. = 0.24, SE = 0.10, p = 0.02, 95% CI (0.05, 0.44)]. Amount of family contact was higher when students perceived recovery maintenance to be more difficult than usual when including the effects of negative affect and school stress [b = 0.27, CI (0.16, 0.37)]. Recovery maintenance difficulty was not associated with students’ odds of having contact with family [OR=1.22, CI(−0.01,0.41)], romantic partners [OR=0.57, CI(−1.27,0.11)], sponsors [OR=1.15, CI(−0,09,0.37)], or CRC peers [OR=0.90, CI(−0.39,0.19)], nor with the amount of daily contact with CRC peers within [b = 0.02, CI(−0.20,0.24)] and outside the CRC [b = −0.04, CI (−0.13,0.05)], after accounting for negative affect and school stress. 92% of students (n = 46) had a 12-step sponsor at some point during the study. These students contacted their sponsors on 40% of days (n = 384) and had recovery-focused conversations on 35.9% of the days on which they had contact with their sponsors (n = 138). Adjusting for campus drop-in center closures, participants stopped by it on 51% of days. When they stopped by, the most frequently reported time spent was 30–60 min (23.1% of days), with the length of time 30 min more on 53.6% of the days they visited. Students spent time with CRC friends outside the center on 74.1% of days (n = 874) on days with contact, the most frequently reported time spent was 4 + hours (56.1% of days). Students were in contact with CRC peers on 80.5% of days (n = 950). On days when they had contact with CRC peers, students had recovery-focused conversations with them on 46.4% of days (n = 440).
Gerber et al. (52) Cost effectiveness 70
Hamner (69) Single-group retrospective; cross-sectional 89 Moderate Time in recovery M = 4.66 (SD = 4.58)
Herold (68) Qualitative/Mixed methods; cross-sectional 31 Weak Drug Avoidance Self-Efficacy: Total increase in score M = 39.13; 70.2% increase from mean total retrospective score. Quality of Life: Total increase in score M = 56.84; represents 56.4% increase from the mean total retrospective score.
Hennessy et al. (15) Single-group retrospective; longitudinal 861 Varied Weak Time in Recovery: 7.08% < 6 months; 7.34% 6 months-1 year; 40.93% 1–5 years; 25.74% 5–10 years; 18.92% 10 or more years
Nichols et al. (53) Single-group retrospective; cross-sectional 861 History:129/187 students (69.0%): mental health problems98/308 students (31.8%): eating disorder78/169 students (46.2%): other compulsive behavior245/318 students (77.0%) any mental health treatment83/188 students (44.1%) anxiety treatment85/188 students (45.2%) depression treatment
Laudet et al. (11) Single-group retrospective; cross-sectional 486 Moderate 5.4% of students AOD use in past month (3.4% used alcohol; 4.5% used drugs). Days since last drank M = 952 days (SD = 962); Days since last used drugs M = 1,053 days (SD = 1,196). Days since last used any drug or drank alcohol (M = 975, SD = 1,073).Restricting analyses to students considering themselves in recovery from AOD use problem (n = 433): 4.8% used AOD in past month (2.3% alcohol, 3.6% drugs). 1 in 6 students reported also being in recovery from a behavioral addiction.Students had engaged in one of more behavioral addiction in the past 90 days: disordered eating (11.3 days), sex and love (11.7 days), compulsive shopping (8 days), self harm/injury (5.3 days), gambling/gambling addiction (5.1 days), internet addiction (3.1 days), and exercise (2.9 days).Considered self in recovery from:Eating disorder (15.6%)Sex/love addiction (9.5%)Self harm/injury (10.5%)Gaming/gambling addiction (2.5%)Compulsive shopping (3.1%)Internet addiction (0.6%).Exercise addiction (3.3%)Some ‘other’ (2.9%)
Laudet et al. (62) Single-group retrospective; cross-sectional 486 Duration of CRP participation M = 7 semesters (SD =2.0): half of students enrolled in CRP when they started at their academic institution.Perceived helpfulness of CRP participation: 28% selected ‘extremely helpful, 31% ‘quite a bit’, 20% ‘moderately, 14% ‘a little’ and 6% ‘not at all’.
Odefemi-Azzan (65) Single-group retrospective; cross-sectional 222 Weak Relapse rates and mental health (p = 0.004, df = 1)26 students—no MH diagnosis—none relapsed115 students—MH diagnosis—25.2% relapsed 33.7% of female students relapsed0% of male students relapsedFemale students relapse more than male students (p < .001, df = 1).
Patterson et al. (63) Single-group retrospective; cross-sectional 40 Moderate In-degree was the only variable related to length of sobriety in the model (p = 0.002): how many nominations a node (a person) received from other network members was strongly associated with length of sobriety. Sobriety M = 16.89 mo. (SD = 18.40; range: 0 to 75 mo.). Students used CRC M = 8.40 h/wk (SD = 11.23): this was positively related to length of sobriety at the bivariate level.In this model, CRC hours was not related to sobriety length (p = 0.318). Increased CRP participation is correlated with higher sobriety rates (r^2 = 0.49, p < .05)
Smith et al. (55) Single-group prospective; longitudinal 334 59 Weak CRP plays a key role in their social life (M = 68.77, SD = 30.03), 8% reported recurrence of AOD use between baseline and first follow-up; 5.1% reported recurrence of AOD use between the first and second follow-ups.Helps maintain their recovery (M = 79.78, SD = 21.66) Quality of lifeBaseline (M = 31.34, SD = 5.26)Follow-up 1 (M = 30.89, SD = 5.61)Follow-up 2 (M = 30.76, SD = 6.52)Accounting for effects of age, race/ethnicity, gender, cohort, and time in recovery, pairwise comparisons indicated that quality of life at baseline was higher than quality of life at follow-up 1 (Mdiff = 0.97, p = .011, n = 93) and follow-up 2 (Mdiff = 1.38, p = .035, n = 63). Engaged in recovery-related activities, hours per weekBaseline (M = 6.32, SD = 5.83)Follow-up 1 (M = 5.81, SD = 5.63)Follow-up 2 (M = 5.19, SD = 5.25) Accounting for effects of age, race/ethnicity, gender, cohort, and time in recovery, pairwise comparisons indicated that hours engaged in recovery-related activities at baseline was not different than at the first (p < .704, n = 100) or second follow-up (p = .054, n = 73).Helps them grow personally (M = 82.70, SD = 21.54).
Smith et al. (54) Single-group prospective; longitudinal 185 Since onset of COVID-19 pandemic, 15.7% experienced recurrence of AOD use; 51.9% attending fewer recovery meetings.Students who were abstinent from their primary substance had higher recovery capital [Pillai's Trace = 0.37, F(10, 159) = 99.23, p < 0.001, η2 = 0.367] than those who were not abstinent Students abstinent from primary substance had better quality of life [Pillai's Trace = 0.10, F(8, 166) = 2.23, p = 0.028, η2 = 0.097] than those who were not abstinent.Overall recovery capital scores were not significantly correlated with COVID-19-related changes in stress levels (r = -0.03, p = 0.722). 66.0% endorsed CRPs as a primary pathway to recovery.
Vest et al. (56) Single-group prospective; cross-sectional 435 SUD severity:Formerly incarcerated students (M = 10.55, SD = 1.01)Non-incarcerated LS-involved students (M = 10.27, SD = 1.31)Non-LS involved students (M = 9.73, SD = 1.38)Significant difference in SUD severity between non-LS involved students and between non-incarcerated LS and between non-LS involved students and formerly incarcerated students (F = 9.01, p < 0.001).AUD severity:Formerly incarcerated students (M = 9.93, SD = 1.70)Non-incarcerated LS-involved students (M = 9.82, SD = 1.84)Non-LS involved students (M = 9.51, SD = 1.77)No significant differences in AUD severity between groups (F = 0.48, p = 0.617). Quality of life:Non-incarcerated LS involved students (M = 32.42, SD = 4.69)Formerly incarcerated students (M = 31.73, SD = 5.44) Non-LS involved students (M = 30.25, SD = 5.65).No significant differences in quality of life between groups (F = 1.74, p = 0.177).
Smith (57) Single-group retrospective; cross-sectional 16 Weak Number of times of relapse M = 6.07 (SD =12.72) prior to entering the CYAAR program. No participants relapsed after entering the program. There is a significant difference in mean ranks of the number of relapses before and after entering the program.
Tuliao et al. (67) Single-group retrospective; cross-sectional 162 Weak 83% (N = 134) participated in mutual help/12-step program.Negative correlation between mutual help participation and intent to use addiction medicine (r^2= −0.02, p > .05).
Wattick et al. (64) Single-group retrospective; cross-sectional 13 Weak Mean craving score (out of 7) was 1.1. Resilience score (out of 5) M = 3.4 (SD = 0.91). Higher nutrition and culinary program participation was associated with ratings of good health and the absence of depression symptoms (P = .03 and .04, respectively).

Bold indicates a “parent” study of a series of linked studies.

**indicates p < .05.

In one cost-benefit analysis of CRPs (51, 74), the authors used secondary data from two national surveys of CRPs at the programmatic level to model CRP-related variables. The mean budget was modeled from 54 CRPs and the mean membership size from 20 CRPs, while the average relapse rates were used from a 2012 survey of 29 CRPs across 19 US states. Model results indicated that the cost of operating a CRP in the base case is $97,586.24 less than the cost of treatment as usual and adds just over 25 Quality-Adjusted Life Years. The incremental cost-effectiveness ratio for the societal model in the base case is -$3,872.75, suggesting a cost savings of $3,872.75 per Quality-Adjusted Life Years gained when implementing a CRP.

In another cost-benefit analysis of CRPs (52), several sources of existing data were used. Utilizing Castedo De Martell's 2019 findings of the average CRP budget [from the thesis underlying the 2022 publication covered above (74)], the CRP costs for this study were set at $191,389. The mean room and board cost to educate students of $9,488 was used as a cost to the university. The national average alumni giving amounts in the United States was determined using alumni giving amounts and volunteering to the college. An underlying assumption was that CRP students might be more willing to give due to the strong tradition of giving back as part of CRP membership while students at the university. A CRP size of 14 members was used in the calculations and a 90% graduation rate was used. These calculations resulted in a societal cost-benefit ratio of 1.94 and an institutional cost-benefit ratio of 1.38. The societal Incident Rate Ratio was 249% and the institutional Incident Rate Ratio was 77%. The results indicate that for every dollar the university spends, there will be a return of $2.26 over the course of ten years. Thus, this study indicated that the benefits of a CRP outweighed the costs, but to what degree a CRP outweighs the costs will depend on the size of the university and the expected CRP student membership.

 Discussion

This systematic review identified 25 unique studies focused on education-based supports for adolescents (RHSs) and emerging adults (CRPs). The combination of a good number of descriptive studies, and a few studies with comparison groups, suggests that students who participate in RHS and CRPs may demonstrate positive outcomes such as reductions in AOD use, and improvements in social and academic outcomes (e.g., grades). Yet, given the lack of randomized controlled designs or even non-randomized comparative prospective research designs, statements about the incremental public health utility of investing in RHSs and/or CRPs relative to some other kind of approach of equal intensity and duration or engagement with services-as-usual, cannot be made with confidence. Similarly, cost-benefit analyses among both types of programs suggest large cost-savings or benefits, which could benefit society overall, but given the lack of comparative prospective investigations, robust estimates of cost-benefit remain unclear. Unlike an RHS, which often requires standalone programming, CRPs represent programming that can be implemented on college and university campuses and potentially integrated with college student health and wellness services. Thus, there seems to be increased potential for their sustainability in these settings given the larger cost savings and potential for synergy with existing structures that these programs represent. Overall, findings suggest possible near-term benefits are likely from RHS and CRP participation. However, much greater research investment is needed to understand more clearly the incremental benefit attributable to these kinds of programs as well for which students in particular, and also to gain greater clarity on the long-term trajectories of youth with SUD histories engaged with such services vs. similar youth who are not engaged with such services.

As might be expected of studies examining recovery supports in education settings, the most commonly measured outcomes for both types of supports were some assessment of substance use and academic outcomes. Surprisingly, given that a key proposed mechanism of both programs is engaging young people in positive social supports and creating recovery-supportive social networks, only a minority of studies examined whether this happened for their participants (5, 13). Capturing social networks can be challenging with limited resources, but it would be an important mechanism for these programs to examine in future research. A few studies also examined mental health outcomes, which epidemiological data has identified as important to understand among individuals experiencing AOD use disorders (75, 76). Only the CRP studies directly examined recovery capital, using a recovery capital measure. This may be due in part to the lack of a developmentally-appropriate measure of recovery capital despite the identified gap in recovery capital measures (23, 77).

At present, given the study designs of the existing research identified in this review, there is limited evidence identifying RHSs and CRPs as the direct cause of improved outcomes for their participants. There is also limited evidence of longer-term outcomes related to positive development and the growth of recovery capital, such as college graduation, employment and career trajectory after college, and marriage or family life. Furthermore, results from the reviewed studies must be considered alongside their methodological limitations which included high potential for selection bias (often inadequately addressed in analysis), study design (majority of studies did not utilize a comparator group), and the lack of appropriate attention to confounders. There are many important gaps in the literature base that will be important to fill. For example, as it is less feasible to conduct RCT designs with these types of recovery supports, it will be important for researchers to conduct rigorous and longitudinal quasi-experimental studies to determine the effect of RHS and CRPs on AOD use and related outcomes, from which point researchers can work to determine which aspects of these programs are most beneficial, for whom in particular, and why. Providing RHS and CRP programs with funds for a data collection infrastructure that includes dedicated personnel and technical support to engage in ongoing program monitoring will help to establish a sustainable evidence base for these services (12, 13). Overall, substantially more research is needed to begin forming conclusions about the utility of these education-based recovery supports.

In addition to efficacy, RHS and CRPs face additional challenges that warrant investigation. One key issue evident from the studies in this review is that students who utilize these supports are predominately white. Although researchers identify racial disparities in addiction treatment as an ongoing issue (9), RHS often do not reflect the demographic breakdown of their school district (46) or their county. In fact, there are more students of color who receive addiction treatment per capita than attend RHS (78). One possible avenue for future research is to examine, among other factors, the ways in which students are referred and considered for program admission to identify barriers that minoritized individuals face in accessing these services, reason for engaging in services when they do, and strategies for surmounting them. Moreover, it is important for researchers to investigate why, despite the millions of adolescents and young adults with substance use disorder who need treatment, many RHS report one of their main challenges to be enrolling enough students (10). When considering that in 2021, there were an estimated 1.9 million adolescents and 8 million young adults in need of specialized substance use disorder treatment who did not receive it (79), it is important for researchers to work to reconcile the paradox of the adolescent and young adult treatment gap with the enrollment struggles of RHS.

Despite a strong theoretical rationale for the need for education-based recovery supports for young people, research on their effectiveness remains nascent, with only a handful of studies examining these potentially integral supports. Well-conducted comparison studies that examine their underlying logic model are needed to confidently assert that such programs are worthy of integration into the broader public education system (13, 14). Then, larger-scale studies might be undertaken that examine the specific mechanisms to determine how these resources confer benefit and for which students, in particular.

Limitations

Several limitations to the methodological approaches employed in this review should be mentioned. First, screening from the original search was not conducted in duplicate; however, the updated search included duplicate screening, and there were no exclusion discrepancies during this screening, so the risk of inadvertently excluding eligible studies is minimal. Second, while we identified some grey literature (e.g., conference abstracts and dissertations), our search primarily used electronic databases. Thus, some relevant studies may be missing from this review. Finally, although the majority of studies' risk of bias were assessed in duplicate using two reviewers, only a single reviewer applied the separate cost-benefit tool to assess the risk of bias in the two cost-benefit studies. In the interest of transparency, we provide all coded responses for this tool in Table 4.

 Conclusions

This systematic review highlights the growing interest and evidence supporting education-based recovery supports (RHSs) and collegiate recovery programs (CRPs) for adolescents and young adults, suggesting potential reductions in substance use and improvements in academic and social outcomes. Despite these promising findings, robust causal evidence is limited, particularly regarding effects on long-term outcomes and the development of recovery capital, indicating a need for more rigorous comparative studies with much longer follow-up and better measurement and data collection. Additionally, disparities in enrollment highlight a critical gap in access for students of minoritized identities, emphasizing the importance of investigating barriers to participation and ensuring these programs reflect the demographics of their communities (70, 71).

 Data availability statement

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

 Author contributions

EH: Conceptualization, Data curation, Formal Analysis, Investigation, Methodology, Project administration, Supervision, Visualization, Writing – original draft, Writing – review & editing, Resources, Validation. SG: Formal Analysis, Visualization, Writing – review & editing, Data curation, Investigation, Validation. MK: Formal Analysis, Investigation, Validation, Writing – review & editing. JO: Formal Analysis, Investigation, Validation, Writing – review & editing. DE: Writing – review & editing. JK: Funding acquisition, Resources, Supervision, Writing – review & editing, Conceptualization.

 Funding

The author(s) declare financial support was received for the research, authorship, and/or publication of this article. Support for this work was provided in part by the Substance Abuse and Mental Health Services Administration and Opioid Response Network (1H79TI085588) and the National Institute on Alcohol Abuse and Alcoholism (K01AA028536 [Hennessy], K23AA027577 [Eddie], K24AA022136-10 [Kelly]).

 Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author(s) declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.

 Generative AI statement

The author(s) declare that no Generative AI was used in the creation of this manuscript.

 Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

 Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fradm.2025.1522678/full#supplementary-material

1Other publications that were identified and linked to this study because they published on a portion of the data or focused on aspects of the study design but were not coded for this review include: (70, 71).

 References Behavioral Health, Statistics and Quality. Substance Abuse and Mental Health Services Administration, Center for Treatment Episode Data Set (TEDS): 2020. Admissions to and Discharges from Publicly Funded Substance Use Treatment Facilities. Rockville, MD: Substance Abuse and Mental Health Services Administration (2022). Available online at: https://www.samhsa.gov/data/data-we-collect/teds-treatment-episode-data-set Substance Abuse and Mental Health Services Administration (SAMHSA). 2018–2019 National Surveys on Drug Use and Health: Model-Based Estimated Totals (in Thousands) (50 States and the District of Columbia). Available online at: https://www.samhsa.gov/data/sites/default/files/reports/rpt32879/NSDUHsaeTotal2019/2019NSDUHsaeTotal.pdf (Accessed September 10, 2024). National Center for Education Statistics. Marijuana Use and Illegal Drug Availability [Internet]. 2021 May. (Preprimary, Elementary, and Secondary Education). Available online at: https://nces.ed.gov/programs/coe/indicator/a15/marijuana-use-drug-availability (Accessed September 10, 2024). ClevelandHHHarrisKSBakerAKHerbertRDeanLR. Characteristics of a collegiate recovery community: maintaining recovery in an abstinence-hostile environment. J Subst Abuse Treat. (2007) 33(1):13–23. 10.1016/j.jsat.2006.11.00517588485 FinchAJFriedenG. The ecological and developmental role of recovery high schools. Peabody J Educ. (2014) 89(2):271–87. 10.1080/0161956X.2014.897106 HennessyEATanner-SmithEEFinchAJSatheNKugleyS. Recovery schools for improving behavioral and academic outcomes among students in recovery from substance use disorders. Campbell Collab. (2018) 14:1–86. 10.4073/csr.2018.9 WhiteWLFinchAJ. The recovery school movement: its history and future. Counselor. (2006) 7(2):54–8. Office of National Drug Control Policy. National Drug Control Strategy 2014. Washington, DC: Office of National Drug Control Policy (2014). Available online at: https://www.ojp.gov/ncjrs/virtual-library/abstracts/national-drug-control-strategy-2014 (cited October 21, 2024). Association of Recovery Schools. The State of Recovery High Schools, 2016 Biennial Report. Denton, TX; 2017. Available online at: https://recoveryschools.org/wp-content/uploads/2016/03/State-of-Recovery-Schools_3-3-16.pdf (cited 2024 October 21). FinchAJMobergDPKruppAL. Continuing care in high schools: a descriptive study of recovery high school programs. J Child Adolesc Subst Abuse. (2014) 23(2):116–29. 10.1080/1067828X.2012.751269 LaudetABHarrisKKimballTWintersKCMobergDP. Characteristics of students participating in collegiate recovery programs: a national survey. J Subst Abuse Treat. (2015) 51:38–46. 10.1016/j.jsat.2014.11.00425481690 VestNde MartellSCHennessyE. Introducing a socio-ecological model of collegiate recovery programs. Drug Alcohol Depend. (2024) 260:110563. 10.1016/j.drugalcdep.2023.110563 HennessyEANicholsLMBrownTBTanner-SmithEE. Advancing the science of evaluating collegiate recovery program processes and outcomes: a recovery capital perspective. Eval Program Plann. (2022) 91:102057. 10.1016/j.evalprogplan.2022.10205735217288 VestNReinstraMTimkoCKellyJHumphreysK. College programming for students in addiction recovery: a PRISMA-guided scoping review. Addict Behav. (2021) 121:106992–106992. 10.1016/j.addbeh.2021.10699234087765 HennessyEATanner-SmithENicholsLMBrownTBMccullochBJ. A multi-site study of emerging adults in collegiate recovery programs at public institutions. Soc Sci Med. (2021) 278:113955–113955. 10.1016/j.socscimed.2021.11395533940434 GranfieldRCloudW. Coming Clean: Overcoming Addiction Without Treatment. Vol. 25. New York: New York University Press (1999). p. 281. GranfieldRCloudW. Social context and “natural recovery”: the role of social capital in the resolution of drug-associated problems. Subst Use Misuse. (2001) 36(11):1543–70. 10.1081/JA-10010696311693955 HennessyEACristelloJVKellyJF. RCAM: a proposed model of recovery capital for adolescents. Addict Res Theory. (2019) 27(5):429–36. 10.1080/16066359.2018.1540694 MaslowAH. A theory of human motivation. Psychol Rev. (1943) 50:370–96. 10.1037/h0054346 MaslowAH. Preface to motivation theory. Psychosom Med. (1943) 5(1):85–92. 10.1097/00006842-194301000-00012 FinchAJ. Recovery High Schools Growth Chart - Assn. of Recovery Schools [Internet]. Association of Recovery Schools. (2021). Available online at: https://recoveryschools.org/rhs-growth-chart/ (cited October 21, 2024). Association of Recovery in Higher Education. Collegiate Recovery Directory [Internet]. Kennesaw, Georgia: Association of Recovery in Higher Education. (2024). Available online at: https://collegiaterecovery.org/collegiate-recovery-directory/ (cited October 21, 2024). HennessyEA. Recovery capital: a systematic review of the literature. Addict Res Theory. (2017) 25(5):349–60. 10.1080/16066359.2017.1297990 PageMJMcKenzieJEBossuytPMBoutronIHoffmannTCMulrowCD The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Br Med J. (2021) 372:n71. 10.1136/bmj.n71 The Methods Group of the Campbell Collaboration. Methodological expectations of campbell collaboration intervention reviews: conduct standards. Campbell Policies Guidel Ser. (2016) 3. 10.4073/cpg.2016.3 AloeAMDewidarOHennessyEAPigottTStewartGWelchV Campbell standards: modernizing campbell’s methodologic expectations for campbell collaboration intervention reviews (MECCIR). Campbell Syst Rev. (2024) 20(4):e1445. 10.1002/cl2.144539376895 JohnsonBTHennessyEA. Systematic reviews and meta-analyses in the health sciences: best practice methods for research syntheses. Soc Sci Med. (2019) 233. (Journal Article):237–51. 10.1016/j.socscimed.2019.05.03531233957 HaddawayNRPageMJPritchardCCMcGuinnessLA. PRISMA2020: an R package and Shiny app for producing PRISMA 2020-compliant flow diagrams, with interactivity for optimised digital transparency and Open Synthesis. Campbell Syst Rev. (2022) 18:e1230. 10.1002/cl2.123036911350 HarrisPATaylorRThielkeRPayneJGonzalezNCondeJG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. (2009) 42(2):377–81. 10.1016/j.jbi.2008.08.01018929686 MamikuttyRAlyASMarhazlindaJ. Selecting risk of bias tools for observational studies for a systematic review of anthropometric measurements and dental caries among children. Int J Environ Res Public Health. (2021) 18(16):8623. 10.3390/ijerph1816862334444374 ThomasBHCiliskaDDobbinsMMicucciS. A process for systematically reviewing the literature: providing the research evidence for public health nursing interventions. Worldviews Evid Based Nurs. (2004) 1(3):176–84. 10.1111/j.1524-475X.2004.04006.x17163895 Armijo-OlivoSStilesCRHagenNABiondoPDCummingsGG. Assessment of study quality for systematic reviews: a comparison of the cochrane collaboration risk of bias tool and the effective public health practice project quality assessment tool: methodological research. J Eval Clin Pract. (2012) 18(1):12–8. 10.1111/j.1365-2753.2010.01516.x20698919 McGuinnessLAHigginsJPT. Risk-of-bias VISualization (robvis): an R package and shiny web app for visualizing risk-of-bias assessments. Res Synth Methods. (2021) 12(1):55–61. 10.1002/jrsm.141132336025 ChiouCFHayJWWallaceJFBloomBSNeumannPJSullivanSD Development and validation of a grading system for the quality of cost-effectiveness studies. Med Care. (2003) 41(1):32–44. 10.1097/00005650-200301000-0000712544542 CampbellMMcKenzieJESowdenAKatikireddiSVBrennanSEEllisS Synthesis without meta-analysis (SWiM) in systematic reviews: reporting guideline. Br Med J. (2020) 368:l6890. 10.1136/bmj.l6890 KellyJFMyersMGBrownSA. A multivariate process model of adolescent 12-step attendance and substance use outcome following inpatient treatment. Psychol Addict Behav. (2000) 14(4):376–89. 10.1037/0893-164X.14.4.37611130156 HenselDJWilburnVG. Daily association of drug use cravings and physical and emotional well-being among students attending a recovery high school during COVID-19: results from an EMA study. J Adolesc Health. (2023) 72(3):S88. 10.1016/j.jadohealth.2022.11.17836229391 GilesM. The Massachusetts recovery high schools: ensuring success in school and in recovery (Phd thesis). Brandeis University, The Heller School for Social Policy and Management (2022). Available online at: https://search.proquest.com/openview/963d364fe3a5cacf2dd2eb11a5970579/1?pq-origsite=gscholar&cbl=18750&diss=y (cited October 21, 2024). KnottsAC. Young people in recovery from substance use disorders: an analysis of a recovery high school’s impact on student academic performance & recovery success (Phd thesis). Indiana University, Indiana (2017). Available online at: https://scholarworks.iupui.edu/bitstream/1805/16280/1/Knotts_iupui_0104D_10282.pdf (cited October 21, 2024). FinchAJTanner-SmithEHennessyEMobergDP. Recovery high schools: effect of schools supporting recovery from substance use disorders. Am J Drug Alcohol Abuse. (2017) 44(2):175–84. 10.1080/00952990.2017.135437828767275 Tanner-SmithEEFinchAJHennessyEAMobergDP. Effects of recovery high school attendance on Students’ mental health symptoms. Int J Ment Health Addict. (2019) 17(2):181–90. 10.1007/s11469-017-9863-732831810 Tanner-SmithEENicholsLMLoanCMFinchAJMobergDP. Recovery high school attendance effects on student delinquency and substance use: the moderating role of social problem solving styles. Prev Sci. (2020) 21(8):1104–13. 10.1007/s11121-020-01161-z32886317 GlaudeMWJenningsSWTorresLRFinchAJ. School of enrollment and perceptions of life satisfaction among adolescents experiencing substance use disorders. J Evid-Based Soc Work. (2019) 16(3):261–77. 10.1080/26408066.2019.1588184 Tanner-SmithEEFinchAJHennessyEAMobergDP. Who attends recovery high schools after substance use treatment? A descriptive analysis of school aged youth. J Subst Abuse Treat. (2018) 89:20–7. 10.1016/j.jsat.2018.03.00329706171 WeimerDLMobergDPFrenchFTanner-SmithEEFinchAJ. Net benefits of recovery high schools: higher cost but increased sobriety and increased probability of high school graduation. J Ment Health Policy Econ. (2019) 22(3):109–20.31811754 LanhamCCTiradoJA. Lessons in sobriety: an exploratory study of graduate outcomes at a recovery high school. J Groups Addict Recovery. (2011) 6(3):245–63. 10.1080/1556035X.2011.597197 MobergDPFinchAJ. Recovery high schools: a descriptive study of school programs and students. J Groups Addict Recovery. (2008) 2. (Journal Article):128–61. 10.1080/15560350802081314 HennessyEAJurinskyJCowieKPietrzakAZBlythSKrasnoffP Visualizing the influence of social networks on recovery: a mixed-methods social identity mapping study with recovering adolescents. Subst Use Misuse. (2024) 59(9):1405–15. 10.1080/10826084.2024.235261838738809 AshfordRDBrownAMCurtisB. Collegiate recovery programs: the integrated behavioral health model. Alcohol Treat Q. (2018) 36(2):274. 10.1080/07347324.2017.1415176 BrownAMAshfordRDFigleyNCoursonKCurtisBKimballT. Alumni characteristics of collegiate recovery programs: a national survey. Alcohol Treat Q. (2019) 37(2):149–62. 10.1080/07347324.2018.1437374 Castedo de MartellSHolleran SteikerLSpringerAJonesJEisenhartEBrownHSIII. The cost-effectiveness of collegiate recovery programs. J Am Coll Health. (2022) 72(1):82–93. 10.1080/07448481.2021.202420635080467 GerberWMHuneNDWangEWKimballTG. Tangible and intangible values of a CRP: a cost-benefit analysis. Alcohol Treat Q. (2022) 40(2):164–73. 10.1080/07347324.2021.1979437 NicholsLMHennessyEABrownTBTanner-SmithEE. Co-occurring mental and behavioral health conditions among collegiate recovery program members. J Am Coll Health. (2021) 71(7):2085–92. 10.1080/07448481.2021.195568734314662 SmithRLBannardTMcDanielJWhitneyJDeFrantz-DuforWStatmanM Examining the associations between collegiate recovery programs and students' recovery capital and quality of life during the COVID-19 pandemic. Poster Abstracts. Alcohol Clin Exp Res. (2021) 45:80–262. 10.1111/acer.14628 SmithRLBannardTMcDanielJAlievFBrownAHollidayE Characteristics of students participating in collegiate recovery programs and the impact of COVID-19: an updated national longitudinal study. Addict Res Theory. (2024) 32(1):58–67. 10.1080/16066359.2023.221645938524726 VestNBellJSNiederASmithRBannardTTragesserS Learning with conviction: exploring the relationship between criminal legal system involvement and substance use and recovery outcomes for students in collegiate recovery programs. Drug Alcohol Depend. (2024) 257:111127. 10.1016/j.drugalcdep.2024.11112738394814 SmithT. Sustaining recovery: the participant perspective of a collegiate recovery program. Diss Abstr Int. (2022) 83(10-B):1–118. BotzetAMWintersKFahnhorstT. An exploratory assessment of a college substance abuse recovery program: Augsburg College’s StepUP program. J Groups Addict Recovery. (2008) 2(2–4):257–70. 10.1080/15560350802081173 ClevelandHHHarrisK. Conversations about recovery at and away from a drop-in center among members of a collegiate recovery community. Alcohol Treat Q. (2010) 28(1):78–94. 10.1080/07347320903436268 WiebeRPGriffinAMZhengYHarrisKSClevelandHH. Twelve steps, two factors: coping strategies moderate the association between craving and daily 12-step use in a college recovery community. Subst Use Misuse. (2018) 53(1):114–27. 10.1080/10826084.2017.132590428813187 KnappKSClevelandHHApsleyHBHarrisKS. Using daily diary methods to understand how college students in recovery use social support. J Subst Abuse Treat. 20210414 ed. (2021) 130:108406. 10.1016/j.jsat.2021.10840634118698 LaudetABHarrisKKimballTWintersKCMobergDP. In college and in recovery: reasons for joining a collegiate recovery program. J Am Coll Health. (2016) 64(3):238–46. 10.1080/07448481.2015.111746426731130 PattersonMSRussellAMNelonJLBarryAELanningBA. Using social network analysis to understand sobriety among a campus recovery community. J Stud Aff Res Pract. (2020) 58(4):401–16. 10.1080/19496591.2020.1713142 WattickRAHagedornRLOlfertMD. Enhancing college student recovery outcomes through nutrition and culinary therapy: mountaineers for recovery and resilience. J Nutr Educ Behav. (2020) 52(3):326–9. 10.1016/j.jneb.2019.11.00631866257 Odefemi-AzzanOA. Factors that affect students enrolled in a midsize collegiate recovery program in the United States (ProQuest Dissertations & Theses). St. Peter's University, Jersey City, NJ, United States (2020). BennettMEMcCradyBSKellerDSPaulusMD. An intensive program for collegiate substance abusers progress six months after treatment entry. J Subst Abuse Treat. (1996) 13(3):219–25. 10.1016/S0740-5472(96)00045-19017564 TuliaoAPTuliaoMDLewisLEMulletNDNorth-OlagueJMillsD Perceptions of addiction medications within collegiate recovery communities. J Addict Med. (2023) 17(6):732–5. 10.1097/ADM.000000000000122437934547 HeroldMA. Student perceptions of the effectiveness of collegiate recovery programs and collegiate recovery communities (ProQuest Dissertations & Theses). The University of Texas Rio Grande Valley, Edinburg, TX, United States (2023). HamnerLC. The Influence of Authentic Leadership on Recovery Capital in a Collegiate Setting. San Antonio, TX: Our Lady of the Lake University (2023). Available online at: https://search.proquest.com/openview/80d034820124e56e7670bc10febe777a/1?pq-origsite=gscholar&cbl=18750&diss=y (cited October 21 2024). BotzetAMMcilvainePWWintersKCFahnhorstTDittelC. Data collection strategies and measurement tools for assessing academic and therapeutic outcomes in recovery schools. Peabody J Educ. (2014) 89(2):197–213. 10.1080/0161956X.2014.89564825018573 HennessyEAFinchAJ. Adolescent recovery capital and recovery high school attendance: an exploratory data mining approach. Psychol Addict Behav. (2019) 33(8):669. 10.1037/adb000052831724415 GroshkovaTBestDWhiteW. The assessment of recovery capital: properties and psychometrics of a measure of addiction recovery strengths. Drug Alcohol Rev. (2013) 32(2):187–94. 10.1111/j.1465-3362.2012.00489.x22882622 VilsaintCLKellyJFBergmanBGGroshkovaTBestDWhiteW. Development and validation of a brief assessment of recovery capital (BARC-10) for alcohol and drug use disorder. Drug Alcohol Depend. (2017) 177. (Journal Article):71–6. 10.1016/j.drugalcdep.2017.03.02228578224 Castedo de MartellS. Cost-Effectiveness Of Collegiate Recovery Programs [Internet]. UTHealth School of Public Health; 2019. Available online at: https://digitalcommons.library.tmc.edu/uthsph_dissertsopen/42/ (cited 2024 October 21). CouwenberghCvan den BrinkWZwartKVreugdenhilCvan Wijngaarden-CremersPvan der GaagRJ. Comorbid psychopathology in adolescents and young adults treated for substance use disorders. Eur Child Adolesc Psychiatry. (2006) 15(6):319–28. 10.1007/s00787-006-0535-616648966 LampsCASoodABSoodR. Youth with substance abuse and comorbid mental health disorders. Curr Psychiatry Rep. (2008) 10(3):265–71. 10.1007/s11920-008-0043-018652796 BunaciuABliucAMBestDHennessyEABelangerMBenwellC. Measuring recovery capital for people recovering from alcohol and drug addiction: a systematic review. Addict Res Theory. (2024) 32(3):225–36. 10.1080/16066359.2023.224532339045096 OserRKarakosHLHennessyEA. Disparities in youth access to substance abuse treatment and recovery services: how one recovery school initiative is helping students “change tracks.”. J Groups Addict Recovery. (2016) 11(4):267–81. 10.1080/1556035X.2016.1211056 Substance Abuse and Mental Health Services Administration (SAMHSA). 2021 National Surveys on Drug Use and Health: Model-Based Estimated Totals (in Thousands) (50 States and the District of Columbia) - Table 28 [Internet]. Available online at: https://www.samhsa.gov/data/report/2021-nsduh-estimated-totals-state (Accessed September 10, 2024).