Implementation Strategies for Telestroke: A Qualitative Study of Telestroke Networks in North Carolina

Abstract

Background: Telestroke services allow under-resourced hospitals to gain access to stroke specialists to improve the timeliness and quality of stroke care. However, limited research is available on how telestroke networks are developed, implemented, and sustained. Understanding the process of telestroke network implementation is critical for developing implementation guidance and for evaluating determinants of implementation effectiveness.

Objective: This study examines the adoption decision process and strategies employed during telestroke network development, implementation, and sustainability.

Research Design: We interviewed representatives from five telestroke networks in North Carolina. Each network consisted of a distant site from which stroke consultation was provided and multiple originating sites at which the patient presented.

Subjects: The sample included 24 telestroke representatives (i.e., 5 network representatives and 19 hospital representatives) and 4 hospital representatives who do not participate in telestroke (i.e., nonadopters).

Measures: The qualitative measures used in this study were based on Roger's stages of the innovation process in organizations. Stages included agenda setting and matching (pre-implementation), restructuring, redefining, and clarifying (implementation), and routinizing (sustainability).

Results: Distant and originating sites employed various strategies in the pre-implementation, implementation, and sustainability stages. Although there are many commonalities across networks, there are also important differences, for example in terms of network structure, quality monitoring, and performance feedback. Some nonadopter hospitals reported difficulty accessing information about telestroke, suggesting that some hospitals are not reached by telestroke network marketing efforts.

Conclusions: Identifying and/or tailoring strategies to support the needs of hospitals in different telestroke network models should be a priority for future research.

Introduction

Telestroke programs aim at increasing access and timeliness of care—which are critical for stroke patients¹—via live video consultations between clinicians in emergency departments (ED) at the site in which the patient presents (originating site) and stroke specialists at a distant site. Without such consultations, ED physicians may be reluctant to prescribe the gold standard for treatment of ischemic strokes, tissue plasminogen activator (tPA),^2,3 due to concerns about potential adverse side effects.⁴ Because telestroke can reduce “door-to-needle time” for tPA⁵ and ultimately stroke-related morbidity and mortality^6–9 in a cost-effective manner,^10,11 it has been recommended by the American Heart Association and American Stroke Association when specialists are not available in person and when teleradiology systems approved by the Food and Drug Administration (or equivalent organization) are used.¹² Increasing access to specialty stroke care via telestroke is particularly promising for those residing in rural areas,¹³ where the stroke mortality rate is about 20% higher than urban areas.¹⁴ Specialized stroke teams are uncommon in rural hospitals,¹⁵ as is the use of tPA,¹⁶ and only half the U.S. population is estimated to reside within 60 min of a primary stroke center.⁶ Telestroke services bridge this gap and enable rural hospitals to provide care comparable to hospitals with in-person stroke specialists.^17,18

Despite the ability of telestroke to improve access to specialty stroke care, little is known about how telestroke programs are developed, implemented, and sustained. Prior studies have documented barriers to implementation—such as physician resistance and lack of local information technology (IT) staff^19,20—but have not examined the implementation strategies used to address these barriers. Implementation strategies are defined as “methods or techniques used to enhance the adoption, implementation, and sustainability of a clinical program or practice”^21–23 and could include, for example, assessing readiness for implementation or developing tools for quality monitoring.^22,23 A better understanding of implementation strategies for telestroke is important for improving the operations of existing networks and for increasing the reach of telestroke services because the strategies employed (and/or the ways in which they are employed) could affect how well telestroke is implemented.

To address this need, our study examines the adoption-decision process and implementation strategies employed during the stages of telestroke network development, implementation, and sustainability. Our findings have implications for policy makers, telestroke program administrators, and researchers.

Methods

Design

We used semi-structured interviews, complemented by secondary data sources from the American Hospital Association and Centers for Disease Control and Prevention (CDC) as well as a brief survey we developed, to examine the development and operations of five telestroke networks in North Carolina (NC). Each network consisted of a primary network site (the distant site from which specialists provide consultations) and multiple originating sites (acute-care hospitals where patients present with stroke symptoms and that receive consultation from distant sites). Networks used various models, including hub-and-spoke (network site provides consultations), third-party vendor (network site hires third-party vendor to provide consultations), or hybrid (a combination of hub-and-spoke and third-party vendor).²⁴

To collect background information on the history and operations of telestroke networks in NC (e.g., the number of originating sites in the network, stroke specialists available to provide consultations, and consultations completed), we administered a survey via email to a representative of each network (Table 1). In addition to these survey data, we used the 2015 American Hospital Association survey and the CDC's WONDER online database from the period 2012–2014 to identify organizational and community characteristics about the hospitals in the network (e.g., county-level stroke hospitalization rate).

Table 1.

Staff Roles of Network and Hospital Representatives (n = 28)

STAFF ROLE	NETWORK OR HOSPITAL REPRESENTATIVE	n
Telestroke Network Director	Network	4
Telestroke Network Medical Director	Network	1
Stroke Program Coordinator	Hospital	16
Emergency Department Manager	Hospital	3
Emergency Department Nurse Manager	Hospital	3
Critical Care Coordinator	Hospital	1

We conducted semi-structured interviews by telephone with network-level (distant site) representatives, originating site representatives, and representatives from acute-care hospitals not participating in telestroke to triangulate data sources.²⁵ Our interview guide was based on Roger's stages of the innovation process in organizations framework,²⁶ with follow-up questions asked to clarify participant responses and reveal key insights.²⁷ The Rogers framework delineates activities within broader implementations stages: pre-implementation, implementation, and sustainability. Pre-implementation includes activities prior to adoption of an intervention, such as conducting a needs assessment (“agenda-setting”) and determining whether the intervention is a good fit for the organization (“matching”).²⁶ Implementation occurs after adoption of an innovation and may include modifying the organization's infrastructure (“restructuring”), modifying the intervention to fit the organization's local context (“redefining”), and refining the implementation process (“clarifying”).²⁶ The sustainability stage involves activities such as modifying policies to include the intervention (“routinizing”).²⁶ Interviews were recorded and transcribed verbatim. Participants were recruited until we had representation from each category (i.e., network representatives and originating site representatives from all five networks as well as nonadopters of telestroke) and until we reached data saturation, defined as the point at which no changes were needed in the codebook²⁸ and no new themes were emerging within our implementation stages framework.

Participants

Our sample included 24 telestroke representatives (i.e., 5 network representatives and 19 hospital representatives) and 4 hospital representatives who do not participate in telestroke (i.e., nonadopters), for a total of 28 individuals. The staff roles of each of the representatives are presented in Table 1. At the network level, we most commonly interviewed the telestroke network director, and at the hospital level, we most commonly interviewed the stroke program coordinator (Table 1). Through online searches and direct contacts with hospitals in the state, we identified nine telestroke networks in NC. We contacted representatives from each of the nine and successfully recruited a representative at five networks to complete an interview, and four of these five also completed the background survey. There were eight originating-site representatives who did not respond to our email invitation and two representatives who refused to participate due to time constraints. Study participants received a $100 gift card. The University of North Carolina at Chapel Hill Institutional Review Board approved this study (IRB No. 16-2890).

Approach

For quantitative survey data, we produced descriptive statistics about the characteristics of the telestroke networks using Stata statistical software (version 15). For qualitative data, we used Dedoose software (www.dedoose.com) as a tool for conducting a thematic analysis approach, using topical codes based on the interview guide (Table 2).²⁷ Two research team members (A.A.T., K.T.) coded five transcripts to come to a consensus on coding. We selected the five transcripts from two network representatives and three hospital representatives. The five transcripts represented three out of the five networks. Once reaching consensus, the same team members divided and coded the rest of the transcripts. Within each stage of innovation, we identified the implementation strategies employed by network and hospital representatives based on Powell et al.'s^22,23 compilation of implementation strategies. After completing the analysis, we sent the findings to 10 study participants to review for accuracy and validity (i.e., member checking).^25,29 We selected the 10 participants based on their level of expertise regarding telestroke implementation, which was based on job tenure. The participants agreed with the findings and did not recommend any changes to the interpretation of the results.

Table 2.

Key Themes, Code Definitions, and Illustrative Quotations

KEY THEME AND CODE DEFINITION	ILLUSTRATIVE QUOTATION
Agenda Setting (Pre-implementation) Apply code when participant discusses factors related to adoption of telestroke such as motivating factors for adoption, gaining awareness of telestroke, or identifying and gaining buy-in among key stakeholders.	“If you're in a small, rural area like what we are, it's difficult to recruit specialists, especially high caliber specialists to this area. Telehealth services offer that on demand.” —Hospital Representative
Matching (Pre-implementation) Apply code when participant discusses assessing whether telestroke is a good fit for their hospital or strategies for planning out telestroke before implementation.	“We will send them a very detailed questionnaire that really covers from A to Z, who are their key stakeholders, we address what is their IT system like, what kind of wireless units do they have.” —Network Representative
Redefining (Implementation) Apply code when participant discusses factors related to modifying or tailoring telestroke to fit their hospital or network's local context.	“We tried to have fun with it [telestroke]. We gave him [the robot] the name of Sheldon and we made him a nice badge and he had a tag on him.” —Hospital Representative
Restructuring (Implementation) Apply code when participant discusses modifying infrastructure (e.g., hiring new staff) or workflow to accommodate telestroke.	“We've more added physicians, which is in part because of the size of our network, but also in part of just the size of our growing inpatient stroke service too.” —Network Representative
Clarifying (Implementation) Apply code when participant discusses activities related to scaling up and putting telestroke into wider use and refining processes for implementation.	“It's important that there is someone at the facility that is collecting the telestroke data. That was what we used as far as determining what needed to be changed in our process. If we had a specific delay, we were able to identify that by looking at those door times month after month.” —Hospital Representative
Routinizing (Sustainability) Apply code when participant discusses strategies to make telestroke a routine part of daily operations and overall program sustainability such as perceived benefits and barriers and planned changes to the program.	“Because of infrequency of use we make sure that the protocols are set out from the very beginning and that they're accessible for when we do actually need them.” —Hospital Representative

Results

Sample Characteristics

Three of the five telestroke networks used a hybrid model (combination of hub-and-spoke and third-party vendor models) (Table 3). On average, telestroke networks employed 6 consulting physicians and had 1.7 full-time equivalent (FTE) nonphysician providers and 0.4 FTE administrative staff. Each network reported providing more than 100 consultations per year. Most of the telestroke networks included some hospitals that were not affiliated with the distant site's healthcare delivery system.

Table 3.

Telestroke Network Characteristics (n = 5)

NETWORK CHARACTERISTICS	MEAN (SD)
Years of operation	6.0 (2.83)
No. of consulting physicians	8.0 (1.73)
No. of FTE nonphysician providers (e.g., nurses)	1.7 (1.04)
No. of FTE administrative staff (e.g., schedulers and administrative assistants)	0.4 (0.48)
No. of originating sites	12.3 (8.73)

NETWORK CHARACTERISTICS	%
Type of model
Hub-and-spoke	0.40
Third-party vendor or hybrid model	0.60
Inclusion of nonaffiliated hospitals in network
Yes	0.67

FTE, full-time equivalent; SD, standard deviation.

The sample included 19 originating-site hospitals. On average, these hospitals had operated their telestroke program for 2.5 years. The majority of hospitals (62.5%) were classified as critical access hospitals, and 41.7% had another telemedicine program (telepsychiatry) (Table 3). On average, the originating-site hospitals had 56.2% of residents in their market living in a rural area with the remaining residents (43.8%) residing in either urban or suburban areas. In addition, most originating-site hospitals were located in counties with 11.8 stroke hospitalizations per 1,000 residents—a rate that is much higher than the national average (3.18 per 1,000).³⁰ Regarding hospital characteristics, hospitals in this sample had an average of 197.7 staffed beds and 45,300 ED visits per year. The average total operating margin (a profitability indicator) of hospitals in the sample was 5.0%, which is higher than the national average.³¹ All originating-site hospitals in the sample were affiliated with a delivery system.

The sample also included 4 hospitals that did not participate in telestroke (i.e., nonadopters) (Table 4). On average, nonadopter hospitals had 417.9 staffed hospital beds, 81,500 emergency room visits per year, and a total operating margin of 0.4%. In terms of community-level characteristics, nonadopter hospitals had, on average, 58.7% of the residents in their market living in a rural area and were located in counties with 12.5 stroke hospitalizations per 1,000 residents. Although there appear to be differences across adopters and nonadopters (e.g., total operating margin), we did not assess statistical significance given the small number of nonadopter hospitals in the sample.

Table 4.

Characteristics of Originating Site and Nonadopter Hospitals (n = 23)

	ORIGINATING SITE (n = 19) MEAN (SD) OR%	NONADOPTER (n = 4) MEAN (SD) OR%
No. of years providing telestroke	2.5 (1.7)	N/A
Percent of residents in market living in rural area	56.2 (37.9)	58.7 (13.9)
County-level stroke hospitalizations	11.8 (2.1)	12.5 (1.4)
Total operating margin	0.05 (0.2)	0.004 (0.1)
No. of staffed hospital beds	197.7 (205.7)	417.8 (247.7)
No. of emergency room visits (in thousands)	45.3 (31.2)	81.5 (33.3)
Classified as critical access hospital	62.5%	50.0%
Presence of telepsychiatry program	41.7%	50.0%

N/A, not applicable; OR, odds ratio.

Telestroke Network Development at The Distant Sites

Pre-implementation stage

Four representatives (4/5) from distant sites (i.e., network representatives) identified a lack of neurologists in rural hospitals as a key motivator for telestroke network development. These representatives discussed similar pre-implementation strategies for developing the network, including conducting a needs assessment (3/5) and conducting consensus discussions with key internal stakeholders (e.g., system and hospital-level leadership) and external stakeholders (e.g., hospitals interested in telestroke) (5/5) (Table 5). All networks had a process for identifying potential originating sites and for assessing readiness of these sites, although the methods and criteria used were not identical across networks. Four networks (4/5) began implementation with hospitals affiliated with their delivery system and/or located in close proximity. Two networks (2/5) recruited hospitals located in other regions of the state that lacked access to telestroke. An originating site's ability to receive stroke patients (e.g., having a computed tomography scanner), interest in receiving telestroke training, and ability to pay fees associated with joining the network were all important considerations. Three networks (3/5) administered a survey to hospitals to assess potential barriers.

Table 5.

Implementation Strategies Used by Distant and Originating Sites at Each Stage of Implementation

	NETWORK SITE	HOSPITAL SITE
Pre-implementation	Conduct a local needs assessment to determine whether there is a need for telestroke (3/5)	Conduct a local needs assessment to determine whether there is a need for telestroke (13/19)
	Conduct local consensus discussions among external and internal stakeholders (5/5)	Conduct local consensus discussions among external and internal stakeholders (19/19)
	Determine which hospitals would be natural partners for the telestroke network (5/5)	Assess readiness and identify barriers (19/19)
	Assess readiness of potential partner sites and identify barriers (3/5)
	Select a telestroke network model (5/5)
Implementation	Create governing structure (5/5)	Select and deploy telestroke technology (15/19)
	Access new funding to cover start-up costs (5/5)	Revise professional roles (7/19) or hire new staff (12/19)
	Create fee structure (2/5)	Upgrade Internet capabilities (10/19)
	Develop a formal implementation blueprint (5/5)	Develop a formal implementation blueprint (18/19)
	Develop and deliver staff training (5/5)	Involve patients and family members in telestroke implementation (16/19)
	Tailor strategies to overcome barriers and honor preferences (3/5)	Develop and distribute educational materials to community members (8/19)
	Hire additional support staff (4/5)	Use mass media to inform community members about telestroke (5/19)
	Develop and organize quality monitoring systems and tools (5/5)	Develop and deliver staff training (19/19)
	Provide performance feedback (5/5)
Sustainability	Stage implementation scale-up (3/5)	Identify and prepare champions (13/19)
	Provide ongoing education (5/5)	Provide ongoing training (19/19)
	Develop and strengthen relationships with hospitals (3/5)	Develop and organize quality monitoring systems and tools (19/19)
		Provide performance feedback through either formal or informal mechanisms (19/19)

[Note: In parentheses, we indicate the number of interviewees from the sample who discussed the strategy]. Powell et al.^22,23 and Krefting et al.²⁵

All networks determined which model to use (i.e., hub-and-spoke, third-party vendor, or hybrid) based on several characteristics of potential originating sites, such as geography, stroke patient volume, and pre-existing relationships. As an example, a hub-and-spoke model was typically preferable for hospitals that had a pre-existing relationship for patient transfers and/or were located in close proximity to the distant site. Conversely, a third-party vendor or hybrid model was used for some hospitals that were located farther from the distant site or that had a low stroke volume.

Network implementation stage

Key considerations reported were establishing a governance structure for medical, legal, and technical issues as well as deciding how to fund the network. Three networks (3/5) used grants for start-up costs (e.g., USDA Rural Development grants) and two networks created fee structures (2/5), such as using a per-click model versus a monthly retainer fee. Three networks (3/5) also charged start-up fees for originating sites.

All networks developed implementation policies, such as criteria for transferring patients, documenting stroke care across electronic health record (EHR) systems, and collecting performance measures from hospitals. Privileging consulting physicians at each originating site was particularly time-consuming as processes varied across sites. All networks used various strategies for preparing originating sites, such as mapping telestroke service workflow, conducting mock code strokes, training staff on stroke care (e.g., tPA and NIH stroke scale administration) and telestroke technology, and providing network representatives onsite during the go-live date.

As networks evolved, most (4/5) hired new staff to support expansion (e.g., administrative assistants). Four networks (4/5) also refined processes for performance monitoring and evaluation, such as setting up quality improvement (QI) committees to review performance data. Two networks (2/5) also refined approaches to delivering performance feedback, such as visualization of performance data. Four networks (4/5) provided performance feedback through phone calls or in-person meetings. Three networks (3/5) tailored the amount of implementation support based on the originating site's capacity; for example, hospitals without a stroke coordinator needed additional assistance with redefining staff roles.

Network sustainability stage

Networks reported common strategies in the sustainability stage. Participants highlighted the importance of (1) staging network scale-up to align the number of originating sites with the amount of network support available (3/5), (2) providing ongoing education and training to originating sites (5/5), and (3) developing relationships with the originating sites to facilitate open communication and help ensure that ED physicians trust the network's consulting physicians (3/5).

Telestroke Implementation by Originating Sites

Pre-implementation stage

Similar to distant sites, originating sites conducted local needs assessments (13/19) and worked to gain buy-in among key stakeholders (19/19). Hospital representatives learned about telestroke services primarily through professional networks, recruitment from a telestroke network, and conducting an environmental scan. Motivating factors for joining a network were similar to those identified by network representatives—lack of neurologists, high prevalence of stroke hospitalizations in their county, and desire to improve stroke care.

Originating sites gained buy-in from external and internal stakeholders. External stakeholders included their hospital boards and local emergency medical services (EMS) agencies—although EMS was less common. Internal stakeholders included the hospital's senior leadership, IT department, medical records department, human resources personnel, and ED personnel (e.g., ED manager, physicians, and nursing staff). Communicating the need for telestroke services (e.g., using past measures of performance in stroke care) and providing examples of how telestroke improves patient care were reported as important for gaining buy-in.

Originating sites evaluated their hospital's readiness for telestroke (19/19), such as having a stroke coordinator or other staff to serve as an implementation leader and having sufficient IT support staff. Several hospitals hired new staff to support the telestroke program (12/19) or redefined existing staff roles to support the telestroke program (7/19). Also important was having sufficient Internet capabilities, such as wireless connection throughout the hospital to enable movement of patients without losing connection with the consulting physician. Another strategy reported was assessing local EMS capability to evaluate stroke patients and alert the hospital of potential stroke patients ahead of their arrival (8/19).

Implementation stage

Four hospitals (4/19) established a request for proposals for telestroke technology, whereas the majority of hospitals (15/19) used the technology selected by their telestroke network. Hospitals also had to address issues identified in their readiness assessment (e.g., appointing a staff member to coordinate stroke care [19/19], upgrading Wi-Fi [10/19]). Nearly all originating sites developed new policies (18/19), for example, to share information within the EHR between originating and distant sites and to coordinate telestroke services between the ED and in-patient care. Some originating sites did not have the same EHR system as the distant site (7/19) and in those instances, the sites had to share patient information through a separate telestroke management system and then upload that information into their EHR system at a later time.

Originating sites aimed at increasing patient and broader community awareness about telestroke. For example, four hospitals held contests among patients to name the telestroke robot (4/19). Most hospitals also developed protocols for introducing patients and their families (16/19) (1) to the consulting physician to establish rapport and (2) to the telestroke technology itself. Some hospitals used local marketing and education campaigns (e.g., community health fairs) to raise awareness among community members and organizations (e.g., primary care providers, large employers) (8/19) and advertised their telestroke program through newspapers, television, and radio (5/19).

Sustainability stage

Most participants reported routinization of telestroke within their ED, with physician/nurse champions (13/19) and training (19/19) (for both new staff during orientation and ongoing for existing staff) being reported as keys to routinization. Most hospitals had formal structures (e.g., QI committees or staff meetings) for providing performance feedback (14/19). Among those 14 hospitals, some hospitals also used ad hoc strategies (e.g., debriefing with physicians and nurses after a stroke case) for providing performance feedback (7/19). There were also a small number of hospitals that did not have any formal structures for delivering feedback and reported only having ad hoc strategies (5/19).

Considerations for hospitals without telestroke

The hospitals in our sample not offering telestroke reported being in an adoption-decision process—either considering telestroke but not having a timeline for adoption (2/4) or already having identified a telestroke network and working toward implementation (2/4). These hospitals identified similar motivators for telestroke as originating and distant sites (e.g., lack of neurologists, ability to treat patients locally). Those working toward implementation described similar preparation strategies (e.g., outreach to EMS, developing a detailed implementation plan), similar facilitators to adoption (e.g., having another telemedicine program, a stroke performance measurement process, and Joint Commission stroke certification), and similar barriers (e.g., not having a stroke coordinator, costs) as hospitals currently serving as originating sites. However, two nonadopter hospitals (2/4) discussed barriers not reported by the hospitals in our sample that were currently participating in a telestroke network. Specifically, these nonadopter hospitals reported having difficulty obtaining specific information about the cost of participation in a network and how much technical assistance would be provided. In addition, these hospitals discussed being unaware of which hospitals in their region were participating in telestroke and, therefore, did not know whom to reach out to learn more about other hospitals' experiences.

Discussion

We explored adoption-decision processes and implementation strategies employed by distant and originating sites in telestroke networks in NC. We categorized these issues and strategies by type of site (i.e., distant or originating) and stage of the implementation process—pre-implementation, implementation, and sustainability. Several common considerations and strategies appeared across networks and across originating and distance sites, but some also were unique to, or applied differently across, networks and sites.

Because distant and originating sites shared similar beliefs about drivers of telestroke adoption and a desire for providing high-quality stroke care, perceived need for telestroke does not appear to be a substantial barrier, at least in NC. Instead, efforts to promote and sustain telestroke networks should focus on resources and capabilities. Both distant and originating sites expressed concerns about start-up and maintenance costs without a clear funding stream, which is consistent with previous studies.^32–34 In addition, sites explained that lack of reimbursement for telestroke services was a key barrier to program sustainability.^32–34 Future studies should test reimbursement models for telestroke services and whether changes in reimbursement affect the profitability and sustainability of telestroke services. For some sites, both costs and access to human and technology resources (e.g., Internet capabilities such as wireless connection and data-transmission speed) are problematic.

Originating sites belonging to networks that provide easy-to-use technology and thorough planning and support likely are in a better position for successful implementation. Network assistance with monitoring and QI is particularly important—a finding consistent with other studies.^32,34 For example, sites in our study reported difficulty determining how to measure time to initiation of telestroke processes (e.g., video-consult initiation) and tracking patient transfers, which are challenges also identified by previous research.^35,36 Notably, hospitals belonging to a network not operated by their delivery system may be treated differently in terms of quality reporting and QI support, as compared with hospitals belonging to a telestroke network operated by their delivery system. Establishing a set of quality standards for telestroke providers and creating a certification program for telestroke programs³⁵ would facilitate greater standardization of care and could prove particularly useful for evaluating the quality of different telestroke vendors.

More research is needed to characterize specific types of network structures and to examine how network structure affects stroke care quality and access, perhaps using spatial and/or network analytic methods. We found that some nonadopter hospitals had difficulty accessing information about telestroke networks. Educational interventions could be developed to increase knowledge about telestroke programs. Also, originating sites in our study reported having strong relationships with distant sites and attributed their success with telestroke to support received from the network. Future studies could evaluate whether specific network characteristics (e.g., strength of relationships between hospitals) are associated with improved performance, which previous studies of other industries have suggested.³⁷

A notable limitation in our study is that the hospitals were located in the Stroke Belt of the United States and from counties with a higher than national average stroke hospitalization rate.³⁰ Therefore, these hospitals may have a greater volume of stroke patients and thus different experiences with telestroke implementation than hospitals located in other regions of the country that have lower stroke prevalence. Future research should assess the extent to which our findings represent the pre-implementation, implementation, and sustainability considerations and activities in other geographic areas. In addition, the nonadopter hospitals that we interviewed were much larger than the participating hospitals on average (417.8 staffed hospital beds vs. 197.7 staffed hospital beds). It is possible that larger hospitals are less likely to adopt telestroke because larger hospitals may be more likely to have a neurologist on staff.¹⁵ However, all of the nonadopter hospitals we interviewed expressed interest in participating in telestroke, suggesting that there is a need for telestroke among these hospitals. Future studies should explore how telestroke needs and barriers to adoption vary based on hospital size.

Conclusions

Our study suggests that concerns about acquisition and maintenance cost are threats to widespread implementation and sustainability of telestroke programs. Proposed policies, such as Furthering Access to Stroke Telemedicine Act in the United States,³⁸ indicate that policy makers are aware of such concerns. However, in addition to cost, our study highlights important differences in terms of barriers, resource needs, and quality monitoring across hospitals, particularly between hospitals that are participating in telestroke networks within their integrated delivery system and those that are not. Evaluating a process for supporting the needs of hospitals in different telestroke network models is an opportunity for future research.

Disclosure Statement

The authors have no conflicts to report.