This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Medical Informatics, is properly cited. The complete bibliographic information, a link to the original publication on http://medinform.jmir.org/, as well as this copyright and license information must be included.
Stroke is a leading cause of death and serious, long-term disability across the world. Urgent stroke care treatment is time-sensitive and requires a stroke-trained neurologist for clinical diagnosis. Rural areas, where neurologists and stroke specialists are lacking, have a high incidence of stroke-related death and disability. By virtually connecting emergency department physicians in rural hospitals to regional medical centers for consultations, specialized Web-based stroke evaluation systems (telestroke) have helped address the challenge of urgent stroke care in underserved communities. However, many rural hospitals that have deployed telestroke have not fully assimilated this technology.
The objective of this study was to explore potential sources of variations in the utilization of a Web-based telestroke system for urgent stroke evaluation and propose a telestroke assimilation model to improve stroke care performance.
An exploratory, qualitative case study of two telestroke networks, each comprising an academic stroke center (hub) and connected rural hospitals (spokes), was conducted. Data were collected from 50 semistructured interviews with 40 stakeholders, telestroke usage logs from 32 spokes, site visits, published papers, and reports.
The two networks used identical technology (called Remote Evaluation of Acute isCHemic stroke, REACH) and were of similar size and complexity, but showed large variations in telestroke assimilation across spokes. Several observed hub- and spoke-related characteristics can explain these variations. The hub-related characteristics included telestroke institutionalization into stroke care, resources for the telestroke program, ongoing support for stroke readiness of spokes, telestroke performance monitoring, and continuous telestroke process improvement. The spoke-related characteristics included managerial telestroke championship, stroke center certification, dedicated telestroke coordinator, stroke committee of key stakeholders, local neurological expertise, and continuous telestroke process improvement.
Rural hospitals can improve their stroke readiness with use of telestroke systems. However, they need to integrate the technology into their stroke delivery processes. A telestroke assimilation model may improve stroke care performance.
Stroke is a leading cause of death and serious, long-term disability in the United States. In 2008, nearly 800,000 people suffered a stroke, resulting in the deaths of more than 134,000 people [
For ischemic (ie, nonbleeding) strokes, a blood-clot dissolving drug tissue plasminogen activator (tPA) greatly reduces the risk of severe disabilities if administered within 4 ½ hours from the onset of stroke symptoms [
Information technology (IT)—in the form of specialized Web-based telemedicine systems that include videoconferencing and supporting applications that enable a remote stroke specialist to view and evaluate a patient—has helped address the challenge of urgent stroke care in underserved communities [
IT utilization (or assimilation) can be defined as “the extent to which the use of technology diffuses across the organizational projects or work processes and becomes routinized in the activities of those projects and processes” [
Recent studies have explored IT adoption in health care organizations [
Recent telestroke literature has focused on the organizational, managerial, financial, technical, and legal issues that influence adoption. The enablers of telestroke adoption include a stroke systems of care model with primary and comprehensive stroke centers of excellence, statewide and local stroke champions, pre-hospital and in-hospital coordination, favorable regulatory and reimbursement policies, stakeholder support and communication, and appropriate IT infrastructure [
Based on purposive sampling [
Recognizing the potential of using telestroke to link hub-based specialists to rural hospitals, a team of GRU neurologists developed the REACH system. The system comprised a mobile, Internet-ready REACH cart (with a mounted adjustable camera, a phone, and a high-resolution monitor) that could be wheeled into the ED room where the stroke patient was being examined. As shown in
REACH Web interface showing a patient's CT scan.
We collected primary data between March and August 2012 by visiting the 2 hubs and 8 selected spokes (
We also collected archival data from the 2 hubs related to REACH consultations with each spoke since the start of the telestroke program. To account for variations in spoke ED volume across hospitals, we adjusted the annual rate of REACH consultations at each hospital by its reported ED volume. We refer to the average adjusted annual telestroke consultation rate (calculated as number of REACH consultations/year per 104 ED volume) as REACH assimilation. Thus, we consider the REACH-enabled consultation rate as a proxy for telestroke assimilation. It must be emphasized that this paper focuses on the decision-making enabled by the telestroke technology; therefore, we have examined REACH consultations rather than the resulting tPA usage.
Primary and secondary data sources.
Primary data sources | Secondary data sources |
15 semistructured interviews at 2 hubs (with neurologists, stroke coordinators, ED nurse managers, stroke service line manager, and data analyst) | 14 published papers [ |
30 semistructured interviews at 8 spokes (with chief executive officers and chief operations officers, stroke coordinators, neurologists, ED directors, ED physicians, ED nurses, quality managers, radiology nurses, and EMS directors) |
15 internal documents related to 8 spokes (including presentations, stroke protocols, emails, and meeting notes) |
5 semistructured interviews at REACH Health Inc (with chief executive officer, chief technology officer, marketing director, business manager, and IT specialist) |
5 internal documents (including presentations, technical specifications, and meeting notes) |
Variation in telestroke assimilation across networks.
Network characteristics and REACH assimilation data.
Telestroke network | Spoke hospital | Joining date | No. of |
Primary stroke center | Stroke coordinator | Local neurologist | REACH assimilationa |
GRU | 1 | 2/1/03 | 72 | No | No | No | 18.70 |
GRU | 2 | 3/1/03 | 47 | No | No | No | 15.31 |
GRU | 3 | 7/1/03 | 50 | No | No | No | 26.24 |
GRU | 4 | 8/1/03 | 10 | No | No | No | 18.12 |
GRUb | 5 | 9/1/03 | 56 | No | No | No | 21.33 |
GRU | 6 | 3/1/04 | 65 | No | No | No | 16.44 |
GRU | 7 | 4/1/04 | 20 | No | No | No | 9.01 |
GRUb | 8 | 2/1/05 | 52 | No | No | No | 17.71 |
GRU | 9 | 3/1/06 | 71 | No | No | No | 13.72 |
GRU | 10 | 1/1/08 | 191 | No | No | Yes | 7.93 |
GRU | 11 | 8/1/08 | 236 | Yes | No | Yes | 7.35 |
GRU | 12 | 6/1/09 | 40 | No | No | No | 19.51 |
GRU | 13 | 10/1/09 | 190 | Yes | Yes | Yes | 21.42 |
GRUb | 14 | 10/1/09 | 196 | Yes | Yes | Yes | 22.65 |
GRU | 15 | 1/1/10 | 180 | Yes | Yes | Yes | 47.40 |
GRU | 16 | 3/1/10 | 163 | No | No | Yes | 8.39 |
GRUb | 17 | 11/1/10 | 192 | No | Yes | Yes | 14.88 |
MUSC | 1 | 5/1/08 | 131 | No | No | Yes | 20.81 |
MUSC | 2 | 5/6/08 | 140 | No | No | Yes | 30.66 |
MUSCb | 3 | 5/7/08 | 453 | No | Yes | Yes | 14.77 |
MUSC | 4 | 9/1/08 | 220 | No | No | No | 4.47 |
MUSC | 5 | 9/18/08 | 124 | No | No | No | 15.89 |
MUSC | 6 | 12/23/08 | 25 | No | Yes | No | 35.10 |
MUSCb | 7 | 1/20/10 | 45 | Yes | Yes | Yes | 41.62 |
MUSCb | 8 | 3/26/10 | 288 | Yes | Yes | Yes | 26.84 |
MUSCb | 9 | 5/19/10 | 121 | No | Yes | Yes | 32.93 |
MUSC | 10 | 7/29/10 | 79 | No | No | No | 28.80 |
MUSC | 11 | 8/26/10 | 231 | No | No | No | 21.69 |
MUSC | 12 | 01/21/11 | 116 | No | No | Yes | 12.64 |
MUSC | 13 | 2/28/11 | 105 | No | No | Yes | 30.30 |
MUSC | 14 | 2/28/11 | 50 | No | No | Yes | 21.00 |
MUSC | 15 | 3/2/11 | 354 | No | Yes | Yes | 27.30 |
aREACH assimilation calculated as number of telestroke consultations/year per 104 ED volume.
bSpokes selected for detailed examination (through field visits).
Based on primary and secondary data analysis, we identified several hub-related practices that can explain the superior telestroke assimilation in the MUSC-REACH network.
Comparison of hub-level practices.
GRU-REACH hub | MUSC-REACH hub |
GRU-REACH hub invited most of the early spokes to become part of the network and subsidized their participation; most recent spokes sought membership without subsidies. | MUSC-REACH hub invited most of the early spokes to become part of the network, but participation was not subsidized; most recent spokes also sought membership without subsidies. |
GRU administration considers telestroke as an ongoing experiment rooted in the vision and goodwill of the stroke specialists who developed REACH. As such, the specialists feel REACH is “taken for granted.” GRU administration does not provide support for telestroke operations. | MUSC administration considers telestroke an integral part of their neuroscience service line, and therefore provides ongoing support (including director’s pay, advertising budget, and administrative salary support for credentialing, billing, operations, and project management). |
There is broad consensus among the hub stroke specialists that network performance would benefit from a full-time telestroke coordinator. | A dedicated telestroke coordinator at the hub has been part of the network from the start. She facilitates coordination and training of the spokes’ ED staff. |
The hub has no established processes for reinforcing telestroke use and related routines at the spokes. There are no continuous quality improvement processes in place. Any problems related to stroke consultations are reported to REACH Health Inc with variable follow-up. | The hub has established processes for reinforcing telestroke use and related routines at the spokes. It has a formal continuous quality improvement process in place. Any problem during telestroke consultation is reported to REACH Health Inc and its resolution is coordinated by the hub staff. |
The hub collects spokes’ telestroke use data, but there is no systematic analysis of the data. | The hub telestroke coordinator collects spokes’ usage data and conducts systematic analysis. |
A hub stroke specialist visits spokes when they go live with REACH and at rare occasions for major upgrades. However, there are no ongoing training and follow-up procedures. | A hub telestroke specialist visits spokes when they go live with REACH and maintains regular communication (with some visits) to spokes to understand concerns and train ED staff. |
The hub stroke specialists rarely conduct ongoing training for spokes. | The hub facilitates occasional breakfast meetings, lunch-and-learn, mock-consults, and dinners with spoke ED physicians and nurses to discuss issues. |
The hub has no formal system to provide site-specific feedback. | The hub provides site-specific performance data. As an MUSC-REACH stroke specialist told us, “The sites love to receive such feedback.” |
Spoke-level REACH assimilation varied from 7.35 in GRU-11 to 47.4 in GRU-15 (average 18.00), and from 4.47 in MUSC-4 to 41.62 in MUSC-7 (average 24.32). We cannot explain these large variations by length of relationship with the hub or size of the spoke. For example, GRU-3 and GRU-4 joined the network within 1 month of each other, but still showed variation in assimilation (26.24 and 18.12). Moreover, GRU-13 joined the network more than 6 years after GRU-5 and both showed similar REACH assimilation (21.42 and 21.33). Furthermore, MUSC-5 and MUSC-9 had similar number of beds (124 and 121), but showed considerable variation in REACH assimilation (15.89 and 32.93).
To explain the observed variations across all spokes, we first considered the availability of local neurological expertise for post-tPA patient supervision. Seven spokes in GRU-REACH and 10 spokes in MUSC-REACH had an on-call local neurologist. As
Next, we considered whether stroke center certification had an impact on telestroke assimilation. The US Joint Commission certifies acute care hospitals as “Primary Stroke Centers” if they have specialized knowledge and infrastructure to treat stroke patients. The certification signifies that a hospital has necessary stroke-related facilities (such as ED, EMS, and stroke unit), services (such as neurological, neuro-imaging, laboratory, and clinical support), personnel (such as acute stroke teams), practices (such as written care protocols, outcome and quality improvement activities, and continuing medical education), and commitment and support of the medical organization [
We also considered the impact of a telestroke coordinator. Such a position may help spokes establish standard processes for stroke care; collect, analyze, and use performance data to continually improve care delivery; and, become a stroke champion in the hospital and in the local community. Four spokes in GRU-REACH and 6 spokes in MUSC-REACH had a dedicated telestroke coordinator. As
To confirm and elaborate these explanations, we conducted an in-depth analysis of telestroke use at 4 selected spokes in each network. Helped by long-standing relationships with the 2 hubs, we visited these spokes and interviewed key stakeholders associated with stroke operations. These interviews provided additional insights into the current and historical context of REACH implementation at these spokes. Accordingly, we identified several notable practices that can further explain variations in telestroke assimilation across spokes.
Impact of spoke characteristics on telestroke assimilation.
Spoke characteristic | REACH assimilationa in GRU-REACH | REACH assimilation in MUSC-REACH | Overall REACH assimilation | |
|
|
|
|
|
|
No local neurologist | 17.61 | 21.19 | 18.80 |
|
Local neurologist | 18.57 | 25.89 | 22.88 |
|
Difference (%) | 5.45 | 22.18 | 21.70 |
|
|
|
|
|
|
No stroke certification | 15.95 | 22.80 | 19.37 |
|
Stroke certification | 24.70 | 34.23 | 27.88 |
|
Difference (%) | 54.86 | 50.13 | 43.93 |
|
|
|
|
|
|
No stroke coordinator | 15.37 | 20.69 | 17.55 |
|
Stroke coordinator | 26.59 | 29.76 | 28.49 |
|
Difference (%) | 73.00 | 43.84 | 62.34 |
aREACH assimilation calculated as number of telestroke consultations/year per 104 ED volume.
In 6 of the 8 spokes that we visited, a combination of local neurological expertise and telestroke provided urgent stroke care. The local neurologists would follow up on patients admitted locally, including post-tPA stroke patients, in the intensive care unit (ICU). In some cases (eg, GRU-5 and MUSC-9), all emergency consultations were handled via telestroke. In other cases, local neurologists also provided acute stroke coverage in the ED either during daytime (GRU-14) or 15 days/month (GRU-17). Overall, the combination of local neurology support and REACH coverage afforded spokes expanded stroke care capability.
Three of the 8 spokes that we visited (GRU-14, MUSC-7, and MUSC-8) had received primary stroke center certification. This suggests that they had established the necessary infrastructure, acquired stroke-related specialized knowledge, and developed standardized protocols and best practices to manage urgent stroke patients. In 2004, GRU-14 became the first spoke in Georgia to receive certification. To achieve that, GRU-14 set up a dedicated stroke unit, hired three neurologists, and developed standardized protocols (such as a written “stroke code”). Prior to the stroke certification, the local EMS “dreaded bringing stroke patients to the hospital because they were not sure that the hospital had capability to deliver urgent stroke care,” and instead took the patients directly to the nearest tertiary medical center. However, as GRU-14 advertised its stroke care capabilities, the local EMS started to bring stroke patients to the hospital. Similarly, after MUSC-7 gained certification in 2010, its acceptance as the preferred stroke care center in the region increased, resulting in a growing number of stroke patients admitted at the hospital. When needed, the ED staff at these hospitals connected to GRU-hub via REACH for consultations.
At 6 of the 8 spokes that we visited (GRU-14, GRU-17, MUSC-3, MUSC-7, MUSC-8, and MUSC-9), a telestroke coordinator set up and developed requisite processes, and facilitated collaboration within the hospital and with the hub. The coordinator provided ongoing feedback and training to ED nurses to reinforce and improve stroke-related processes, and conducted systematic spoke performance analysis. The coordinator helped to develop best practices (such as taking blood samples for laboratory analysis while the patient was in the CT scan room), which helped to reduce delays in stroke treatment. A full-time coordinator at GRU-17 reviewed each stroke case and reported any deficiencies (eg, missed stroke diagnosis, or delays in CT scan). Spokes (eg, GRU-8) that had no dedicated telestroke coordinator used the services of a part-time coordinator. At GRU-5, MUSC-7, MUSC-8, and MUSC-9, the coordinator conducted community awareness initiatives (including health fairs, and advertisements in the local newspapers, radio, and television) to provide information about stroke symptoms and related services available at the hospital.
Senior leadership support was critical to establishing and fostering telestroke capability at the spokes. In 5 of the 8 spokes that we visited (GRU-14, GRU-17, MUSC-7, MUSC-8, and MUSC-9), the senior leadership realized the value of telestroke and encouraged the ED and other staff to make it an integral part of urgent stroke care. They also provided requisite IT infrastructure and resources, and facilitated a culture of continuous improvement. In contrast, at GRU-8, several years of managerial neglect had led to a situation where the ED staff routinely referred stroke patients to other hospitals. Over time, they lost their stroke-handling skills. A nurse manager elaborated on the situation:
A few years ago, the ED staff knew what to do in case of a stroke patient. Now, I am not sure they do. I guess they don’t know when to trigger the REACH system.
A stroke committee—consisting of a telestroke coordinator, ED physicians and nurses, radiology staff, and EMS—proved essential to improving telestroke practices at GRU-14, GRU-17, MUSC-3, MUSC-7, and MUSC-8. Emphasizing the need for coordination, the chief of medical staff at MUSC-7 said, “We consider stroke to be a team event.” In some spokes, the committee also facilitated a cultural change. An ED physician at GRU-14 explained:
When I arrived here 3 years ago, we did not have a stroke care culture. The stroke committee took ownership of the stroke program and led the change in culture from within. Now, stroke is a source of identity for the hospital.
The committees met regularly to discuss issues and to find ways to enhance stroke readiness. The role of ED physicians and nurses in stroke committees was critical. In some spokes (eg, GRU-5), the nurses encouraged the ED physicians to initiate the REACH call, while at others (eg, GRU-14), the ED physicians themselves contacted the remote specialist. In all cases, however, the ED physicians made a decision (to treat locally or transfer patients) based on availability of local neurology support and neuro-ICU facilities in their hospital. Deliberate engagement of the local EMS in some spokes (MUSC-7 and GRU-14) improved stroke performance by reducing patient transportation time. Similarly, a pro-active EMS became an integral part of stroke care at MUSC-8. At MUSC-9, the hospital-owned EMS became the “voice of the hospital.”
Spokes with superior stroke performance (eg, GRU-14 and MUSC-7) focused on improving their stroke delivery processes. Their stroke committees had developed protocols and training procedures to sustain and improve urgent stroke care. Stroke care-related staff at GRU-14 and GRU-17 regularly exchanged best practices and updates with colleagues in other hospitals. The chief financial officer at GRU-17, who trained as a Six Sigma Master Black Belt, had initiated several quality improvement initiatives to improve stroke care. Over time, GRU-17 fostered shared responsibility for stroke care and created a systematic basis for continuous improvements. In contrast, GRU-8 did not have established routines or process improvement initiatives to develop their urgent stroke care capability. At GRU-14, MUSC-7, and MUSC-8, the process improvement initiatives helped achieve the coveted primary stroke center accreditation.
The existing IT literature emphasizes how organizational factors enable technology utilization in key processes to enhance business performance [
The identified hub factors included (1) institutionalization of telestroke by making the technology an integral part of stroke delivery, (2) providing required resources for telestroke program, (3) ongoing support for stroke readiness of spokes, (4) telestroke performance monitoring with site-specific feedback, and (5) continuous process improvement to improve telestroke delivery. Similarly, the identified spoke factors included (1) managerial telestroke championship, (2) stroke center certification, (3) dedicated telestroke coordinator, (4) stroke committee consisting of key stakeholders, (5) availability of local neurological expertise, and (6) continuous telestroke process improvement. These empirical findings suggest a telestroke assimilation model (
Proposed model of telestroke assimilation.
Existing telestroke studies support some of our findings and related elements of the assimilation model. On the hub level, Cho et al [
On the spoke level, Rogove et al [
Thus, our findings add to the literature in a number of ways. To our knowledge, this is the first study to focus on telestroke assimilation as a key activity in determining how technology contributes to urgent stroke care performance. Second, we have distinguished between hub- and spoke-level factors as the key organizational antecedents to telestroke assimilation. This is particularly important because most studies have focused on spoke-related factors. Finally, we have leveraged our empirical findings to propose a comprehensive model of telestroke assimilation in hospitals that have already deployed the technology.
An important limitation relates to the scope of this study. We considered the impact of organizational factors on telestroke assimilation, but did not explore policy-related (eg, reimbursement and incentive structures), technology-related (eg, reliability, ease-of-use, broadband connectivity, and level of integration of telestroke with other IT in the hospital), or behavioral factors (such as physician attitudes toward thrombolysis and technology, and local neurologists’ buy-in). Furthermore, we assumed that the patient population characteristics were similar across the spoke hospitals’ service areas. It is also important to note that not all hospitals may have the financial resources to hire a neurologist or a dedicated stroke coordinator (which may explain their reluctance or inability to use telestroke). Our findings draw on a comparative case study of two telestroke networks involving a particular technology. Although a case study design has limited generalizability [
Our study suggests some future research directions. First, researchers can validate and improve the proposed telestroke assimilation model by considering additional factors (eg, policy-related, technological, and behavioral) across different networks. Second, researchers can adapt the model to examine postdeployment utilization of telemedicine and other IT (such as EMR and health information exchanges) in health care organizations. Third, the literature provides several examples of maturity models for IT adoption and assimilation. The term “maturity” relates to the degree of repeatability and optimization of processes, from ad hoc practices, to formally defined steps, to managed result metrics, to active optimization of processes [
EDs in rural hospitals with limited neurological expertise face significant challenges in evaluating patients with stroke symptoms. These hospitals need to either transfer stroke patients to larger regional medical centers or hire local neurologists. Recent telemedicine innovations have enabled rural hospitals to connect virtually to regional medical centers for urgent stroke evaluation. However, many hospitals that have deployed telestroke have not assimilated the technology, that is, they have not integrated it into their regular stroke delivery processes. Consequently, neurologic expertise is not used optimally, opportunities for tPA administration may be lost, and patients are transferred out unnecessarily. Based on a detailed examination of variations in telestroke assimilation across two networks, this exploratory research proposes a telestroke assimilation model that includes specific hub- and spoke-related characteristics that can potentially increase IT assimilation by spokes and lead to improved stroke readiness.
computed tomography
emergency department
emergency medical services
Georgia Regents University
intensive care unit
information technology
Medical University of South Carolina
Remote Evaluation of Acute isCHemic stroke
tissue plasminogen activator
This research was funded by Genentech.
JAS has served as a consultant for Genentech and REACH Health Inc and RJA is a cofounder of REACH Health Inc and has cofounder’s equity.