Economic benefits of substance use disorder treatment: A systematic literature review of economic evaluation studies from 2003 to 2021

Abstract

Introduction:

The economic burden of substance use disorder (SUD) is significant, comprising costs of health care and social services, criminal justice resources, loss of productivity, and premature mortality. This study assembles and synthesizes two decades of evidence describing the benefits of SUD treatment across five main outcome domains; 1) health care utilization; 2) self-reported criminal activity by offense type; 3) criminal justice involvement collected from administrative records or self-reported; 4) productivity assessed through working hours or wages earned; and 5) social services (e.g., a day spent in transitional housing).

Methods:

This review included studies if they reported the monetary value of the intervention outcomes, most commonly through a cost-benefit or cost-effectiveness framework. The search included studies from 2003 to the present day as of this writing (up to October 15, 2021). Summary cost estimates were adjusted using the US Consumer Price Index (CPI) to reflect the 12-month benefits per client in USD 2021. We followed the PRISMA methodology for study selection and assessed quality using the Checklist for Health Economic Evaluation Reporting Standards (CHEERS).

Results:

The databases yielded 729 studies after removing duplicates, and we ultimately selected 12 for review. Studies varied widely regarding analytical approaches, time horizons, outcome domains, and other methodological factors. Among the ten studies that found positive economic benefits, reductions in criminal activity or criminal justice costs represented the largest or second largest component of these benefits (range $621 to $193,440 per client).

Conclusions:

Consistent with previous findings, a reduction in criminal activity costs is driven by the relatively high societal cost per criminal offense, notably for violent crimes, such as aggravated assault and rape/sexual assault. Accepting the economic rationale for increased investment in SUD interventions will require recognizing that more benefits accrue to individuals by avoiding being victims of a crime than to governments through budget offsets resulting from savings in non-SUD program expenses. Future studies should explore individually tailored interventions to optimize care management, which may yield unexpected economic benefits to services utilization, and criminal activity data to estimate economic benefits across a broad range of interventions.

1. Introduction

The 2020 National Survey on Drug Use and Health (NSDUH) reports that more than 20% of people over 12 used illicit substances in the past year, and more than 40 million have a substance use disorder (SUD). The economic burden of SUD is significant, comprising costs of health care and social services, criminal justice resources, lost earnings, and premature mortality (Degenhardt & Hall, 2012; Rehm et al., 2017; WHO, 2021). The National Institute on Drug Abuse (NIDA) reports that the aggregate cost of SUD in the United States is more than $700 billion per year (NIDA, 2020). Interventions to reduce or mitigate the negative consequences of SUD could result in significant reductions in these societal costs. As preventive and treatment interventions for SUD are developed and tested, economic impact analysis remains essential for informing stakeholders about the feasibility and sustainability of implementing evidence-based interventions. Moreover, economic evaluation studies of an SUD intervention’s costs and monetary benefits provide evidence for determining which programs represent the best value while considering how the interventions can vary substantially across health care systems, populations, and specific client needs.

Studies of the economic benefits of SUD interventions have increased in the past two decades, reflecting the desire to demonstrate to policymakers that evidence-based interventions are effective, cost-effective, and/or cost-saving. We conduct a systematic literature review of economic evaluations to take stock of what the field has learned about the economic impact of SUD interventions over the past two decades and to assess how the field should move forward to address measurement challenges and other limitations notable across these studies.

This article updates a study by McCollister and French (2003) (McCollister & French, 2003), which reviewed cost-benefit studies of SUD treatment from 1988 to 2002, highlighting the relative contribution of economic outcome domains to the total economic benefits of treatment. That study found that avoiding criminal activity generated the greatest economic benefit of SUD interventions and contributed more, as a separate outcome domain, to the total economic benefit of SUD interventions than any other domain. Reduced utilization of health care services was also a noteworthy economic benefit of SUD interventions. It seems prudent to ask if more recent evidence supports these findings or if other areas of impact are emerging as important sources of economic benefits.

Treating SUD can improve clinical and societal outcomes across many domains. From an economic perspective, important domains beyond health care services and criminal-legal sectors include productivity, social services, infectious disease incidence, infant health outcomes (e.g., fetal alcohol syndrome, neonatal abstinence syndrome), and quality of life. Economic benefits can capture reductions in service utilization, disease incidence or adverse health outcomes, and productivity and quality of life improvements. In summarizing results from this review, we also identify gaps in measuring economic benefits and other methodologic issues to include as priority areas on the research agenda in this field going forward.

2. Methods

2.1. Search strategy

We follow the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) methodology (Moher et al., 2009; Page et al., 2021). We registered the protocol with PROSPERO before the title and abstract reviews began (CRD42021246424). We identified studies by searching the following electronic databases: PubMed, Cochrane, Embase, EconLit, PsycINFO, and Web of Science. We first searched in PubMed for journal articles published in English from January 1, 2003, to October 15, 2021 (See Appendix A for a detailed PubMed search strategy).

2.2. Selection process

EF initially screened articles based on the title and abstract, removing ineligible articles. Next, two authors (EF and IDM) conducted full text reviews of the articles. BRS and KEM reviewed selected articles and resolved any discrepancies regarding the final list of articles to be featured in this study. The PRISMA flow diagram in Figure 1 shows the electronic search process. The first search identified 745 articles (670 from PubMed, 30 from Web of Science, 20 from Embase, 19 from Cochrane Library, and six from PsycINFO). Five additional articles are included based on a review of references in the studies selected for a full-text review. We saved the articles in EndNote X9^® (Clarivate Analytics) and exported the search into a systematic review web-based software tool, Covidence^®, automatically removing duplicates. In total, we screened 729 articles, ultimately excluding 614.

Figure 1.

Flow diagram of the study.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram describing article selection. Primary reasons for exclusion are presented in terms of which exclusion criteria became evident during screening.

2.3. Inclusion and exclusion criteria

To be included in the systematic review, studies had to be in English; evaluate the cost-benefit or cost-effectiveness of an intervention targeting SUD; and include data on both intervention costs and the economic consequences in monetary terms (expressed as benefits, reduced societal costs, or savings) of interventions targeting the use and misuse of alcohol, cannabis, opiates, or other illicit substances. Economic consequences could include any combination of outcomes across health care, criminal-legal, social services, infectious diseases, productivity, and quality of life domains. Studies calculated economic consequences or benefits by applying a unit cost to counts of services (e.g., nights staying in a hospital) or events (e.g., days incarcerated in prison or jail). Productivity outcomes are already expressed in dollars as earnings. For nonmonetary outcomes, studies calculated economic benefits by applying unit costs from various sources, including reimbursement rates, administrative data, and published literature.

The exclusion criteria were: (1) no estimation of economic benefits/societal costs; (2) studies published in a language other than English; (3) the articles were reviews, protocol papers, letters, editorials, conference abstracts, poster presentations with insufficient details, or case reports; (4) and published before 2002.

2.4. Data extraction and analysis plan

We developed data extraction templates for each study selected for full-text review. The summary table of economic benefits includes currency and reference year for each study and five main outcome domains: 1) health care utilization; 2) self-reported criminal activity by offense type; 3) criminal justice involvement collected from administrative records or self-reported (i.e., specific contacts with the justice system such as an arrest or a day incarcerated); 4) productivity assessed through working hours or wages earned; and 5) social services (e.g., a day spent in transitional housing). A sixth category, “other consequences,” was included in two studies—one estimated societal cost attributable to automobile accidents (Logan TK, 2004), and another one estimated the costs of drugs purchased (Elarabi et al., 2021).

For comparability across different assessment timeframes, economic benefits were annualized using a straight-line method to represent 12-month benefits and then adjusted to 2021 US dollars using the GDP implicit price deflator (U.S. Bureau of Economic Analysis, 2022). The GDP implicit price deflator represents all activity in the US economy, including government spending and entrepreneurial investments. As noted by Dunn and collaborators, “to compare the amount of societal resources expended in different periods (i.e., opportunity cost), the preferred index is the GDP implicit price deflator, which captures overall economic activity rather than consumer spending.” (p. 184) (Dunn et al., 2018).

2.5. Study quality assessment

The authors reviewed the economic evaluations and identified the reporting quality of each study using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist. The 28-item CHEERS checklist was updated in 2022 (Fayanju et al., 2021; Husereau et al., 2022) and is designed to improve reporting of economic findings. Two reviewers (EF and IDM) assessed each of the included articles for reporting quality independently against the criteria to calculate a score out of 28 (or the number of applicable items). Each item on the checklist was assigned one point if it fully met and half a point if an article partially met the criteria. Three reviewers (EF, IDM, KEM) then discussed differences in criteria ratings to reach a consensus, and then calculated a percentage score for each study. If a criterion did not apply to a particular economic evaluation, “Not Applicable” was indicated. We defined a study as good quality if it scored 71% or higher, average quality if equal or between 32 and 70%, and studies scoring below 32% were classified as poor quality.

A second area of quality assessment is based on study design and type of data used, with evidence from randomized clinical trials (RCTs) being preferred to observational studies for evaluating causality between intervention and outcomes. These implications are discussed in the Results, Discussion, and Limitations sections that follow.

3. Results

As noted in Section 2.2, the search identified 729 articles for screening, ultimately excluding 614. Most of these excluded studies did not include any measures of economic benefits (or societal costs) associated with an SUD intervention. The authors conducted a full-text review of 115 eligible papers. Of these, we excluded 103 for not meeting the criteria related to study type (i.e., evaluation of a SUD intervention) or because they did not include any economic outcomes. The remaining 12 articles comprise this systematic literature review. Of these 12 articles, five were published between 2004 and 2009 (Drummond et al., 2009; Ettner et al., 2006; Koenig et al., 2005; Logan TK, 2004; Parrott et al., 2006), two were published between 2010 and 2016 (Anglin et al., 2013; Polsky et al., 2010), and five were published between 2017 and 2021 (Collins et al., 2017; Dunlop et al., 2017; Elarabi et al., 2021; Horn et al., 2017; McCollister et al., 2018).

The authors summarize the results in three tables below. In each table, studies are grouped into 4 categories reflecting the SUD continuum of care (CoC) framework (Stanojlović & Davidson, 2021): i) Primary prevention for all levels of SUD; ii) Early intervention involving screening for substance use problems at an early stage and providing brief intervention to prevent more severe SUD delivered in a variety of settings such as school clinics, primary care offices, and behavioral health clinics; iii) Pharmacotherapies with FDA-approved drugs that include medication-assisted treatment (MAT) and medications for opioid use disorders (MOUD); and iv) Behavioral intervention that focuses on changing individual behaviors concerning substance misuse (e.g., cognitive behavioral therapy -CBT-, interpersonal therapy, self-help groups, motivational interviewing, and relapse intervention).

Table 1 provides information about each study, including the type of intervention(s) being evaluated, the study population, the study design, comparison condition(s), and a summary of the main findings. Table 2 provides an overview of each study’s economic analysis methods including analytic perspective, time horizon, health and economic outcomes, and type of economic evaluation. Table 3 then presents the core findings of this review by summarizing the average economic benefits across (up to) six unique outcome domains. Results supporting causal links between intervention and economic benefits are highlighted. A final table provides results from the CHEERS checklist on overall study quality.

Table 1.

Characteristics of the 12 studies included in the review.

Study and intervention	Population	Comparator	Study design	Results
Primary prevention
Anglin et al. 2013 CBA of the implementation of Substance Abuse and Crime Prevention Act (SACPA).	88,962 offenders, 34 years old on average, convicted of nonviolent drug offenses or probation/parole violations: 47.2% white	A time-lagged cohort of SACPA-eligible individuals’ pre-implementation	Observational	SACPA generated savings primarily through reduced incarceration.
Early Intervention
Collins et al., 2017 CEA/CUA of a UK-based Drug Intervention Program (DIP).	266 individuals ages 18 and older that were in contact with the DIP	Pre-DIP outcomes	Observational	DIP associated with reduced costs of crime, increased quality-of-life, and reduced subsequent drug use.
Drummond et al., 2009 CEA/CUA stepped care for alcohol use disorder	1 794 male primary care patients, 42 years old on average, screening positive for alcohol use disorder	Nurse-led minimal intervention	RCT	The stepped care intervention resulted in greater cost savings compared with the minimal intervention.
Logan et al., 2004 CBA of adult drug courts in Kentucky	745 individuals 33 years old on average (73% male); 65% African American and 34% were white	Individuals who were assessed for the Drug Court programs but did not enter the programs	Observational	For graduates, Drug Court involvement was associated with reductions in incarceration, mental health services, and legal costs, as well as increases in earnings and child support payments.
Horn et al., 2017 CBA of SBIRT in emergency departments [screening, assessment referral (SAR) or SAR plus brief intervention]	1 285 (69.3% male and 31% female) between 8 to 72 years old enrolled in the SMART-ED clinical trial; 51.6% white	Minimal screening only (MSO)	RCT	The SBIRT models implemented in this trial did not generate net economic benefits. No statistically significant differences in costs or outcomes across intervention and control groups.
Parrott et al. 2006 CEA/CUA of direct access to alcohol detoxification service.	103 individuals admitted for alcohol detoxification at two treatment centers in the UK	Postdetoxification (clients served as own controls)	Observational	Both services delivered a flexible needs-based service to very disadvantaged population at a reasonable cost and were associated with statistically significant reductions in drinking.
Medical treatment (MOUD and MAT)
Dunlop et al., 2017 CEA of take-home buprenorphine-naloxone (BNX).	50 individuals 38 years old on average diagnosed with heroin use disorder	Wait list control	RCT	Intervention group had significantly lower crime costs and improved mental health and quality of life relative to those remaining on the wait list.
Elarabi et al., 2021 CBA BUP/NX-F with incentivized abstinence and adherence monitoring.	141 patients 18 and older with active criminal involvement with a likelihood of incarceration	BUP/NX-F in usual care	RCT	The experimental intervention applying contingency management and using BPN home-doses/prescriptions as the incentive generated higher net economic benefits compared to usual care.
Ettner et al., 2006 CBA of California Treatment Outcome Project (CalTOP)	Direct costs of 43 substance abuse treatment providers in 13 countries in California.	Pre-post treatment (clients served as own controls)	Observational	CalTOP shows that allocating taxpayer dollars to substance use disorder treatment pays off in reduced incarceration costs and increased employment earnings.
Polsky et al. 2010 CEA of buprenorphine-naloxone (BPN) versus 14-day detox	152 individuals 19 years old on average who met DSM-IV criteria for opioid dependence	14-day detox taper	RCT	BUP has a high probability of being cost effective at a threshold of $100 000 per QALY gained.
Behavioral Interventions and treatment
Koenig et al., 2005 CBA of SUD treatment in Ohio	1 832 individuals (57% male and 43% female) 18 years and older enrolled in treatment in Cuyahoga County, OH; 73% were African American	Pre-post baseline (clients served as own controls)	Observational	Treatment was cost beneficial regardless of the number of times a client entered treatment in the baseline or follow-up periods. Clients who entered residential treatment and then stepped down to less intensive care showed greater treatment benefits than clients who only received residential treatment.
McCollister et al 2018 CBA juvenile drug court/reclaiming futures (JDC/RF).	245 participants across 5 sites	Standard JDC	Observational	JDC/RF relative to standard JDC generated savings from reduced physical/mental health problems, illegal activity, and missed school days.

CBA, cost-benefit analysis; CUA, Cost-utility analysis; CEA, cost-effectiveness analysis; SUD, Substance Use Disorder.

Table 2.

Economic Methods Across the 12 Studies Included in the Review

Study	Perspective(s)	Time horizon	Health outcome(s)	Economic outcome(s)	Economic evaluation	Discount rate	Year and currency
Primary prevention
Anglin et al. 2013	Societal and justice system	30 months	NR	HC, CJ	CBA	NR	United States/USD (2009)
Early Intervention
Collins et al., 2017	Societal and healthcare system	12 months	QALYs (SF-12)	HC, CJ	CUA, CEA	NR	Great Britain/GBP (2013)
Drummond et al., 2009	Societal	6 months	QALYs (EQ-5D, SF-12)	HC, CJ, SS	CEA, CUA	NA	United States/USD (2001)
Logan et al., 2004	Societal	12 months	NR	HC, CA, W, O	CBA	NR	United States/USD (1999)
Horn et al., 2017	Societal and healthcare system	12 months	NR	HC, CA, W	CBA	NR	United States of America/USD (2015)
Parrott et al. 2006	Societal	8 months	QALYs (EQ-5D)	HC, CA, SS	CUA, CEA	NA	Great Britain/GBP (2004)
Medical treatment (MOUD and MAT)
Dunlop et al., 2017	Societal	3 months	NR	HC, CA	CBA	NA	Australia/AUD (2009)
Elarabi et al., 2021	Societal	5-, 4-, and 1-month inpatient treatment, 4 months outpatient treatment	NR	CJ, W, O	CBA	NA	United Arab Emirates/AED (2012)
Ettner et al., 2006	Societal	9-month follow-up interview	NR	HC, CA, CJ, W	CBA	NA	United States of America/USD (2001)
Polsky et al. 2010	Societal and Healthcare system	12 months	QALYs (EQ-5D)	HC, CA, W	CEA	NR	United States of America/USD (2006)
Behavioral Interventions
Koenig et al., 2005	Societal	30 months	NR	HC, CA, W, SS	CBA	NR	United States of America/USD (1997)
McCollister et al 2018	Societal and Juvenile justice	12 months	NR	HC, CA, W	CBA	NR	USA, USD (2012)

EQ-5D, Euroqol–5 Dimensions; SF-12, Short-Form 12; CBA, cost-benefit analysis; CUA, Cost-utility analysis; CEA, cost-effectiveness analysis; HC, healthcare; CA, criminal activity; CJ, criminal justice services; SS, social services; W, work earnings; O, other; NA, Not applicable & NR, Not reported.

Table 3.

Average (per client) annual economic benefits of addiction interventions in 2021 USD

Study	Avoided healthcare utilization	Avoided Criminal Activity	Avoided Criminal Justice	Productivity	Avoided Social Services	Other Consequences	Average cost per client	Economic benefits per clienta	Net economic benefit per client b
Primary prevention
Anglin et al., 2013  CA Substance Abuse and Crime Prevention Act (SACPA)	$ (110.28)	NR	$ 8,355	NR	NR	NR	$ 665	$ 8,245	$ 7,580
Early intervention
Collins et al., 2017  Drug Intervention Programs	NR	$ 3,470	NR	NR	NR	NR	$ 2,324	$ 3,3470	$ 1,146
Drummond et al., 2009  Stepped care for alcohol use disorder	$ 47	NR	$ 7,454	NR	$ 140	$6,975	$ 932	$14,615	$13,683
Horn et al., 2017
Screening and referral vs. minimal screening only	$ (1,029)	$ (3,880)	NR	$ (751)	NR	NR	$ 103	$ (5,660)	$ (5,763)
Screening, referral, brief intervention vs. minimal screening only	$ (2,183)	$ (621)	NR	$ 507	NR	NR	$ 168	$ (2,298)	$ (2,465)
Logan et al., 2004
Drug court graduates	$ 1,119	$ 14,783	NR	$ 24,495	NR	$ 2,037	$ 7,908	$ 42,430	$ 34,522
Drug court terminators	$ 404	$ 1,914	NR	$ 1,014	NR	$ 455	$ 2,760	$ 3,780	$ 1,028
Parrot et al., 2006
Smithfield detoxification service	$ 2,707	$ 3,187	NR	NR	$ 128	NR	$2,853	$ 6,022	$ 3,169
Plummer detoxification service	$ 1,120	$ 2,530	NR	NR	$ 400	NR	$ 2,702	$ 4,050	$ 1,348
Medication treatment (MAT and MOUD)
Dunlop et al., 2017  Take-home buprenorphine-naloxone	$ 6,129	$ 50,961	NR	NR	NR	NR	NR	$ 44,832	$ 44,832
Elarabi et al., 2021  Buprenorphine/naloxone film, contingency management	NR	NR	$ 107,016	$ 12,515	NR	$ 93,143	$ 137,306	$ 212,674	$ 75,368
Ettner et al., 2006
All SUD treatments	$ 1,329	$ 11,229	$ 3,537	$ 6,631	NR	NR	$ 3,131	$ 22,726	$ 19,595
Methadone maintenance	$ 6,596	$ 2,059	$ 55.39	$ 1,745	NR	NR	$ 5,415	$ 10,455	$ 5,041
Outpatient treatment	$ 455	$ 11,355	$ 1,636	$ 4,453	NR	NR	$ 1,658	$ 17,899	$ 16,242
Residential treatment	$ 2,116	$ 12,208	$ 7,055	$ 10,782	NR	NR	$ 5,694	$ 32,161	$ 26,639
Polsky et al., 2010  buprenorphine-naloxone (BupNx)	$ (15,925)	$ (2,623)	NR	$ (14,830)	$ 26,125	NR	$ 3,000	$ (2,007)	$ (5,006)
BupNx Tapier in Detox	$ (8,702)	$ (60,659)	NR	$ (13,214)	$(20,964)	NR	$ 773	$ (59,707)	$ (58,934)
Behavioral intervention
Koenig et al., 2005  SUD treatment in Ohio	$ 727	$ 3,626	NR	$ 4,731	$ 1,543	NR	$ 2,700	$ 10,628	$ 7,928
McCollister et al., 2018  Juvenile Drug Court/Reclaiming Futures (JDC/RF)	$ 193	$ 193,440	NR	$ 268	NR	NR	$ 60,429	$ 193,901	$ 133,472
Standard JDC	$ 350	$ 122,855	NR	$ 204	NR	NR	$ 40,159	$ 122,408	$ 82,249

^aEconomic benefit per client is the summation of avoided healthcare utilization, avoided criminal activity, avoided criminal justice, productivity, avoided social services, and other consequences (i.e., accidents, all services). The original studies used various empirical models to estimate total economic benefits. Thus, the numbers presented here do not exactly match those found in the initial studies as we applied the inflation factor to 2021 USD, used the exchange currency for each foreign research, and performed these calculations annually.

^bNet economic benefit per client equals average treatment benefit minus average treatment cost. NR = not reported. NA = not applicable.

3.1. Study characteristics

3.1.1. Interventions, populations, and study designs

Table 1 provides an overview of the 12 included studies, which were conducted in the United States (n=8) (Anglin et al., 2013; Drummond et al., 2009; Ettner et al., 2006; Horn et al., 2017; Koenig et al., 2005; Logan TK, 2004; McCollister et al., 2018; Polsky et al., 2010), United Kingdom (n=2) (Collins et al., 2017; Parrott et al., 2006), Australia (n=1) (Dunlop et al., 2017), and United Arab Emirates (n=1) (Elarabi et al., 2021). The populations represented in these studies feature a broad age range (15 and older), but only six of the studies reported sex and/or gender (Dunlop et al., 2017; Ettner et al., 2006; Horn et al., 2017; Koenig et al., 2005; Logan TK, 2004; Parrott et al., 2006). Of those, two reported the percentages of both sexes included in the study (Horn et al., 2017; Koenig et al., 2005), two included only males (Dunlop et al., 2017; Logan TK, 2004), and one had only females (Parrott et al., 2006). Four studies reported race/ethnicity data for participants (Anglin et al., 2013; Horn et al., 2017; Koenig et al., 2005; Logan et al., 2004).

Study designs were of two primary types: randomized controlled trials (RCT; n=5) or observational cohorts (n=7). The observational studies created comparison groups using various statistical approaches, or studies employed pre-post designs and clients served as their own controls. Most studies found their featured interventions were associated with positive net economic benefits, although statistical significance of these findings varied. For instance, two studies showed no statistical differences between experimental or control conditions for any of the economic outcomes suggesting that the more costly interventions were not a good value relative to a low-cost control condition.

3.1.2. Economic methods across studies

Table 2 shows the economic evaluation methods employed in these studies that include both cost-benefit analysis (CBA; n = 8) (Anglin et al., 2013; Dunlop et al., 2017; Elarabi et al., 2021; Ettner et al., 2006; Horn et al., 2017; Koenig et al., 2005; Logan et al., 2004; McCollister et al., 2018), and cost-effectiveness or cost-utility analysis (CEA/CUA; n = 4) (Collins et al., 2017; Drummond et al., 2009; Parrott et al., 2006; Polsky et al., 2010). Cost-effectiveness analysis that includes quality-adjusted life years as a primary outcome is called cost-utility analysis (CUA). CEA and CUA studies are not required to translate outcomes into dollars. Still, many CEA/CUAs are framed from the societal perspective, which incorporates costs (or reductions in costs during intervention) across multiple outcome domains such as health care utilization, productivity, and criminal justice services. We opted to include CEAs that calculated the broader societal costs of a treatment program or intervention as these are comparable to cost-benefit results.

The time horizon in two of the included studies was 2.5 years (30 months) (Anglin et al., 2013; Koenig et al., 2005), but the study did not apply a discount rate in the cost-benefit calculations, and provided no justification. In five studies, the time horizon was one year (12 months) (Collins et al., 2017; Horn et al., 2017; Logan TK, 2004; McCollister et al., 2018; Polsky et al., 2010), and the rest of the studies reported findings over a period of less than 12 months, in which case a discount rate is not typically applied (Husereau et al., 2022).

3.2. Annual economic benefits

Table 3 summarizes the average annual economic benefits of the 12 included studies, organized by Continuum of Care (CoC) framework category (Stanojlović & Davidson, 2021). Economic benefits estimated from RCTs are discussed in terms of potential causal links to SUD interventions and benefits generated from observational studies are discussed in terms of potential associations with interventions. This distinction is obviously critical for evaluating the significance of these reported findings for resource allocation decisions.

3.2.1. Primary prevention

Anglin and colleagues (Anglin et al., 2013) estimated the net economic benefits of the California Substance Abuse and Crime Prevention Act (SACPA). SACPA, implemented on July 1, 2001, mandates SUD treatment combined with probation or continued parole instead of incarceration for adult offenders convicted of nonviolent drug offenses and/or probation/parole violations. The authors used data from the California Department of Justice and other state administrative databases to construct a time-lagged cohort of individuals meeting SACPA eligibility criteria in the year before the program was enacted (July 1, 1997–June 30, 1998; n = 47,355) to compare with the intervention group comprising SACPA-eligible nonviolent drug offenders convicted within the first 12 months of SACPA implementation (July 1, 2001–June 30, 2002; n = 41,607).

The economic benefits were estimated from reduced health care utilization and reduced criminal justice system costs. Criminal justice outcomes included days in prison and jail, days on parole, days on probation, arrests, and convictions. A multivariate difference-in-differences (DID) approach estimated the effect of SACPA implementation on economic outcomes. SACPA was associated with an average (per person) 12-month economic benefit of $8,245, comprised of increased health care utilization costs ($110.28) and reduced criminal justice costs ($8,355). After accounting for the costs of SUD treatment mandated by SACPA ($665 per person), the average net benefits would be $7,580. Results were statistically significant with the exception of criminal justice conviction costs. The authors acknowledge that the main challenges in conducting this study were reliance on a time-lagged comparison cohort and that health care utilization was available only for individuals enrolled in California’s Medicaid plan.

3.2.2. Early intervention

Collins and colleagues (Collins et al., 2017) used observational data to create a Monte Carlo simulation model to estimate the costs and benefits of the Drug Intervention Program (DIP), a UK-based diversion program for adult offenders with SUD. The premise of DIP is to direct adult offenders out of prison into treatment programs for their SUD. The authors conducted their analysis from the societal and health care system perspectives using administrative data from a borough in West North England with approximately 320,000 people. Simulation models create hypothetical cohorts of patients to predict costs and benefits over a range of scenarios and timelines.

The study followed individuals in contact with the DIP for six months, from April to September 2013. The study estimated economic benefits from reduced criminal activity, changes in drug treatment costs, and quality-adjusted life-years (QALYs). Comparing pre- and post-DIP costs showed a statistically significant decrease in self-reported criminal offenses valued at $3,470 per client. The cost of DIP was $2,324 on average, thus the net economic benefit per client was $1,108. Due to the nature of the data analyzed in this study and issues regarding regression to the mean, the authors conducted a sensitivity analysis in which they found that clients who had no further DIP contact, often because they refused further assessment, still showed a significant crime reduction.

Drummond and colleagues (Drummond et al., 2009) used data from an RCT to evaluate the cost-effectiveness of screening for alcohol use disorder (AUD) and a stepped-care intervention. The CEA was performed from a societal perspective. Economic outcomes were estimated comparing baseline to 6-months postintervention and included health care utilization, criminal justice events, social services, and vehicle accidents. The authors used the Alcohol Use Disorder Identification Test (AUDIT) to screen 1,794 male primary care attendees at six South Wales, UK, practices. The study assessed incremental cost per quality-adjusted life-years gained (QALYs) using the Euro-Qol 5D (EQ–5D).

The average 12-month economic benefit associated with the intervention (extrapolated from 6-month follow-up data), was estimated to be $14,615, generated by reduced criminal justice services ($7,454) and reduced automobile accidents categorized as “other consequences” ($6,975). Economic benefits also included small reductions in health care utilization ($47) and social services utilization ($140). The average intervention cost was $932; thus, the net economic benefit per client was estimated to be $13,683. None of these findings were statistically significant, however. Study limitations included the fact that only one-quarter of potential participants gave consent to participate in the study. The authors also found that those who agreed to participate were older and had higher AUDIT scores, indicating that participants may not represent the general population of primary care patients. This selection bias may also show differences in willingness to participate in research and the desire to receive alcohol interventions if offered.

Horn and colleagues (Horn et al., 2017) estimated the economic benefits of screening, brief intervention, and referral to treatment (SBIRT) in U.S. Emergency Departments (SMART-ED). The study performed a series of sensitivity analyses to evaluate the robustness of their assumptions and selected unit cost estimates. The design was a three-arm RCT in which the study randomly assigned patients reporting alcohol misuse to one of three treatment conditions: 1) minimal screening only (MSO); 2) screening, assessment, and referral to treatment (SAR); and 3) SAR plus brief intervention, with two telephone follow-up booster sessions (BI-B). Participants in the MSO group received an informational brochure about substance use (its consequences and types of treatment) and no further intervention. Participants in the SAR group received the same informational brochure, minimal scripted feedback about their substance use, and a recommendation to seek treatment if indicated or requested. Individuals in the BI-B group also received the informational brochure, feedback, advice to seek treatment (if indicated or requested), and an additional 30-minute brief intervention comprising up to two 20-minute booster sessions. The analysis was framed from the societal and health care perspectives and based on a sample of 1,285 trial participants.

Self-reported outcome data were collected at 3, 6, and 12 months following baseline assessment. Regarding MSO, participants in SAR reported decreased productivity and increased health care utilization, social services utilization, and criminal activity following the intervention. The comparison between BI-B and MSO outcomes also showed an increase in health care utilization, criminal activity, and social services utilization but an average increase in productivity of $507 per participant. Differences in economic benefits were not statistically significant for any comparisons. The results indicate that these interventions were not cost-saving over the initial 12 months following the baseline. Study limitations noted by the authors included reliance on self-reported data and assumptions about program costs, which may differ if the programs were implemented in community-based settings (i.e., transitioned from a clinical trial) and/or implemented in other locations.

Logan and colleagues (Logan TK, 2004) conducted a cost-benefit analysis of three adult Drug Court programs in Kentucky using observational data on Drug Court program participants. The authors compared drug court participants to a matched control group of individuals eligible for drug court but who never entered the programs. Authors included participants who graduated (n=222) and terminated (n=371) from these Drug Court programs from 1996 to 1998. Data collection for the evaluation began in November 1999. The study used multivariate predictive models to estimate economic benefits.

The economic benefits per drug court graduate were estimated to be $42,430, resulting from the increase in annual earnings ($24,491), reduction in criminal activity ($337), reduction in motor vehicle accidents, reductions in child support payment deficits ($2,037), and a reduction in healthcare utilization ($1,119). After accounting for average treatment costs ($7,908 per graduate), graduating from Drug Court was associated with an average net benefit of $34,522 per client. For drug court terminators, the average cost of drug court was $2,760. Terminating the drug court program was associated with positive economic benefits ($3,780) and net economic benefits ($1,028), which includes a reduction in criminal activity ($1,914), an increase in annual earnings ($1,014), a reduction in accidents and child support payment deficits ($455), and a reduction of health care utilization ($404). Economic benefits were statistically significant except for estimated benefits attributable to reduced traffic accidents.

In an observational cohort study, Parrot and colleagues (Parrott et al., 2006) estimated the costs and benefits of two services for alcohol-dependent adults in the UK: the Smithfield Court, a direct-access alcohol detoxification service in Manchester, and the Plummer Court, an NHS partial hospitalization program in Newcastle. All admissions for alcohol detoxification during eight months were eligible for study recruitment. A medically supervised detox center at Smithfield Court admitted 283 patients over the recruitment period. Of those, 145 (51%) entered the study. At Plummer Court, a detox center admitted 113 patients, of whom 77 (68%) entered the study. This study adopted health care and societal perspectives. By comparing pre- and post-discharge data on changes in criminal activity, health care utilization, and social services utilization, authors estimated economic benefits for both groups using a comprehensive regression model on societal costs but changes in costs were not statistically significant for either site. Authors report that Smithfield Court had an average 12-month economic benefit of $6,022, comprising a reduction in avoided health care utilization ($2,707), avoided criminal activity ($3,187), and avoided social services ($128). The Plummer Court intervention had an estimated annual benefit of $4,050, resulting from a decrease in criminal activity ($2,530), health care utilization ($1,120), and social services ($400). Had these results been significant, both sites would be associated with positive net economic benefits (Smithfield $3,169; Plummer $1,348).

3.2.3. Medication treatment (MAT and MOUD)

Dunlop and colleagues (Dunlop et al., 2017) performed an economic evaluation alongside an Australian RCT of increasing access to buprenorphine-naloxone (BupNX), an opioid agonist treatment for opioid use disorder (OUD). Data came from an opioid treatment clinic in Newcastle, Australia. The study examined whether patients with heroin dependence, who were dispensed BupNX weekly, had more significant reductions in heroin use and related adverse health effects 12 weeks after commencing treatment than patients on a waitlist (the study control condition). The study was conducted from the societal perspective using a recruited sample of 50 patients with DMS-IV heroin dependence (and no other substance dependence).

The study assessed trial subjects at baseline (day zero) and the end of week four (day 28), week eight (day 56), and week 12 (day 84) for heroin use, other substance use, health outcomes, quality of life, and criminal activity. The intervention group (n=25) received take-home self-administered sublingual buprenorphine-naloxone weekly, and the control group (n=25) received no clinical intervention. The total economic benefit of the BupNX intervention amounted to $50,961 per participant, generated by reductions in criminal activity. The opioid treatment clinic did not report BupNX costs, but total health care costs were higher in the intervention group ($6,129 over 12 months). Subtracting health care costs from the benefits of reduced crime results in a statistically significant net economic benefit of $44,832. The authors noted that the enrollment criteria could impact the generalizability of the findings. These enrollment criteria did not include participants dependent on alcohol, benzodiazepines, amphetamines, or cocaine.

Elarabi and colleagues (Elarabi et al., 2021) estimated the economic burden of OUD in the United Emirates Arab and the costs and benefits of buprenorphine/naloxone treatment in a 16-week outpatient RCT of 141 adults with OUD stabilized on buprenorphine/naloxone film (BUP/NX-F), approximately half of which were randomized to BUP/NX-F with incentivized abstinence and adherence monitoring (experimental n=70; control group n = 71). Data came from the Work and Social Adjustment Scale (WSAS), a five-item self-reported scale used to quantify functional impairment related to personal, social, and occupational factors associated with patients’ OUD. BUP/NX-F with Incentivized Abstinence and Adherence Monitoring generated an average 12-month economic benefit of $212,674 per participant, comprised of reduced criminal justice costs ($107,016), reduced drug purchases captured under the category “other consequences” ($93,143) and increased productivity ($12,515). After accounting for the intervention costs ($ 137,306), the average net benefits per client were $75,368. These results are statistically significant. The authors acknowledge that one main limitation of this study is that they did not account for the health costs averted.

In an observational cohort study, Ettner and colleagues (Ettner et al., 2006) conducted a cost-benefit analysis of substance use disorder treatment modalities from 43 providers in 13 counties in California during 2000–2001. The authors utilized data from the California Drug Abuse Treatment Cost Analysis Program (CalTOP) to calculate weighted and unweighted average per diem treatment costs over a 9-month follow-up period. Health care utilization, criminal-legal outcomes, and work productivity comprised economic benefits.

The authors grouped data into four categories: i) all treatment modalities, ii) methadone maintenance, iii) outpatient treatment, and iv) residential treatment. All treatment modalities (n=2,567) showed a decrease in health care utilization ($1,329), criminal activity ($11,229), criminal justice contacts ($3,537), and an increase in productivity ($6,631). After accounting for the treatment costs ($3,131), the average net benefits were $19,595. Methadone maintenance (n=115) was associated with a decrease in health care utilization ($6,596), criminal activity ($2,059), criminal justice contacts ($55.39), and an increase in productivity ($1,745), but these results were not statistically significant. Outpatient treatment (n=1,585) showed a significant decrease in health care utilization ($455), criminal activity ($11,355), criminal justice contacts ($1,636), and an increase in productivity ($4,453). After accounting for the outpatient costs ($1,658), the average net benefits were $16,242. Residential treatment (n=867) generated a significant decrease in health care utilization ($2,116), criminal activity ($12,208), criminal justice contacts ($7,055), and an increase in productivity ($10,782). After factoring in the cost of residential treatment ($5,694), the net benefit per client was $26,639.

Polsky and colleagues (Polsky et al., 2010) estimated the benefits and costs of buprenorphine-naloxone (BupNx) treatment versus brief detoxification (BupNx-DETOX) treatment in opioid-dependent youth enrolled in a randomized controlled trial in the United States. The authors framed the analysis from societal and health care sector perspectives, and estimated benefits for four outcomes: health care utilization, productivity, social services, and criminal activity (assault, robbery, auto theft, shoplifting, and drug offenses). The study enrolled 152 participants aged 15–21 randomized between July 2003 and December 2005 to either a 2-week BupNx-DETOX or a 12-week course of BupNX. The study collected measures at baseline, weeks 4, 8, and 12, and months 6, 9, and 12. For this review we compared societal costs generated by BupNX and DETOX over the 1-year follow-up. The BupNX intervention cost $3,000 on average and was associated with an increase in health care utilization ($15,925), a loss in productivity ($14,830), and increased criminal activity ($2,623). This group did show a reduction in social services costs ($26,125), leading to a 12-month economic loss of $2,007 per client. The BupNx DETOX intervention had relatively higher societal costs: $8,702 in health care utilization, $60,659 in higher criminal activity costs, $13,214 in lost productivity, and $20,964 in higher social services costs. Total economic losses for BupNX-DETOX were $59,707. The authors acknowledge that while BupNX had relatively lower costs, none of the differences in economic outcomes were statistically significant.

3.2.4. Behavioral interventions

Koenig and collaborators (Koenig et al., 2005) conducted an observational study estimating the long-term costs and benefits for SUD clients who received treatment in Cuyahoga County, OH, from the societal perspective using data from the Persistent Effects of Treatment Studies (PETS). To test the continuum of care, they developed two regression models to predict the net costs to the nontreated population and net costs to society. Benefits included reduced crime-related costs, health care costs, social services costs, and increased client earnings over a 12-month follow-up period. Based on a sample of 1,237 clients who completed a baseline interview and 595 clients who completed the follow-up interview at 12 months, the average cost per client for SUD treatment was $2,700. Treatment was associated with average economic benefits of $10,628 [reduced health care costs ($727), greater earnings for clients reporting any income ($4,731), reductions in criminal activities ($3,626), and reductions in social services utilization ($1,543)]. Net economic benefits would be $7,928 per person. Results were statistically significant.

Study limitations noted by the authors included missing data on arrests and incarceration and sample selection biases using a pre-treatment/post-treatment analytic framework. Pretreatment costs may not be a reasonable estimate of societal costs in the absence of treatment because many clients may be entering treatment at an atypical time for their earnings, use of government financial assistance, criminal activity, and other measures. Moreover, some clients have chosen to voluntarily seek treatment, whereas others are required to enter treatment as a condition of probation or parole.

McCollister and collaborators (McCollister et al., 2018) examined the economic impact of juvenile drug court (JDC) programs for juvenile offenders with substance use problems. As part of the National Cross-Site Evaluation of JDC and Reclaiming Futures (RF), an observational cohort study of the economic analysis of five JDC/RF programs was conducted from a multisystem and multiagency perspective. The study highlights the direct and indirect costs of JDC/RF and the savings generated from reduced health problems, illegal activity, and missed school days based on follow-up interviews focused on a 12-month timeframe. The average treatment client enrolled in the JDC/RF program was associated with a total 12-month economic benefit of $193,901, comprised chiefly of reduced criminal activity costs ($193,440); other benefits included reduced health care utilization ($193) and greater earnings ($268). After factoring in the cost of JDC/RF ($60,429), the average net economic benefits would be $133,472. The standard JDC program was associated with total economic benefits of $122,408, also comprised chiefly of reduced criminal activity costs ($122,855), with additional benefits in decreased health care utilization ($350) and greater earnings ($204). After accounting for treatment costs ($40,159), the average net benefits associated with the standard JDC program were $82,249. The authors acknowledge several study limitations. First, they found some inconsistencies in reporting of costs across sites resulting from different levels of detail provided in financial records and other data; second, outcomes were based on self-reported data; finally, the authors did not have complete expenditure information for all substance use disorder treatment received by JDC/RF participants.

3.3. Quality assessments

The twelve studies in this SLR were determined to adhere to 86% (median =88%, range = 80–90%) of applicable items from the CHEERS checklist (Table 4). Appendix A shows the items that were fully or partially satisfied for the CHEERS checklist. The average quality of the twelve studies in this review was good. Four studies (Drummond et al., 2009, Parrot et al., 2006, Dunlop et al., 2017, Ettner et al., 2006; Elarabi et al., 2021) did not report a discount rate (criterion #10) because the time horizon was less than 12 months; eight studies (Anglin et al., 2013; Drummond et al., 2009; Horn et al., 2017; Logan et al., 2004; Dunlop et al., 2017; Elarabi et al., 2021; Polsky et al., 2010; McCollister et al., 2018) did not have a formal model as applicable to the analysis (criterion #16). We also identify that none of the studies reported a formal health economic analysis plan (criterion #4). Finally, not all studies described how they addressed uncertainty and other biases in the analyses (e.g., through a sensitivity analysis) (criterion #20).

Table 4.

CHEERS percentage applicability of each study

Author	% Applicable CHEERS criteria met	Overall reporting quality
Primary Prevention
Anglin et al., 2013	67%	Average
Early Prevention
Collins et al., 2017	86%	Good
Drummond et al., 2009	83%	Good
Elarabi et al., 2021	78%	Good
Ettner et al., 2006	78%	Good
Horn et al., 2017	71%	Average
Logan et al., 2004	83%	Good
Parrot et al., 2006	82%	Good
Medication treatment (MAT and MOUD)
Dunlop et al., 2017	78%	Good
Polsky et al., 2010	83%	Good
Behavioral intervention
Koenig et al., 2005	71%	Average
McCollister et al., 2018	78%	Good

4. Discussion

The studies featured a broad variety of interventions, populations, study designs, analysis time horizons, locations, and economic outcomes. More than half of the studies included in this review (7 of 12) relied upon data from observational cohorts, which raises concerns regarding missing data, omitted variables bias, confounding, and causal interpretations. Although we include summaries of the estimated costs and economic benefits for all studies, any findings based on observational cohorts should be viewed as associative and not causal. Of course, RCTs have nonsignificant results and when applicable we have highlighted this in the Results section.

Eight studies were CBAs and four were CEA/CUAs. In the future, more CEA studies may include the economic outcomes that can estimate benefits based on the recommendations of the Second Panel on Cost-Effectiveness in Health and Medicine published in 2016 (Sanders et al., 2016). The Second Panel recommends that CEAs adopt both a health care sector perspective and a societal perspective (2 reference cases). The panel created an impact inventory of costs and outcomes, including several types of economic consequences across many domains and including all of the domains featured in this review. These recommendations have helped to move CEA closer to CBA in terms of the results presented (i.e., reductions in societal costs). The characteristics of these studies reflect the heterogeneity and complexity of economic evaluations of behavioral health interventions, making it challenging to compare methods and results. Future studies should consider using recommended instruments for measuring economic outcomes to support the standardization and/or harmonization of economic findings (McCollister et al., 2018; Papp et al., 2021).

Studies focused on the economic impact of early intervention (5), MOUD (3), behavioral interventions (3), and primary prevention (1). Domains representing criminal activity or criminal justice costs were included in all studies: 9 included criminal activity from self-reported offending, and 7 had criminal justice costs. Surprisingly, many studies included only 1 or 2 economic outcome domains. This fact is clearly driven by limited access to data that report counts of services or events in a way that supports the monetization of these outcomes. Among the 10 studies that identified positive economic benefits, reductions in criminal activity or criminal justice costs represented the largest or second largest component of these benefits (range $621 to $193,440 per client). These studies included results from observational data, which supports only associative links between intervention and economic benefits. Among the RCT studies included in this review, two studies did not find significant positive economic benefits for the intervention groups (Polsky et al., 2010; Horn et al., 2017).

Studies were conducted in countries including the United States, Australia, United Arab Emirates, and the UK, suggesting that benefits from reducing crime and health care utilization are context dependent. It is obvious that this can influence criminal justice costs (and laws and practices underlying them) and health care costs and practices that differ across countries.

In this review, seven studies found savings associated with reduced health care utilization, of which two evaluated alcohol use disorder interventions. Three other studies reported increases in health care utilization costs. Health care utilization savings were generally lower than criminal activity or criminal justice cost savings, ranging between $34 and $9,898 per client. Six studies found improved productivity associated with interventions ranging between $185 and $18,151. Four studies estimated moderate reductions in the utilization of social services ranging between $99 and $14,216. Two studies identified other consequences: decreased drug purchases following intervention in the United Arab Emirates, and reduced motor vehicle accidents and improved adherence to child support payments in Kentucky. Future studies may reveal that reduced reliance on social services, productivity impacts, and other benefits can significantly contribute to the economic benefits of interventions. Still, we need more data to understand the relative importance of these domains.

5. Limitations

Our review has several limitations. First, this review was limited to journal articles in English. It did not consider grey literature, such as reports, newsletters, white papers, and dissertations/masters theses, which may report program costs and benefits estimates. Second, we applied a straightforward method to adjust the economic benefit estimates to reflect the 12-month total and net benefits in 2021 dollars if benefits will accrue at a fixed rate over time. Third, we followed previous systematic literature reviews by calculating CHEERS checklist scores. Also, not all checklist categories may apply in some of these studies, likely because the CHEERS statement’s development was back in 2013, and most papers included in this review were published before. Thus, an unfavorable score may not reflect critical weaknesses in specific domains. Fourth, results presented in this review come from point estimates that often have vast confidence intervals.

The latest can be substantial uncertainty in economic evaluation results, driven by statistical data limitations due to relatively small sample sizes and the nature of some measures like right-skewed criminal activity (driven by many respondents reporting 0). While sensitivity analyses can examine the degree of uncertainty associated with estimating costs and benefits, it is still important to note that the underlying uncertainty in the results of the included studies limits our findings. Finally, the time horizons of most studies did not allow them to identify societal benefits that take longer to accrue as individuals reduce their use of substances, such as greater workforce participation and reduced use of hospital and emergency department health care services (Drummond et al., 2009; Dunlop et al., 2017; Ettner et al., 2006; Horn et al., 2017; Koenig et al., 2005; Logan TK, 2004; McCollister et al., 2018; Parrott et al., 2006). Similarly, without longer time horizons, we cannot observe the potential decay of intervention effectiveness over time that might translate into decreasing economic benefits. A 12-month time horizon, however, is more likely consistent with the time horizon of many policymakers and can inform annual projections about the budget impact and potential return on investment from these interventions.

6. Conclusion

This systematic review identified 12 papers with information on the economic benefits of interventions and treatments for SUD. Consistent with the prior review, reductions in criminal activity accounted for the most significant societal benefits over one year, representing crucial policy implications. Accepting the economic rationale for increased investment in SUD interventions will require recognizing that more benefits accrue to individuals by avoiding being victims of a crime than to governments through budget offsets resulting from savings in non-SUD program expenses. More comprehensive research is necessary to assess societal impacts where budget offsets could occur, including multiple domains beyond criminal activity and health care utilization, for which we found fewer results. Future economic analyses should also examine intervention benefits for specific subpopulations for which budget offsets might be more relevant. Several studies have shown that treatment needs can differ for specific ethnic groups, pregnant women, LGBTQ+ populations, veterans, and criminal justice populations. Future economic studies should also expand to other settings, such as syringe service programs, overdose prevention centers, and other harm reduction programs. Finally, studies should also explore individually tailored interventions to optimize care management, which may yield unexpected economic benefits.

HIGHLIGHTS.

• The greatest economic benefits over one year resulted from a reduction in criminal activity

• Alcohol use disorder interventions were associated with a decrease in healthcare utilization costs

• Reduced productivity following intervention was generally due to reduced labor force participation and lost earnings

APPENDIX A. material to: “Economic benefits of substance use disorder treatment: A systematic literature review of economic evaluation studies from 2003 – 2021”

TABLE A.1:

PubMed search strategy

1	“substance related disorders”[MeSH Terms]
2	“behavior, addictive”[MeSH Terms]
3	1 OR 2
4	“cost benefit analysis”[MeSH Terms]
5	“cost effectiveness”[All Fields]
6	“economics”[All Fields]
7	“cost*”[All Fields]
8	“costs and cost analysis”[MeSH Terms]
9	“costs and cost analysis”[All Fields]
10	4 OR 5 OR 6 OR 7 OR 8 OR 9
11	3 AND 10
12	Date limited: 2003 to 2021

Note. This PubMed/MEDLINE strategy was translated for the other database searches and was reviewed by all authors. Key words related to specific types of substances (e.g., alcohol, opioids, methamphetamine) were not included to reduce complexity. Conflicts were resolved through discussions among the four investigators contributing to this study.

TABLE A.2:

Methodology quality assessment of the included studies using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS).

	Primary intervention	Early intervention					Medication treatment				Behavioral treatment
Criteria	[1]	[2]	[3]	[4]	[5]	[6]	[7]	[8]	[9]	[10]	[11]	[12]
TITLE
1 Title	1	1	1	1	1	1	1	1	1	1	1	1
ABSTRACT
2 Abstract	1	1	1	1	1	1	1	1	1	1	1	1
INTRODUCTION
3 Background and objectives	1	1	1	1	1	1	1	1	1	1	1	1
METHODS
4 Health economic analysis plan	0	0	0	0	0	0	0	0	0	0	0	0
5 Study population	1	1	1	1	1	1	1	1	1	1	1	1
6 Setting and location	1	1	1	1	1	1	1	1	1	1	1	1
7 Comparators	1	1	1	1	1	1	1	1	1	1	1	1
8 Perspective	1	1	1	1	1	1	0	1	1	1	1	1
9 Time Horizon	1	1	1	1	1	1	1	1	1	1	0	1
10 Discount rate	0	0	NA	0	0	NA	NA	NA	NA	0	0	NA
11 Selection of outcomes	0	1	1	0	1	1	1	1	1	1	0	0
12 Measurement of outcomes	0	1	1	0	1	1	1	1	1	1	0	0
13 Valuation of outcomes	0	1	1	0	1	1	1	1	1	1	0	0
14 Measurement and valuation of resources and costs	1	1	1	1	1	1	1	1	1	1	1	1
15 Currency/price conversion	1	1	1	1	1	1	1	1	1	1	1	1
16 Rationale and description model	NA	1	NA	NA	NA	1	NA	NA	1	NA	1	NA
17 Analytics and assumptions	NA	0	NA	NA	NA	1	NA	NA	0	NA	1	NA
18 Characterizing heterogeneity	1	1	1	1	1	1	1	1	1	1	1	1
19 Characterizing distribution effects	1	1	1	1	1	1	1	1	0	1	1	1
20 Characterizing uncertainty	NA	1	NA	NA	NA	0	NA	NA	0	NA	1	NA
21 Approach to engagement with patients and others affected by the study	0	1	0	1	1	1	0	0	0	0	0	1
RESULTS
22 Study parameters	NA	1	NA	NA	NA	1	NA	NA	1	NA	1	NA
23 Summary of main results	1	1	1	1	1	1	1	1	1	1	1	1
24 Effect of uncertainty	0	1	0	0	0	0	0	0	0	1	1	0
25 Effect of engagement with patients and others affected by the study	0	1	0	1	1	1	0	0		0	0	1
DISCUSSION
26 Study findings, limitations, generalizability, and current knowledge	1	1	1	1	1	1	1	1		1	1	1
OTHER RELEVANT INFORMATION
27 Source of funding	1	1	1	1	1	1	1	1		1	1	1
28 Conflicts	1	0	1	0	0	0	1	0		1	0	1

Note. Score of 1 was assigned for positive responses 1 and a score of 0 for negative responses. NA as for Not appliable, NR as for not reported, score 1 is equal to a positive answer and score 0 is equal to a negative answer.

[1] Anglin et al., 2013, [2] Collins et al., 2017, [3] Drummond et al., 2009, [4] Horn et al., 2017, [5] Logan et al., 2004, [6] Parrot et al., 2005, [7] Dunlop et al., 2017, [8] Elarabi et al., 2021, [9] Ettner et al., 2006, [10] Polsky et al., 2010, [11] Koening et al., 2005, [12] McCollister et al., 2018.