We created a search strategy from January 2019 to June 2024 using the following databases: (1) MEDLINE (as of 10th June 2024), (2) EMBASE (as of 10th June 2024), (3) Scopus, (4) PsycINFO, (5) EconLit, and (6) Tufts CEA Registry. We tailored the search algorithms to the specifications of each electronic database and employed validated filters to obtain suitable designs as required. Additionally, we examined the reference lists of previous systematic reviews for potentially eligible studies.

We included comprehensive economic evaluations addressing our research question (see structured PICO question). These evaluations conducted comparative analyses of interventions, examining costs and consequences (outcomes and effects) through cost-effectiveness and cost-utility analysis. Our review encompassed model-based studies using a lifetime horizon and empirical health economic studies using shorter horizons, including economic evaluations based on randomized and non-randomized trials. We excluded cost–benefit studies, partial economic evaluations, conference abstracts, letters to the editor, and studies not published in English.

Two reviewers (LYR, AS) initially assessed search results based on titles and abstracts, followed by full-text reading. Disagreements were resolved by consulting a third reviewer. One reviewer (LYR) extracted the main characteristics of included studies in a pre-designed form, including (1) general information (authors, publication year, country, conflicts of interests); (2) study characteristics (type of intervention, substance, comparator); (3) methodology (model, type of economic evaluation, perspective, time horizon, discounting rate, currency, sensitivity analysis, sources of information); and (4) cost-effectiveness outcomes. A second reviewer performed a quality control of the extraction process (AS).

We assessed the quality of the primary studies that were included by using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) 2022 checklist developed by ISPOR (15). The checklist consists of 28 items grouped into six categories: (1) title and abstract, (2) introduction, (3) methods, (4) results, (5) discussion, and (6) others. Each item received one point if it met the quality criteria and zero points if it did not, resulting in a maximum score of 28 for each study. The results were categorized as “Excellent” if all items were present in the study, “Good” if at least 80% were satisfied, “Fair” if at least 70% of the items were satisfied, and “Average” if at least 60% were satisfied. Two researchers (AS, YR) independently assessed the quality of each study. The results were reviewed by FS. Any discrepancies were resolved through discussion and consensus and final decisions with a third researcher (FS).

We identified 390 individual records through our search process. After removing duplicates, we screened 248 records based on their titles and abstracts. Subsequently, we evaluated 28 studies in full text and ultimately included eight studies in our review (16–23). Figure 1 depicts the PRISMA flow chart, visually representing our screening process. Supplementary Table S1 provides detailed reasons for excluding specific studies.

Table 1 summarizes the characteristics of the included studies. Five studies were conducted in the USA, two in Korea, and one in Canada. Two studies focused on specific contexts in Connecticut and Massachusetts (19, 22). All included interventions concentrated solely on opioids, with none addressing benzodiazepines or stimulants. Costs were expressed in the respective country’s currency. Most studies utilized simulated cohorts constructed with data from relevant sources alongside national and local population estimates. One study was a cost-effectiveness analysis based on a randomized controlled trial (RCT) (22), while three were retrospective analyses using a National Registry (20, 23) and US Army soldiers’ data (21). The articles involved a diverse group of 439 to 2.5 million adults, comprising both men and women exposed to opioids. The only exception was the RCT, in which only women participated. The authors of the five studies declared no conflicts of interest, and funding sources were mainly from government or not-for-profit organizations.

Table 2 presents the characteristics of the economic evaluations and their outcomes. Two studies used Markov-based models, two used Markov and decision-tree models, and one used a discrete event simulation (DSE). Two studies used specific models to simulate populations, including RESPOND and SOURCE. Regarding the economic evaluation (EE), five studies were cost-utility analyses, two conducted cost-effectiveness analyses, and one performed a budget impact analysis (19). We categorized EE reporting cost-effectiveness results using quality-adjusted life-years (QALYs) as “cost-utility” analyses. Seven studies conducted sensitivity analyses, with six performing deterministic and probabilistic analyses and one conducting probabilistic analysis only. The most common discount rate used was 3%. The time horizon varied across interventions, ranging from 6 months to a lifetime. Regarding the perspective of the EE, eight studies reported from a healthcare perspective, one adopted a societal perspective, and one considered third-party payers.

In terms of outcomes, all studies reported incremental cost-effectiveness ratios (ICER). Five studies reported quality-adjusted life years (QALYs), and one showed cost savings. Additionally, four studies reported prevention of opioid overdose deaths, two showed preventions of non-fatal opioid overdose, and two reported the number of days/months of drug abstinence. Only one study reported utilizing the CHEERS reporting guidelines.

The interventions targeted diverse populations in various settings, including community pharmacies (16, 17), primary care centers (21), reproductive centers (22), detox centers (19), and other healthcare settings (20). One intervention involved mixed settings, such as emergency departments (ED) and community centers (18). Most interventions focused on ambulatory individuals with opioid use disorder (OUD) who were prescribed opioids. One intervention specifically targeted high-risk prescription opioid users (16). Five interventions were categorized as secondary prevention (16, 18–20, 23), two as primary prevention (21, 22), and one mixed prevention (involving aspects of both primary and secondary preventions) (17). To facilitate the description of the included interventions, we grouped them as follows:

We identified three studies examining the effectiveness and economic impact of modifying opioid prescribing behaviors (20, 21, 23). A retrospective cohort used a Markov decision process model to evaluate a policy intervention aimed at reducing the length of an initial opioid prescription to 3 days in a military population. This intervention resulted in $3.1 million in cost savings over 2 years and produced around 4,500 additional opioid-free months (21). One South Korean study found that “The Network System to Prevent Doctor-Shopping for Narcotics” program was considered cost-effective over 30 years, with a cost of US$227 per QALY and a WTP threshold of US$31,362 per QALY from a healthcare system perspective. The program was determined to be 100% cost-effective, even with a WTP threshold of US$900 per QALY (20). Another South Korean study that assessed the same intervention using DES found that this strategy was cost-effective, with an estimated ICUR of $2,227/QALY (23).

These interventions focus on strategies to scale up or expand the distribution of NLX. In this review, two studies evaluated the cost-effectiveness of scaling up NLX distribution (11, 12). The interventions were (a) pharmacy-based distribution of intranasal (IN) NLX (one-time and biannual) in the U.S., and (b) pharmacy-based distribution of both IN and intramuscular (IM) NLX in Canada. The one-time and biannual intranasal NLX distribution strategies were cost-effective, with ICERs of $56,699 per QALY gained and $84,799 per QALY gained, respectively. These values are below the accepted willingness-to-pay (WTP) threshold of $100,000 per QALY gained. Additionally, the one-time distribution prevented 14 additional overdose deaths, while the biannual distribution prevented 107 overdose deaths per 100,000 people.

We identified two interventions focused on expanding MOUD treatment in the U.S. (18, 19) MOUD includes buprenorphine (BUP), methadone, or injectable extended-release naltrexone. The interventions included (a) Initiating MOUD during the detox process and ensuring linkage to outpatient care and (b) MOUD + “treatment add-ons.”

The MOUD initiation in detox centers with perfect linkage turned out to be cost-effective compared with the standard of care, with an ICER of $55,600 per QALY, reducing opioid overdose deaths by 4.5% (19). Additionally, one study explored the use of any MOUD in combination with potential “treatment add-ons,” which refer to additional therapies or supports provided in conjunction with a primary treatment to enhance its effectiveness (18). Increasing the capacity and duration of MOUD, particularly BUP, coupled with the provision of additional therapies such as psychotherapy, contingency management, or telehealth, led to an increase in QALYs gained and effectively prevented 8,570 opioids fatal overdoses and 75,900 non-fatal overdoses per 10,000 person-year.

A cost-effectiveness analysis was conducted using a randomized controlled clinical trial comparing electronic (e) and clinician-delivered SBIRT (Screening, Brief Intervention, and Referral to Treatment). This intervention aims to reduce primary substance use among women receiving treatment in reproductive health centers in New Haven, CT (United States). The results suggest that e-SBIRT could be a cost-effective approach from both healthcare provider and patient perspectives, increasing the days of abstinence during the 6-month follow-up period by 16.66.

Table 3 presents the quality assessment characteristics using the 2022 CHEERS checklist. None of the studies were rated as excellent; four were judged to be of good quality, and four were rated as fair quality.

This systematic review distinguishes itself from others (8–11) by focusing exclusively on strategies to prevent PDM, particularly on articles published within the last 5 years. This contemporary focus not only highlights the review’s relevance in addressing the cost-effectiveness of the latest interventions on PDM but also provides valuable insights into the ongoing challenges and gaps. Likewise, for reporting results, we adhered to the recommendations described in the ISPOR Criteria for Cost (−Effectiveness) Review Outcomes (CiCERO) Checklist, ensuring a comprehensive and transparent presentation of the economic aspects of our review. As part of our review, we assessed the study quality using the 2022 checklist “Consolidated Health Economic Evaluation Reporting Standards (CHEERS)” developed by ISPOR.

Our review has some limitations. We restricted our search to scientific publications and did not include gray literature or reports from health technology agencies. As a result, our findings may not cover all available evidence, and we cannot definitively claim that these are the only relevant results. In addition, we cannot conclude that the results of the included interventions apply solely to patients who were prescribed opioids, as some of the studies, especially those related to the use of MOUD, included populations exposed to illicit drugs as well. We faced difficulties classifying interventions into three levels of prevention due to incomplete descriptions of the population characteristics. It is crucial to accurately understand the history and the risk of opioid use disorder among participants receiving drug prescriptions or within simulated populations in order to categorize an intervention as secondary or tertiary prevention properly. Without this context, it is challenging to determine the precise level of prevention.

Finally, we excluded studies published in languages other than English, which may have limited the scope and comprehensiveness of our search. By not including research published in other languages, we may have overlooked important studies that could have provided valuable insights, particularly from regions where English is not the primary language. This exclusion may have affected the generalizability of our findings.

Despite their limitations, the findings of this review hold substantial implications for healthcare policy, practice, and future research directions. There is a critical need for research into the economic evaluations of PDMPs, the impact of clinician adherence to CDC guidelines on PDM prevention, especially in chronic conditions, and the cost-effectiveness of strategies to mitigate or prevent the misuse of prescription benzodiazepines and stimulants. Moreover, advancing economic models, conducting rigorous cost–benefit analyses, and evaluating real-world implementation strategies are critical to ensure evidence effectively informs policy decisions, not only from a healthcare perspective but also from societal and payer perspectives.