Over the past decade, the rates of opioid-related harms have been increasing in North America [1-3]. Opioid-related harms may include opioid use disorder, adverse drug reactions and events, opioid toxicities, and death [4-7]. Since the early 2010s, there has been growing awareness of these potential opioid-related harms across North America [8,9]. Although most recent opioid-related harms are due to the unregulated drug supply (ie, fentanyl) [10-12], a long history of opioid prescribing practices for acute and chronic pain has contributed to harms [8]. In response to the increasing numbers of opioid-related harms in the United States and Canada, professional standards for opioid prescriptions were revised in 2016 and 2017, respectively, to include recommendations for more conservative opioid prescription practices [13-15].

e-Prescribing is designed to help facilitate the safe and appropriate prescribing of medications. e-Prescribing (in the Canadian context) is the secure electronic creation and transmission of a prescription between an authorized prescriber and a patient’s pharmacy of choice [16]. It uses clinical point-of-service solutions to integrate clinical workflow and software. e-Prescribing has shown some promising benefits at the patient, clinician, and health system levels. At the patient level, e-prescribing has improved patient safety [17-22] and patient experiences with accessing medications [23-25]; for example, the implementation of e-prescribing resulted in decreased rates of adverse drug events and prescribing errors [17-22]. It has also been shown to improve patient experience through easier access to medications and reduced wait times for dispensing [19,23,24]. At the clinician and health system levels, e-prescribing can improve workflow efficiency (eg, facilitating communication between prescribers and dispensers [26] and improve the rates of medication adherence, measured by prescriptions being filled [27-31]), resulting in both reduced health care costs and improved health outcomes [32,33]. Although these benefits of e-prescribing have been well described, there remains a gap in understanding the extent to which e-prescribing can influence safe and appropriate opioid use as well as clinically relevant experiences and outcomes.

To address this gap, a rapid scoping review was undertaken to answer the following question: what are the direct impacts of e-prescribing for opioids on experiences and outcomes? The specific objectives of this review were to (1) explore how e-prescribing has been used clinically with opioids; (2) examine the effects of e-prescribing of opioids on clinical outcomes, patient or clinician experience, service delivery, and policy; and (3) identify any gaps in the literature to inform future studies and recommendations.

A rapid scoping review was conducted following the guidance of the JBI 2020 scoping review methodology [34] and the World Health Organization guide for rapid reviews [35]. Streamlined methods to conduct the rapid review followed the steps outlined by the Cochrane Rapid Reviews Methods Group in 2020 [36]. Reporting aligns with the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) statement [37]. The PRISMA-ScR checklist can be found in Multimedia Appendix 1 [37]. The protocol for this review was registered with OSF Registries [38].

Eligibility criteria for the review evolved during the screening stages. During the title and abstract screening, inclusion criteria were as follows: (1) e-prescribing programs that were targeted to opioid use or misuse, including those that were accompanied or complemented by clinically focused initiatives, and (2) a primary research study of experimental (eg, randomized controlled trials), quasi-experimental (eg, nonrandomized controlled trials, controlled before-and-after studies, or interrupted time series), observational (eg, cohort studies, case-control studies, or cross-sectional studies), qualitative, or mixed method design. At the full-text screening phase, an additional inclusion criterion was added: an ambulatory component of e-prescribing (eg, e-prescribing of opioids occurred at discharge from acute care, in the emergency department, or in the community). This criterion was not included until the full-text stage to ensure that all relevant articles were included for review because abstracts were not likely to clearly specify the involvement of an ambulatory component. The exclusion criteria for all stages of peer-reviewed article screening included (1) prescribing that occurred within 1 hospital system (eg, within an acute care ward), (2) articles that did not look at the impact of e-prescribing on opioid use, (3) digital solutions for prescribing that did not include e-prescribing (eg, digital fax), (4) not a primary research study (eg, commentaries and opinion pieces), and (5) conference materials (eg, abstracts). Gray literature was included if the aforementioned criteria were met; however, articles were not required to be a research study.

A literature search was conducted by an expert librarian (Leah M Boulos) on articles published from database inception until November 16, 2022. Three databases were electronically searched: MEDLINE (Ovid), Embase (Ovid), and Scopus (Elsevier). Gray literature was searched using a string of key terms in Google and ProQuest Dissertations & Theses Global. No filters or language limitations were applied.

The search strategies were developed based on 2 key concepts (e-prescribing and opioids) in consultation with the expert librarian who ran the search (Leah M Boulos). Previously published systematic reviews on opioids were also searched to identify relevant opioid-related terms [39-41]. Search strategies for the databases and gray literature can be found in Multimedia Appendix 2. A second librarian reviewed the search strategy using the Peer Review of Electronic Search Strategies (PRESS) checklist [42].

Before deduplication, records from MEDLINE and Embase were removed from the Scopus database search using the AND NOT function to ensure that all relevant articles could be exported to EndNote (Clarivate); Scopus has a 2000-record export limit. Deduplication of the resulting list of articles from the 3 databases was then conducted in EndNote using the method developed by Bramer et al [43]. The literature review software, Covidence (Veritas Health Innovation Ltd), was used to streamline the article screening process. At the title and abstract screening phase, a pilot test using 20 articles was conducted by the reviewers (SRC, SJTG, JR, Shreya Mahajan, Shanzeh Chaudhry, and Alyssa Yang). After the pilot test, the team met to review the inclusion criteria, which were updated to ensure clarity. During this phase, eligibility criteria were kept broad to ensure that as many relevant articles were included as possible (eg, if there was uncertainty about the ambulatory component, articles were moved to full-text review). All articles were screened independently by 2 reviewers (SRC and JR), with any conflicts resolved through team discussion.

Once the title and abstract screening was completed, 10 full-text articles were randomly selected for pilot testing to ensure consistent application of the eligibility criteria across all reviewers. At this phase, articles that did not include an ambulatory component (ie, did not involve opioids prescribed at acute care discharge, in the emergency department, or in the community) were excluded. All full-text articles were independently screened by 2 of the 5 reviewers (SRC, JR, Shreya Mahajan, Shanzeh Chaudhry, Alyssa Yang) using the updated criteria, which included the ambulatory component.

Using Google, gray literature was manually searched by 1 reviewer (SRC). After reviewing the last relevant citation, an additional 20 citations were reviewed to ensure that all relevant materials were included. For dissertations and theses, this same process was completed by 2 reviewers (SRC and JR) using ProQuest Dissertations & Theses Global.

Data extraction, using the Covidence Data Extraction 2.0 form, was conducted once the full-text screening was completed. Key data that were collected from the articles included study characteristics, population characteristics adapted from the Cochrane PROGRESS-Plus equity variable recommendations [44] (sample size, age, sex, gender, ethnicity or race, religion, income, education, geographic location, and social capital), the description of e-prescribing (design, prescriber context, intended recipients, indication for opioids, and accompanied or not accompanied by clinically focused initiatives), study outcomes, and findings (eg, the descriptions of data-driven activities or analysis for managing the prescribing of opioids or informing better policy and interventions, opioid dependency, opioid-related death, health care use owing to opioids, economic costs owing to opioids, fraud, and the transparency of prescription history). A pilot test was conducted by 4 reviewers (JR, Shreya Mahajan, Shanzeh Chaudhry, and Alyssa Yang) with the extraction of 1 assigned article per person. Each of the pilot articles was spot-checked by an independent trained reviewer (SRC) to ensure consistency in extraction across reviewers. The remaining data extraction was conducted by the 4 reviewers (JR, Shreya Mahajan, Shanzeh Chaudhry, and Alyssa Yang), with quality checks conducted by the independent reviewer (SRC). A quality assessment of the articles was not conducted as per scoping review standards [34].

The findings from the included articles were synthesized using descriptive approaches. Descriptive summaries of the study characteristics, population characteristics, study outcomes, and findings were conducted. Summaries of the findings were developed by collating study findings that reported on similar topics (eg, e-prescribing setting and the rates of prescribing opioids). Once the information was organized, a section header was developed based on the subject matter of each section. This organization process was carried out by 2 members of the authorship team (SRC and JR) in conjunction with members of the senior research team (SJTG, MT, LMM, and TG).

The literature searches yielded 1183 articles (refer to Figure 1 for the PRISMA [Preferred Reporting Items for Systematic Reviews and Meta-Analyses] diagram). After the removal of duplicates from the 1183 articles, 939 (79.4%) were included in the title and abstract review. After this initial screening phase, 161 (17.1%) of the 939 reports were sought for retrieval; however, of the 161 articles, the full text of 1 (0.6%) article could not be retrieved, leaving 160 (99.4%) full-text articles assessed for eligibility. With respect to gray literature, 16 articles were identified: 12 (75%) dissertations or theses and 4 (25%) potentially relevant reports found via Google. After the screening of the total 176 articles, 32 (18.2%) full-text articles [28,45-75] and 2 (1.1%) gray literature documents (thesis: n=1, 50%; report: n=1, 50%) [76,77] met the inclusion criteria and were included in the rapid review. The characteristics of the identified studies (32 full-text studies and 1 thesis) are described in the following subsections, followed by a description of the gray literature report.

Data were obtained through a variety of collection methods, with the most common being electronic medical records (13/33, 39%) [45-49,54-58,60,66,77] and hospital or health care setting databases (12/33, 36%) [52,53,59,61,62,64,67-69,72-74]. Other data were obtained through a variety of methods and approaches, such as surveys (5/33, 15%) [54,63,66,70,71], structured interviews (2/33, 6%) [59,66], opioid prescribing rate maps (1/33, 3%) [51], iScribe (an e-prescribing system used for outpatient settings; 1/33, 3%) [28], data from the US Drug Enforcement Administration’s Automation of Reports and Consolidated Orders System (1/33, 3%) [51], a computer-generated data set (1/33, 3%) [65], a large pharmacy benefits management company (1/33, 3%) [28], and treatment orders (1/33, 3%) [50].

With respect to the populations being studied (Table 2), most were clinical populations (24/33, 73%) [45-60,62,64,67,69,72,73,75,77], the general population (7/33, 21%) [28,61,63,65,66,68,74], and clinical prescribers (2/33, 6%) [70,71]. e-Prescribing settings varied among the included studies, with ambulatory settings (eg, emergency department and outpatient department) being the most common (16/33, 48%) [28,46-49,53,54,58,61,65-68,71,74,75]. Other settings included acute care discharge (12/33, 36%) [45,52,55-57,59,60,64,69,72,73,77], hospital pharmacy (2/33, 6%) [50,62], community pharmacy (1/33, 3%) [71], and primary care (1/33, 3%) [63]. Of the 33 studies, 1 (3%) did not report the setting [51].

Among the 13 studies that specified participant age, 7 (54%) studied adults (aged ≥18 y) [57,59,60,62,64-66], 3 (23%) studied a population comprising older adults (aged ≥65 y) [53,55,58], and 3 (23%) included a pediatric population (aged <18 y) [52,56,61]. Of the 33 articles, 16 (48%) [45,46,48,49,53,56,58-60,62-64,67,69,71,72] reported the sex of the participants (most were male). A few studies reported ethnicity or race (10/33, 30%) [46,48,49,53,60,63,64,67,71,77], comorbidities (5/33, 15%) [45,59,63,66,67], gender (3/33, 9%) [28,66,77], marital status (1/33, 3%) [64], employment status (1/33, 3%) [64], or geographic location (1/33, 3%) [63]. Income, education, the place of residence, social capital, and religion were not reported.

The opioids that were studied included oxycodone (14/33, 42%) [47,48,52,53,56-58,60,61,64-66,69,77], codeine (8/33, 24%) [45,48,52,57,60,61,69,77], morphine (8/33, 24%; immediate release: n=2, 25% [47,52]; controlled release: n=2, 25% [47,52]; intravenous: n=2, 25% [53,58]; and unknown: n=4, 50% [48,55,69,77]), hydromorphone (7/33, 21%) [47,48,52,57,58,64,77], tramadol (6/33, 18%) [47,48,57,69,72,77], hydrocodone (6/33, 18%) [48,52,60,65,69,77], fentanyl (5/33, 15%) [47,58,60,62,77], meperidine (3/33, 9%) [60,69,77], oxycontin (2/33, 6%) [52,54], oxymorphone (2/33, 6%) [69,77], opioid in combination with acetaminophen (hydrocodone: 3/33, 9% [56,57,60]; codeine: 2/33, 6% [57,60]; and oxycodone: 2/33, 6% [56,60]), butorphanol (1/33, 3%) [77], dihydrocodeine (1/33, 3%) [77], and tapentadol (1/33, 3%) [69].

Specific components of e-prescribing were reported to influence opioid prescribing, including alerts and default order sets.

Of the 33 studies included in this review, 2 (6%) described clinicians’ experiences and perceptions with using e-prescribing for opioids [70,71]. Thomas et al [71] explored barriers associated with the adoption and use of EPCS using a quantitative survey (n=246; 64% response rate). When asked about their expectations of e-prescribing systems for opioids, prescribers expected this technology to improve patient management and practice efficacy [71]. However, prescribers were hesitant to use new prescribing technologies owing to their reservations with patient confidentiality or the learning curve to use e-prescribing systems [71]. In the second study by Thomas et al [70], a survey was conducted to understand the experiences of prescribers (n=102; 68% response rate) after EPCS implementation. For prescribers currently using an e-prescribing system, they indicated that it was easy to use, improved the accuracy of prescriptions, improved workflow, improved coordination, and limited the number of calls from pharmacists [70]. With respect to satisfaction with the system, age, comfort with using a computer, the number of patients per week, and the belief that the system improved patient management were associated with increased odds of being satisfied with the system [70]. Both studies described technical issues such as computer crashes, lag time between transmitting and receiving prescriptions, and pharmacist follow-up to confirm e-prescription details as barriers to using the e-prescribing system for opioids [70,71]. Two additional barriers to the implementation of EPCS were the need to keep a security token in their possession to access the system [71] and the lack of community pharmacies using the e-prescribing system [70]. No studies explored the experiences and perspectives of patients or caregivers.

e-Prescribing mandates were associated with the reduction of both opioid prescriptions [49,67] and opioid dose [67]. The mandates were implemented in 2 states in the United States (New York and Massachusetts) [49,67].

The single report identified in the gray literature search suggested that mandatory national use of EPCS could save the US government a projected US $53 billion [76]. The cost savings were based on several factors, including reduced costs owing to opioid-related fatalities (between US $18 billion and US $37 billion saved); decreased health care costs, including treatment costs; increase in workplace productivity; reduced criminal justice costs (between US $7 billion and US $14 billion saved); and savings from improved efficiencies in physician offices and pharmacies (eg, reduced calls between prescribers and pharmacists regarding prescription clarifications and decreased wait times for patients to fill prescriptions; US $1.6 billion saved) [76].

This rapid scoping review examined how e-prescribing has been used clinically for opioids; investigated the impact on experiences, and outcomes; and identified several gaps in the literature. Overall, we identified a limited number of articles that met our inclusion criteria (n=34). Despite a comprehensive search, we identified minimal research examining e-prescribing for opioids and related outcomes. Although the results showed promising findings, such as a reduction in prescription errors and identifying drug-drug interactions, there remain important clinical, implementation, effectiveness, and policy-relevant areas for further exploration.

Most studies examined e-prescribing being initiated within hospital-based care or an affiliated ambulatory clinic. Thus, most of the evidence found in this review reflects hospital settings and closed health systems. The main data systems used within the hospital systems were the CPOE system and EPCS. Only 1 (3%) of the 34 articles focused on e-prescribing in primary care, using the CPOE system [63]. In addition, there was minimal reporting of prescriber and pharmacist characteristics, clinical characteristics, or sociodemographic information. Furthermore, we identified a large variation across the included studies examining the effects of e-prescribing on experiences, and outcomes. Most of the outcomes were focused on prescription-level metrics such as prescription rates, prescription errors, and discontinuation rates.

Despite the variation, there seem to be promising findings with respect to e-prescribing; for example, 1 (3%) of the 34 studies showed a reduction in prescribing errors when compared with handwritten notes (eg, 20.6% handwritten errors vs 1.2% e-prescription errors) [59]. A few studies (2/34, 6%) also highlighted promising effects of alerts and order sets on reducing errors; 2 (6%) of the 34 studies demonstrated the usefulness of e-prescribing mandates in reducing opioid prescriptions [49,67] and reducing dose [67]. Given the increasing rates of opioid-related harms in North America [1], these findings suggest that e-prescribing may be a promising approach to address prescribing errors. However, it is important to understand the nature and related implications of reducing the number, dose, and rapid discontinuations because there may be unintended risks of reducing access to opioids or reducing doses too quickly [78-80].

In the single gray literature report identified, the mandatory national use of EPCS has been projected to have a potential cost savings of approximately US $53 billion annually for the US government [76]. Despite uncertainty around cost savings, there is potential for these cost savings to be reallocated to fund educational programs for prescribers, patients, and the public. However, it is important to note that the unregulated opioid drug supply is the main cause of opioid-related deaths in Ontario, and the generalizability of this review to the Canadian context should be made with caution [12].

One of the challenges in reviewing the literature is the substantial shift in practice guidelines for opioid therapy management that occurred in North America after 2016 [14,15]. As such, studies published before this date examining e-prescribing and opioid use may not reflect current practices or needs. This review identified several gaps, particularly related to implementation and effectiveness considerations. Future research is warranted to expand the current knowledge of e-prescribing systems and opioid-related outcomes. First, e-prescribing needs to be assessed across broader health systems and larger populations, such as in community and primary care. Only a single study was found that assessed e-prescribing in primary care [63]. This study included data collected before 2016, when significant practice guideline changes were released that have an impact on opioid prescribing and patterns, suggesting the data only available from primary care likely do not reflect current practice or needs. Second, the perspectives, experiences, and health care outcomes from a wide variety of stakeholders (such as prescribers, clinicians, pharmacists, patients, and pharmacy managers) should be explored and examined through mixed methods and qualitative studies; for example, qualitative studies with community stakeholders would provide insight into the fear regarding the e-prescribing of opioids that has previously been reported to affect the prescribing rates of primary care physicians [81-84]. Third, the development of a common set of quality indicators to guide the reporting of outcomes would likely be useful to ensure the consistent implementation and evaluation of e-prescribing across varying studies. Finally, more studies are needed to understand implementation considerations such as barriers and facilitators for e-prescribing to inform adoption and larger scalability. There are well-established factors that influence the implementation of interventions and their effectiveness; for example, the Consolidated Framework for Implementation Research consists of 5 key domains that are known to influence implementation [85]. To inform the adoption and uptake efforts of e-prescribing technology, it will be important for future work to understand for whom e-prescribing might be working, how, and in what circumstances, which may be completed through a realist evaluation [86]. This review identified important questions that remain, such as the following: (1) Are there certain oppressed groups where this technology might be particularly useful to support safe and effective opioid therapy management? (2) Are there certain prescribers and pharmacists who might benefit more from this technology and in what clinical settings? (3) Are there certain aspects of the e-prescribing system that are more beneficial or harmful (eg, questions related to the alerts, order sets, and interaction features)? Of note, there are known risks to rapid dose reductions with opioids [78], and it would be important to explore further an understanding of potential harms. A key aspect for consideration is how e-prescribing might be implemented for new prescriptions to prevent short- and long-term risks among persons compared with how it might be implemented for repeat prescriptions among those experiencing chronic pain. These implementation factors should be considered in future work examining e-prescribing.

Overall, there was a lack of consistency in the types of outcomes reported, and it is unclear whether the outcomes reported align with established quality indicators (eg, a consideration of dose within the clinical context of acute or chronic care). Several of the outcomes may be problematic, such as nonadherence and discontinuation, because they may not accurately reflect an improvement in outcomes; for example, with nonadherence, it is important to consider differences in “taking medication when needed” versus “taking medication on a prescribed schedule.” With respect to discontinuation, the timing needs to be considered (eg, discontinuing the same day vs discontinuing within the prescription period). Same-day discontinuations are likely due to errors by the prescriber, as seen in the study conducted by Hickman et al [54]. Tapering guidelines for chronic pain suggest that the discontinuation of opioids may lead to the risk of inadvertent or unintentional overdose risk, if not carried out properly [87,88]. It is suggested that patients follow a gradual morphine equivalent dose decrease of 5% to 10 % every 2 to 4 weeks with frequent follow-up. However, if the prescription is for acute pain, tapering is not necessarily needed [87,88]. Finally, there was an absence of studies exploring the perceptions of e-prescribing for opioids from different stakeholder groups (eg, clinicians, prescribers, and patients) from a qualitative perspective, which would also inform meaningful outcomes and potential indicators of quality e-prescribing.

The limitations of this study are consistent with those common to rapid reviews. It is possible that articles were missed. Despite the time constraint, a rigorous selection process was undertaken with double screening present at each stage of the process, and grey literature was searched. Of note, 15 (44%) of the 34 articles were published in 2016 or earlier, which would not reflect the dramatic shifts that occurred in opioid therapy management in the last several years. In addition, the quality of the studies was not assessed, which is typical of a scoping review, and as such, this review does not integrate the strength of the evidence [35].

Although relatively few studies were identified, this scoping review highlights preliminarily promising results with e-prescribing and opioid therapy management. e-Prescribing, including key features such as alerts and dose order sets, may contribute to a reduction in prescribing errors. A key aspect for consideration is how e-prescribing might be used and the differences in outcomes by using this tool based on medication prescription being newly initiated or chronic. Among new prescriptions, there may be potential to decrease initiation, quantities, and doses as per best practice guidelines to minimize short- and long-term risks. Conversely, there may be important and different considerations with e-prescribing for people who are taking opioids on a chronic basis to minimize disruptions with access and sudden dose changes. These important nuances were missed from the research reviewed and highlight gaps in the literature. It is important that future work explores the experience of prescribers, pharmacists, and patients using e-prescribing for opioid therapy management, with an emphasis on prescribers in the community and primary care. Integrating the thoughts, perceptions, and beliefs of these parties into the literature is important because they are directly affected by technology use in health care. Developing a common set of quality indicators for e-prescribing with opioids will help inform future research and build a stronger evidence base. Furthermore, understanding implementation considerations will be required as the technology is adopted and integrated into clinical practice and health systems.