Introduction
On March 26, 2020, the provincial government in British Columbia (BC), Canada, released what was termed risk mitigation guidance [
1] to support people who use substances, an initial attempt at prescribed safer supply (PSS). The guidance document was introduced in order to promote self-isolation and thus prevent the spread of COVID-19 among people who use substances, as well as offering a medicalized harm reduction response to the increasingly toxic unregulated drug supply [
2]. Risk mitigation guidance permitted prescribers to prescribe opioids, stimulants, and benzodiazepines to individuals who were at high risk of or had a confirmed or suspected COVID-19 infection and/or high likelihood of substance-related and systemic harms [
2]. While introduced as an interim emergency measure, it subsequently became a provincial policy in July 2021, for prescribers that were not a part of specific health authorities in the province and prescribed other safer supply options [
3]. An update was released in January 2022 [
4] reporting preliminary evidence for patient outcomes and diversion, highlighting that PSS was not a direct contributor to the rising rate of illicit drug toxicity deaths.
This guidance was disseminated through voluntary online educational materials and didactic lectures, thus constituting a “passive” dissemination strategy. Initially, the guidance was published as an online educational material when it was released, with voluntary webinars recorded and made available online on April 9, 2020, as well as the slide presentation used [
5‐
7]. The guidance was subsequently updated in 2022 and disseminated similarly. Such strategies have been found to be insufficient for driving system-level change [
8‐
10]. A 2008 review of systematic reviews suggested passive dissemination is not sufficient, and that at least one, but preferably multiple targeted implementation strategies, such as educational outreach, interactive educational interventions, or practical recommendations, should be employed to create awareness of new guidelines or policies [
11]. A more recent review classified distributing educational material passively as having mixed effectiveness for increasing health professionals’ knowledge, attitudes, and willingness to follow guidelines, but being generally ineffective (less than one-third of studies demonstrated a positive effect), for process-related outcomes which included prescription of medications [
12]. A mixed-methods study from Quebec found passive dissemination of practice recommendations was successful when delivered by an opinion leader or self-identified champion, if recipients had a high level of expertise, and when there were sufficient professional resources, such as retaining nurses and other clinical staff [
13].
Implementation science focuses on active and planned efforts to mainstream an innovation [
14‐
16]. Diffusion of innovation theory has been applied within implementation science to focus on the natural and passive spread of innovation through adopters and design strategies to engage target groups to adopt specific innovations [
14,
16]. Healthcare provider networks play an important role in the process of implementing innovations in clinical practice [
17]. Previous work has demonstrated patterns of influence among physicians in primary care practices [
18]. Social network analysis (SNA) is increasingly used to provide evidence for interpersonal influences theorized in diffusion theory [
19], by considering the connections among actors within a network to understand their patterns, influence, and relationships [
20]. SNA can be used to pilot test the feasibility of an implementation [
21]; design communication strategies [
22] tailoring dissemination to opinion leaders, network isolates, and other prescribers with few clinical contacts; and evaluate implementation efforts [
23]. Health administrative data can be used with SNA to model physicians’ social networks, creating opportunities to evaluate and inform the implementation and diffusion of new policies and interventions on a population level [
24,
25]. Previous research on diffusion of new policies has shown the influence of networks on tobacco regulations among countries [
26]. In cancer treatment, physicians connected to previous prescribers of bevacizumab had higher odds (adjusted odds ratio 1.64; 95% confidence interval, 1.20–2.25) of prescribing the following year [
27], with other medications having similar effects [
28]. These studies are based on well-established, and less contentious, treatment effectiveness than PSS. Therefore, PSS presents a unique opportunity to understand the spread of a new emergency policy that faced pushback from addiction medicine prescribers, and other specialties, thus challenging the influence exerted by peers.
Treatment specific to opioid use disorder such as methadone and buprenorphine-naloxone, the primary modalities of opioid agonist treatment (OAT), is available in both specialized treatment centers and office-based settings [
29]. Relaxed restrictions on OAT prescribing have allowed family physicians, nurse practitioners, and registered nurses to start opioid agonist treatment for opioid use disorders without any required waivers [
30,
31]. Given that the risk mitigation guidance was applicable to all prescribers caring for people who use substances, a large diversity of prescribers, operating in different settings with varying levels of experience, caseloads of clients who use substances connections to other prescribers caring for people who use substances were eligible to participate. Analyzing the diffusion of PSS uptake among providers can provide vital information on how to adjust strategies to ensure intervention reach and adoption are optimized. We aimed to characterize the diffusion of PSS adoption and determine the extent to which PSS uptake operated through established networks of prescribers.
Discussion
We evaluated the diffusion and adoption of a prescribed safer supply program implemented in BC and found that prescribers with larger SUD client load and higher proportions of peers that previously provided dispensations under PSS associated with increased odds of uptake of the PSS program. The importance of peer effects in the PSS adoption decision was reinforced by our findings on PSS discontinuation, as those with more peers who were still prescribing PSS were much less likely to discontinue dispensations. While risk or case of COVID-19 was eligibility criteria for PSS prescription, indications of diagnosis within clinicians’ case mix were only modestly positively associated with PSS adoption, thus further reinforcing network influences as the dominant. Overall adoption of the program was limited to 9.1% of all indicated prescribers, with 57% of prescribers who initiated PSS discontinuing dispensations later in the study period, and 63% of all early adopters forgoing PSS. Passive dissemination strategies may have been insufficient for subsequent adopters, those with lower client loads, and less peer prescribers knowledge updates and client’s needs weren’t met by the guideline or by the timeliness of available resources [
48,
49].
Of critical importance in this application is the fact that the risk mitigation guidance was not explicitly recommending evidence-based practices [
2]. This guidance was implemented on an emergency basis to protect against an anticipated, and ultimately observed, escalation in toxic drug-related deaths due to potential disruptions in illicit drug supply channels, guidance for self-isolation which could lead to a higher incidence of using alone, closures of supervised consumption facilities, and restricted access to other forms of treatment and care [
50]. These findings nevertheless align with studies from other disease areas which demonstrate that the prescribing behaviors of a physician are impacted by the prescribing behaviors of their peers [
27,
28,
51]. Aside from barriers in the outer context, adoption can still be hindered due to the inner context [
52], systematic reviews have identified prescriber knowledge, attitudes and beliefs can hinder the success of implementation strategies for clinical practice guidelines [
53], and attitudes and beliefs were no doubt influential in this particular application. The regulatory institution for physicians in the province expressed concerns regarding the limited empirical evidence base; the possibility of diversion, limited training, and expertise in providing prescribed safer supply within their scope of practice; and preference for team-based practices, which may deter adoption of new guidelines among prescribers in preference of well-established treatments [
54‐
56]. As the evidence base for PSS emerges, future implementation strategies to improve PSS uptake will require engagement with both regulatory institutions and prescribers, acknowledging and addressing the concerns of both parties [
57].
Our findings nevertheless highlight critical considerations for future implementation strategies on new policies for the clinical management of substance use disorders, which would benefit from leveraging or targeting networks of prescribers. First, our findings demonstrated prescribers who had a greater number of SUD clients were more likely to initiate PSS, suggesting these clinicians may have had greater awareness of announcements related to innovations in SUD care. Prescribers with smaller SUD client loads therefore may be harder to reach with passive dissemination strategies such as online educational materials and didactic lectures to learn about new clinical guidance and may require more active dissemination strategies such as academic detailing [
24,
25] or interactive educational meetings [
8].
Our findings provide further evidence that diffusion of clinical practice requires social reinforcement and draws attention to the need to identify which prescribers to target in dissemination strategies [
27]. Recognizing this impact of diffusion among peers, and those on the outer edges of the network who are unlikely to be impacted by peers, emphasizes the need to consider differential strategies to engaged clinicians in larger and less-connected practice settings, respectively. Though some evidence is available on the effectiveness of specific singular active strategies [
53,
58,
59], multifaceted implementation strategies, which include a combination of approaches such as educational changes, reminder systems, and organizational shifts for multidisciplinary collaboration, have demonstrated effectiveness [
53,
59] in successfully disseminating guideline and policy changes in other disease areas. Future policy changes and clinical guideline updates should employ several different strategies to engage prescribers. Such strategies are particularly urgent in the present context given the pervasiveness of the toxic drug supply public health emergency, first issued in 2016 [
60], which has increased in intensity due in large part to the changing composition and increasing potency of the illicit drug supply [
61]. These changes in the underlying disease will require ongoing updates and revisions to clinical practice, which may otherwise suffer from waning attention with more frequent communications.
This study was not without limitations. Though we used population-based administrative data with true population coverage within the province of British Columbia, regardless of prescription drug insurance status, there was potential for some misclassification, particularly in identifying location as the location of prescriber and location of their clients may not coincide and clients may live in other health authorities. As with any observational study, our inferences may be subject to unmeasured confounding. In particular, we were unable to observe clinicians’ motivations for uptake and discontinuation. However, we hypothesize that motivations would be mediators on the causal path between peer influence and prescribing medications as individuals’ motivations will be influenced by their peers’ decisions, motivations, and opinions and therefore believe our interpretations were unaffected by these omissions. We otherwise do not observe a range of individual-specific characteristics of clinicians, including their year of graduation. Our constructed measures of the number of years of experience treating SUD clients likely adjusted for much of these potential confounding effects. Furthermore, though we had the number of positive COVID-19 cases within a prescriber’s caseload, we did not have information on who would have been isolating due to a close contact being positive, thus potentially diminishing the influence of COVID-19 on PSS uptake. Finally, while we have captured PSS implementation at the population level in the province of British Columbia, Canada, BC remains the only setting in Canada or elsewhere to have implemented such a program; if other settings choose to implement a prescriber-based safer supply program, both structural and epidemiological conditions may influence PSS uptake. The transportability of these results to other settings should be considered carefully.
Acknowledgements
Our research took place on the traditional, unceded, and continually occupied lands of the over 200 First Nations in what is colonially known as British Columbia. Data for this publication was provided by Vital Statistics BC, the BC Ministry of Health (PharmaNet, Medical Services Plan, the National Ambulatory Care Reporting System and the BC Discharge Abstract Database), Perinatal Services BC, the Ministry of Public Safety and the Solicitor General, Ministry of Social Development and Poverty Reduction, and BC Coroner’s Service. All inferences, opinions, and conclusions drawn are those of the authors and do not reflect the opinions or policies of the data stewards.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.