Findings demonstrating differences between rural and urban EDs are presented below by 4 of the CFIR domains, with the external context taxonomy substituted for the CFIR’s outer setting domain. Specific constructs are italicized within the paragraphs. Table 2 displays the themes categorized by domain and construct, which are also defined.

Table 2 Definitions and source for implementation domain and construct names identified in the analysis Intervention characteristics

The perceived evidence supporting peer services and the source of pre-existing ED-based peer programs was an important facilitator of adoption discussed at T1 interviews. Vendors described awareness of research on peer supports as influencing their decisions to apply for RCPSI funding. For instance, a representative from RV1 spoke about research they looked to when considering applying:

I did a little research on it [peer services], and it made a lot of sense. It works for AA [Alcoholics Anonymous] and NA [Narcotics Anonymous] to get people who’ve been through it [12-step programming] to be your coach and support [i.e., sponsor]. But, then this [peer services in the ED] takes it to another level because they’re [the peers are] more involved with connecting to the [treatment] resources around you [in the community]. (T1)

While the above statement applies logic from other types of peer supports to the decision to implement ED-based peers, other vendors directly discussed knowledge of early successes of ED-based peer support programs they had learned about as influencing their decisions. While urban vendors discussed similar reasons for applying for the RCPSI funding at T1, a key difference was that some urban vendors (e.g., UV3, UV4, and UV6) physically visited sites that had implemented ED-based peers to learn directly from them. While all vendors were aware of early successes of other ED-based peer programs, rural vendors were the only ones that displayed skepticism about the ability to translate these services directly to their settings. As noted by one rural vendor, “[a] lot of this research [on ED-based peers] is only hypothetically applicable to rural [areas]…” (RV2; T1).

Across vendors, the addition of peer services was viewed as having a relative advantage compared to usual care provided to OUD patients presenting in their EDs. This is because most hospitals did very little for these patients after ED discharge: “Honestly…[prior to applying for RCPSI funding] we do not do anything post ED [discharge]” (RV4; T1). As such, vendors viewed peer supports as a valuable resource for patients that would provide needed relief to ED staff: “at this point [T1], the physicians and nurses do what they can. They just see it [treating OUD patients] as too much for them, so they are eager to get additional resources [from the peers]” (UV2). However, as the implementation progressed, urban vendors continued to discuss relative advantage as it related to the ED environment, while rural vendor discussions shifted to described how the advantage of peer services extended beyond the ED’s boundaries. For instance, data from RV2 demonstrated they were using the peer to fill a need for outpatient services: “We had a great need for peer recovery coaching in the outpatient side. So, [the peer has] actually been helping with our outpatient addictions program too, and then just coming over here [to the ED] when called” (T3).

Interviews also demonstrate the high degree of adaptability of peer services. While all vendors discussed ways in which the peer position was different from those on which the RCPSI was partially based and how the peer role evolved over time, rural vendors’ discussions demonstrated they made more considerable modifications to the scope of peer services in an effort to address the fact that peers had very little work due to low numbers of OUD patients being admitted to the ED. The previous quote from RV2 demonstrates how their peer’s scope was extended to include outpatient services, and an earlier interview also showed how the peer was working with other departments: “[our peer] also does [work in] the OB [obstetrics department], and she also is involved with our court program [for justice-involved patients]” (T2). In another example, RV3 contracted with a local off-site mental health provider that designated a peer to respond to the ED when needed since it did not have enough patients presenting with OUD to justify employing a full-time ED-based peer.

Rural vendors found the costs related to peer services to be a greater barrier at all interview points. Early in the implementation, RV1 and RV4 both expressed concerns for how they would support peers because they did not have enough target patient volume to cover the salaries. However, this was demonstrated to be resolved at T2 when the funding model was changed from service reimbursement to one that directly covered peer salaries. At T1, RV2 had concerns about the costs getting “completely out of control” because their hospital had only one part-time, on-call psychiatrist available who warned they were overwhelmed with their peer’s referrals. It was noted in later interviews that the administration hired a full-time psychiatrist to address this issue, which was an unexpected cost at the time they applied for RCPSI funding. The cost of technology also later derailed RV2’s original plans to implement telehealth-based peer services as an addition to their program, as they had no monies left to do this after using RCPSI funds to update their electronic health record to document peer service contacts. In contrast, urban vendors directly stated they had no or very little concerns regarding implementation costs, and they did not discuss any unexpected costs to have occurred over the course of implementation.

External context

Target population needs greatly impacted peer service feasibility at some rural sites. Specifically, OUD patients’ needs that were not anticipated at baseline resulted in missed engagement opportunities and low peer caseloads. One aspect of this was that rural vendors did not have resources to cover evening hours when many overdoses presented to the ED. As an example, RV2 and RV3 peers missed patients due to restricted hours, with RV3 stating in their final interview that no overdose patients had been admitted to the ED when a peer was available. Another unexpected issue was that rural hospitals often transported overdose patients to urban hospitals that could provide services for which rural vendors were unequipped to handle. Because these patients were often unconscious prior to transfer, peers were unable to engage them or obtain a release required to speak about their case to staff at the hospital to which they were transported: “For what we see in our emergency room, [overdose patients are] either very ill, intubated, [or] transported to a higher level of care [at another location]” (RV3; T3). Finally, rural vendors also expressed in later interviews that the majority of patients who could benefit from peer services had other substance use disorders, which were outside of the RCPSI’s scope: “We have a strong uptick in stimulants here in this part of the world…I suspect that’s a big reason why we’re not seeing a lot of [opioid overdoses]” (RV2; T3). While urban vendors also noted the need to serve patients with other substance use disorders, enough OUD patients presented to the ED to make up a considerable portion of peer caseloads.

Connected to the above issue of patient transportation to urban hospitals are factors associated with the external relational climate, which includes the quality of relationships with other hospitals and providers. Despite rural hospitals having relationships and procedures necessary to transport patients to urban hospitals, they had not developed protocols that would allow peers to follow-up regarding a patient’s status. This lack of information sharing was a major barrier to implementation for rural hospitals because peers needed to communicate with patients to link them with treatment and services. As an example, RV3 expressed how their peers had no way of knowing if the patient was transferred to another hospital or if they are “gonna come back in our community” (T2) where they should be connected to local resources by the peer. Information sharing with other health providers is also a matter of the policy and legal climate, as this sharing is governed by external regulations, including rules, policies, and laws, that impact implementation. This was not an issue in urban hospitals, which had the resources necessary to address most OUD patients’ care needs internally.

Rural vendors also described their local infrastructure as lacking needed OUD services and resources. While urban vendors also described such issues, it was far more detrimental for rural ones. For instance, where urban vendors discussed issues finding transportation assistance to help patients attend referral appointments, rural vendors did not even have sufficient providers in their local areas to which they could refer patients. Indeed  most rural vendors  had to rely on a single MOUD prescriber, and this was a concern for them over the entire course of implementation. For instance, at T1 RV3 expressed concern about their sole MOUD prescriber who they felt was already overwhelmed with referrals. In the most extreme case, RV4 was not able to find any MOUD provider to work with them: “that was my biggest hurdle because I did not have access to a nurse practitioner or anyone [to prescribe MOUD]” (T3). This was a major factor leading to RV4’s not following through with RCPSI implementation.

Inner setting

Networks and communication among inner setting actors were key to implementation for all sites, with the fundamentally important line of communication demonstrated to be the one existing among peers and the ED’s medical staff. It is through this communication that peers are notified if and when they can approach an ED patient. All vendors had some growing pains establishing initial lines of communication; however urban vendors describe minimal issues or had largely solved any recognized problems by their final interview. For instance, at baseline, UV1 discussed their plan for integrating peers in the ED but described difficulties getting ED staff to understand that communicating with the off-site peers was not a violation of patient confidentiality at T2. However, this issue was largely solved by T3. Regarding communication issues at rural sites, RV1’s ED staff were supposed to call the peer when an appropriate patient presented. Though, they never identified a mechanism to ensure the peer was alerted. This resulted in the peer needing to “make more of an effort” (T2) to identify patients without ED staff assistance. Likewise, RV2 interviews demonstrated difficulty developing effective lines of communication between the ED staff and peer: “they [ED staff] would kind of just stare at her [the peer] and not talk to her” (T2). It was further explained that RV2’s ED staff would contact the social workers when OUD patients presented instead of alerting the peer. For all three rural vendors who carried through with implementation, they discussed these communication issues as persisting through their final interviews.

Differences in ED culture were highly noticeable when comparing rural and urban vendors. This was most apparent when it came to attitudes and actions of ED staff toward behavioral health care workers, including the peer. RV2 explained “[The ED staff] like things a certain way, and it’s hard to fit behavioral health into that box sometimes…This has been a historical thing for the hospital” (T2). RV3 also described a “bias of [i.e., against] addiction” (T3) among ED staff in which they hesitate to contact the peer because they do not believe there is anything that can be done to benefit overdose patients. By contrast, site leaders of urban vendors largely reported more welcoming ED staff attitudes, with the largest issue being the introduction of the peer role and its fit within the current ED hierarchy: “in the medical realm, the more initials you have behind your name, the higher status you have…Many of them [peers] don’t have any letters behind their names…but I feel like they’ve gained and earned respect within our health system” (UV5, T3).

The implementation climate was another important factor. Shared receptivity for RCPSI programming was lacking at RV2, where implementation leaders struggled to put necessary processes in place with the finance department and ED because there was a “resistance to behavioral health” (T2) that resulted in difficulties integrating peer services into pre-existing systems and workflows, and this issue was not fully solved by T3. RV3 attribute ED staff implementation resistance to the fact that they could not “hardwire” (T3) peer services into the pre-existing workflow since OUD patient volume was too low to justify a need for change among ED staff. In contrast, urban vendors and hospitals displayed more capacity for change (particularly regarding technological change), receptivity to peer services, and positive expectations about them among administration, ED staff, and behavioral health staff of involved organizations. As an example, UV2’s ED physicians were stated to be “very eager to hear [about the program]. They were interested in Narcan [the opioid overdose reversing drug] being used, they were very happy to hear about the [MOUD] clinic opening up and then, to hear about actually having a coach [i.e., peer] on site” (T1) from time they first learned about the peer services.

Characteristics of individuals

ED staff members’ knowledge and beliefs about peer services resulted in barriers to implementation initially for two urban (UV4 and UV6) and two rural (RV2 & RV3) vendors. In the case of the two urban sites, individual-level resistance receded over time. One interview with UV4 provides an example of something a physician said that highlights this: “[a physician said] I was really against this [peer supports] at first, but I kind of see this working, so I think I’m gonna try this with some of my patients” (T2). In contrast, individual-level physician resistance at the two rural sites persisted. At T2, an RV3 physician was described as being “reluctant to open up just to anybody and say that [he can prescribe] Suboxone [a band name formulation of the MOUD buprenorphine]” because he did not think he could meet the potential demand that would be created by the peer services. Additionally, RV2 “never could really get the [ED] staff on board with why you would call a peer recovery coach [i.e., peer] over an LCSW [Licensed Clinical Social Worker] when the LCSW has more training” (T3), suggesting they did not believe a peer could provide services of a similar quality.

Implementation process

Both rural and urban vendors recognized the importance of engaging health providers in various roles, including ED staff and local MOUD providers. One already discussed difference for rural sites was that they lacked MOUD providers in their communities to whom they could refer patients. Due to this lack of physicians, RV3 and RV4 sought out providers who they could support to obtain training necessary to prescribe MOUD, something no urban vendors needed to do. Although both rural and urban vendors described a process of engaging with and winning over the ED staff, mainly nurses, rural vendors experienced more difficulty with this over the course of implementation: “ [the peer service program] hasn’t been accepted [by ED frontline staff] quite as well as I thought it would be in the beginning” (RV2; T2). Engaging external organizations was also important for peer implementation. Both rural and urban vendors disseminated information about their peer services to local physicians, law enforcement agencies, and community organizations; however, rural vendors were much more focused on law enforcement. For instance, RV2 and RV3 developed relationships with local drug courts and probation departments, with RV2 using this as a means of providing more work for their peer given the low patient volume in the ED previously discussed: “she [the peer] also does some case management for our people in our court program” (T2).

In executing the implementation, both rural and urban vendors described peer hiring challenges and difficulties incorporating them into the ED workflow, but overall, rural vendors were less successful following through, with RV4 discontinuing and RV2 still trying to gain ED staff cooperation by the end of the first year of programming. As previously stated, rural vendors experienced greater difficulty getting ED staff to alert peers when patients presented who were eligible for their services, as with RV3: “[for] My [ED staff], still, [the peer is] not top-of-mind there yet, and she’s gone down there and spoken to them, and shadowed them for a shift, a couple different things, but yeah, if anybody has any ideas on how to get more buy in from your [ED staff], I’d be interested” (T3). RV2 described great difficulty setting the electronic health record system up to track peer services, which was not a problem for any of the urban vendors who discussed requiring similar technological adaptations. While urban vendors also described some challenges executing the implementation, these discussions focused more on successes and their underlying facilitators, as they indicated having more support from ED staff, more helpful technology departments with better resources, and an easier time integrating peer services into the pre-existing workflow. A typical example of the smoother execution of the implementation plan among urban vendors is demonstrated by a selection from in UV4’s T3 interview where participants listed barriers they had encountered and overcome:

The first barrier was training up a group of recovery coaches [i.e., peers] that we could look at to hire. We overcame that barrier…a few times we had some issue with ‘Oh my gosh how’s the person [patient] gonna get their medication?’…but we were able to utilize [our foundation] and different things to help….

Such discussions of successful resolutions to major implementation barriers were not a feature of later rural vendor interviews.