Abstract
BACKGROUND:
Telestroke provides access to vascular neurology expertise for hospitals lacking stroke coverage, and its use has risen rapidly in the past decade. We aim to characterize consultations, spoke behavior, and the relationship between spoke telestroke utilization (number of telestroke consults per year) and spoke alteplase treatment metrics in an academic telestroke network.
METHODS AND RESULTS:
We analyzed prospectively collected data on all telestroke consults from 2003–2018. Trends in network performance and spoke characteristics were analyzed using generalized estimating equations and Kendall’s τβ nonparametric tests as appropriate. Unadjusted and adjusted linear regression models determined associations between telestroke utilization and treatment metrics. The network included 2 hubs and 43 spokes with 12,803 consults performed during the study period. Network growth overall was +1.8 spokes per year and median duration of spoke participation was 7.9 years. The numbers of consults and alteplase-treated patients increased annually, even after adjusting for the number of spokes in the network (p<0.01 for both). While times from last seen well to spoke emergency department (ED) arrival and to consult request increased, door-to-needle time, time from teleconsult request to callback, and time from teleconsult to alteplase administration all decreased (all p<0.01). With time, the network included more spokes without a Primary Stroke Center designation. In adjusted analyses, for every 10 telestroke consults requested by a spoke, the spoke door-to-needle decreased by 1.8 minutes (p=0.02), number of patients treated with alteplase was an additional 1.7 (p<0.01), and the percent of eligible patients treated with alteplase increased by 8%. (p=0.03)
CONCLUSIONS:
Telestroke network size and utilization increased over time. Increased use of teleconsults was associated with increased and timely use of alteplase. Over time, the delivery of timely emergency care has improved significantly among EDs participating in this telestroke network. Replication of these findings in other networks is warranted.
Keywords: teleconsultation, acute stroke
SUMMARY:
The adoption of telestroke services within this large academic network increased with time. Greater use of consultation by spokes was associated with more eligible patients treated with alteplase in a faster manner. Overall, pre-hospital times have increased with time but the timeliness of care delivery, once a telestroke consult is initiated, has significantly improved. Additional stroke education efforts in the community may be needed to increase awareness of stroke symptoms and available treatment options prompting urgency to travel to the hospital. Further studies are needed to replicate these findings and determine their generalizability to better understand the key factors that drive quality improvement in telestroke care over time.
Although there are evidence-based treatments for acute ischemic strokes such as intravenous thrombolysis with alteplase1, 2 and endovascular thrombectomy3, shortage of stroke neurologists in many geographic regions limits their administration. Forty-five percent of Americans (135.7 million) do not have access to a Primary Stroke Center(PSC) within 60 minutes.4 There are 717 strokes per board-certified vascular neurologist in the U.S.5, and in the absence of their involvement, alteplase is under-utilized by non-specialists due to complication concerns.6 From 2005 to 2007, 63% of the 4,750 U.S. hospitals did not administer alteplase once.7–9
Telestroke consultation brings neurovascular expertise to local community hospitals which lack stroke specialists.10, 11 Telestroke networks commonly consist of an academic12 or non-academic tertiary care hospital serving as the “hub” with a pool of vascular neurology experts available to providers at smaller community hospital “spokes”. Consultants remotely assess thrombolysis eligibility and evaluate for transfer to a Comprehensive Stroke Center(CSC) for possible endovascular therapy. After demonstration of safety of telestroke-guided alteplase administration8, 13, 14 and triaging, telestroke programs expanded rapidly in the US.9
While multiple studies worldwide have shown the value of telestroke,13, 15–22 these have been generally small studies spanning short periods of time, have demonstrated safety more so than efficacy, or have not accounted for spoke-specific characteristics. In these studies, the benefit of process improvement that can occur in a longitudinally affiliated network is not as well-captured. The longitudinal TEMPiS network in Germany has demonstrated improvement in alteplase treatment metrics, but it is unclear whether this was attributable to implementation of telestroke.23 There has only been one study of an established U.S. telestroke network across a decade, the Remote Evaluation for Ischemic Stroke network, consisting of nonteaching hospitals.24 There was no significant difference in rates of alteplase administration by telestroke utilization, however the degree of telestroke utilization was not quantified.
To address this knowledge gap, we studied an academic hub-and-spoke telestroke network over a 15-year period to: 1.) describe trends in telestroke consultations utilization and alteplase treatment, 2.) explore dynamic network changes at the spoke level over time, and 3.) analyze the relationship between degree of spoke telestroke utilization and performance on alteplase delivery after adjustment for hospital-level variables. We hypothesize that alteplase treatment rates will be positively correlated with the degree of telestroke utilization in a longitudinal cohort of telestroke consultations within a large, academic telestroke network.
METHODS:
This retrospective study was approved by the Partners Institutional Review Board. The corresponding author has full access to the study data and takes responsibility for its integrity and analysis. Reasonable requests to access the dataset from qualified research investigators trained in human subject confidentiality may be sent to the corresponding author.
Partners TeleStroke Consultation Network
Our TeleStroke service was founded in 2001 after publication of our results establishing the validity and interrater reliability of the National Institutes of Health Stroke Scale (NIHSS) score administered over telestroke.25, 26 We initiated a 2-year pilot in 2001 between our tertiary stroke center and one small rural island hospital to demonstrate feasibility. In 2003, we launched our full program offering service to other sites. All telestroke consultations begin with a phone call to assess the diagnosis. If the patient is possibly eligible for either thrombolysis or mechanical thrombectomy, the phone call is converted to a real-time high-quality interactive videoconference. No formal recommendation for either therapy is offered without videoconferencing. Our equipment and approach comply with the American Heart Association scientific statement on Telemedicine Quality and Outcomes in Stroke.27 The spoke hospitals function at a level equivalent to acute stroke ready hospitals or PSCs. We require a video evaluation for NIHSS assessment and discussion of the risks and benefits of treatment with the patient or family if alteplase is recommended. Spoke hospitals are encouraged to call a telestroke consultation for all patients considered possible candidates for acute stroke therapies. Vascular neurology and neurocritical care fellows respond to the triage phone call and determine if a full video consult is required. All video consults are conducted by vascular neurology attendings licensed in the state and credentialed at the facility where the patient is located. All consultations by phone or video are documented in the secure online documentation portal we developed and maintain in-house (https://www.massgeneral.org/teleneurology). Consultants view imaging obtained at spokes through an online image-sharing interface. Draft recommendations are faxed immediately to the spoke ED and the final consults are available through a secure online connection to spoke hospitals upon record completion. In 2003, all cases treated with alteplase were transferred to our hospital for admission, but over time we have worked closely with spoke hospitals to develop the capacity to retain uncomplicated alteplase cases.
Consultation Data:
This study was a retrospective analysis of prospectively collected data of telestroke consultations rendered from 2003 to 2018 by two CSC hubs in Boston, Massachusetts to their network of affiliated spoke hospitals. A map of the hubs and spokes was generated using the publicly available Google Maps API. Data were entered by hub vascular neurology attending physicians and extracted from structured fields (free text or hard-coded dropdown options) within the telestroke consultation documentation portal. All consultations were included in the analysis and treated as independent observations to capture trends in utilization and process measures over time.
All the fields in telestroke documentation used in this analysis were standardized except medical comorbidities. Gender and ethnicity were clickable fields while age was calculated using manually entered dates of birth and consult. Any prior history of prespecified comorbidities was ascertained by checkboxes in the portal and by searching free text notes for common keywords or abbreviations indicating history of vascular risk factors and vascular events (hypertension, hyperlipidemia, diabetes, coronary artery disease, atrial fibrillation, carotid disease, cigarette use, prior stroke/transient ischemic attack, intracranial hemorrhage). The final diagnosis determined by the consult physician is captured in the portal by a hard-coded choice of diagnoses. Consult physicians enter time points in the pre-hospital and ED phases of care by typing in designated fields. These yield the following intervals: last seen well (LSW) to ED arrival, LSW to consult request, LSW to alteplase administration, ED arrival to time of consult request, ED arrival to alteplase administration (door-to-needle), consult request to consultant first contact, and consult request to alteplase administration.
Spoke Hospital Characteristics:
During the 15-year study period, forty-three hospitals participated in the network at some point, in a geographic area spanning 242 miles in the north-to-south cardinal direction and 175 miles in the east-to-west direction (Figure 1). Twenty-nine of 43 (68%) of the hospitals were state or nationally certified PSCs in Massachusetts, New Hampshire, or Maine.
Figure 1.
Map of locations of the telestroke spokes (black dots) and hubs
The independent variable of interest was the number of consults requested per spoke per year as a marker of the spoke-hub relationship strength. The primary spoke outcome measures of interest were 1.) median spoke door-to-needle time during the study period, 2.) number of consults treated with alteplase per year at each spoke, and 3.) percentage of alteplase eligible patients ultimately treated at each spoke during the study period. We adjusted for the following factors at the spoke level selected a priori due to the potential impact on alteplase delivery based on prior studies28–30 including hospital bed capacity,12 rurality of the spoke as denoted by the Rural-Urban Commuting Area or RUCA code using 2015 tract definitions,31 academic status,12 PSC certification designation,12 median age of consult patients, percentage of female consult patients, percentage of Caucasian consult patients, median NIHSS, and percentage of consult patients with atrial fibrillation.
Statistical Analysis:
All statistical analyses were performed in SAS 9.4 (Cary, NC). We performed descriptive analysis of all consultations in aggregate within the network. Medians with interquartile ranges and numbers with percentages were presented for continuous and categorical variables, respectively. The rationale for presenting medians was that the distributions of all the continuous variables in this study were non-normal by Kolmogorov-Smirnov testing (p ≤ 0.05). Negative time intervals were classified as missing, and missing values were not imputed. Generalized estimating equations (GEE) were utilized to explore the trends in stroke care over time in the network in terms of the absolute number of telestroke consults requested per year within the network, the number of alteplase cases per year, and the number of transfers per year.32 We defined model parameter significance at a threshold of p ≤ 0.05. To determine if timeliness of care changed over the 15 years of care delivery, we used the Mann-Kendall test (with a null hypothesis of no change over time) to assess for a monotonic trend, a nonparametric test given the non-normal distribution of the time data.33
Next, we described patterns of spoke telestroke utilization, characteristics, and performance over time. The rates of network growth, spoke entry, and spoke exit were calculated as slopes generated from linear regressions modeling the association between the number of spoke hospitals within the network, joining, and exiting the network as a function of calendar year. We divided these into 2 time epochs: 2003–2011, and 2012–2018. Piecewise analysis was performed due to the nonlinearity of the data and an inflection point in spokes exiting the network observed at 2012. Statistical significance of the patterns noted with the transformed data was gauged by employing a spline regression model with 1 knot at the year 2011 and 1 degree supporting our use of linear functions only within each epoch. Spokes characteristics and performance metrics were assessed over time by the Mann-Kendall test.
To assess the relationship between spoke telestroke utilization (consults called per year) and the outcomes of alteplase delivery measures (median door-to-needle times, number of consults treated with alteplase per year, % of eligible patients treated with alteplase), linear regression models were created at the spoke level. A spoke was deemed an outlier and excluded from a regression analysis if the difference in fits (DFFITS) associated with an individual observation, a marker of influence, was greater than 2.34, 35 Adjustment covariates included bed size, rurality (RUCA code > 1), academic status, PSC certification status, median consult age, % consults who are Caucasian, female, and have atrial fibrillation, and median consult NIHSS.
We describe the outcomes of 100 randomly selected patients who underwent telestroke consultation at a spoke and were subsequently transferred to one hub from 2017 to 2018. Chart review was performed to determine final diagnosis, percentage treated with alteplase at the spoke, transfer indication, percentage who underwent endovascular therapy, and neurologic functional outcomes at discharge and at outpatient follow-up 1–3 months after discharge by modified Rankin scale.
RESULTS:
Telestroke Consultations in the Network:
A total of 12,803 telestroke consultations were performed from 2003 to 2018 (Table 1). Patients evaluated by telestroke consult were a median of 66 years old, 66% Caucasian, and 51% female. The proportion of subjects with vascular risk factors were within expected ranges. Nearly two- third of consultations were resolved by phone, while the remaining one-third necessitated live video. On presentation, the median NIHSS was 5 (IQR 2–12). Fifty-one percent of consultations were determined as acute ischemic strokes within 9 hours by the telestroke consultant (Supplementary Figure 1).
Table 1.
Telestroke Consultation Characteristics from 2003–2018
Consultation Characteristics | N (%) or Median (IQR) (N=12,803) |
---|---|
Age (N=7,823) | 66 (47–80) |
Female Sex (N=12,291) | 6,296 (51.2) |
Race/ethnicity | |
Caucasian | 8,455 (66.0) |
African American | 235 (1.8) |
Asian/Pacific Islander | 88 (0.7) |
American Indian/Eskimo | 10 (0.1) |
Other | 1,969 (15.4) |
Unknown | 2,046 (16.0) |
Comorbidities | |
Hypertension | 4,826 (37.7) |
Hyperlipidemia | 1,299 (10.2) |
Diabetes | 2,022 (15.8) |
Coronary artery disease | 1,773 (13.9) |
Atrial fibrillation | 1,773 (13.8) |
Carotid disease | 432 (3.4) |
Cigarette use | 545 (4.3) |
Prior stroke/transient ischemic attack | 1,919 (15.0) |
Intracranial hemorrhage | 736 (5.8) |
Unwitnessed onset (N=11,066) | 3,808 (34.4) |
Modality | |
Phone only | 8,415 (65.7) |
Phone plus video | 4,388 (34.3) |
Systolic blood pressure (N=4,091) | 148 (130–165) |
Antiplatelet use | 1,593 (12.4) |
Anticoagulant use | 320 (2.5) |
NIHSS* (N=3,584) | 5.00 (2–12) |
Significant neurologic deficit expected to result in long term disability | 2,280 (17.8) |
Key Timepoints | |
LSW to symptom discovered, hours (N=11,066) | 0.0 (0.0 to 0.5) |
LSW to ED arrival, hours (N=5,753) | 1.5 (0.8–3.6) |
LSW to request, hours (N=11,185) | 2.3 (1.4–4.5) |
LSW to alteplase, hours (N=1,902) | 2.3 (1.8–2.9) |
ED arrival to consult request, minutes (N=6,340) | 43.2 (24.0–84.0) |
ED arrival to alteplase, minutes (N=1,089) | 73.0 (55.0–100.0) |
Consult request to start of consult, minutes (N=12,508) | 1.0 (0.0–3.0) |
Start of consult to alteplase, minutes (N=1,813) | 39.0 (30.0–54.0) |
Alteplase eligible (N=11,163) | 2,821 (25.3) |
Alteplase administered (N=11,162) | 2,105 (18.9) |
Potentially eligible for endovascular therapy, (N=9,859) | 996 (10.1) |
Potentially eligible for hemicraniectomy (N=10,162) | 218 (2.1) |
Disposition | |
Initiate transfer | 4,642 (36.3) |
Remain at referring hospital | 7,725 (60.3) |
Unable to determine | 436 (3.4) |
One-quarter of consultations were deemed eligible for alteplase and 19% resulted in alteplase administration (75% of all eligible patients). Reasons for not treating with alteplase were not mutually exclusive including: mild symptoms (61%), rapid symptom improvement (33%), delay in patient arrival (31%), and delay in stroke diagnosis (20%) (Supplementary Table 1). About 10 percent of consultations were eligible for endovascular thrombectomy and 2.1 percent for hemicraniectomy (Table 1). Nearly 36% of consultations were triaged to transfer to another hospital, which could include the hub or another nearby facility. Eighty-one percent of alteplase- treated patients were subsequently transferred.
From 2003 to 2018, the annual number of telestroke consultations requested increased from 64 to 1,878 at a rate of 120 consultations/year (Figure 2; p ≤ 0.0001), after adjusting for the number of spokes within network during each calendar year. The annual number of consultations resulting in alteplase administration increased at a rate of 15 per year, after adjustment for the number of spokes associated with the network each year (p ≤ 0.0001). After adjusting for the number of spokes and consults requested, the number of patients transferred to another hospital for advanced care decreased over time by 27 patients per year (p=0.044).
Figure 2.
Number of telestroke consults requested per year (black) and the number of telestroke consultations resulting in alteplase administration (gray) (both significant at p ≤ 0.0001 by GEE time trends). Note that the 2018 values are annualized based on activity through 7/25/2018.
The median times of action reported in Table 1 are organized chronologically. Time from last seen well to ED arrival was 1.5 hours (IQR 0.8–3.6). Median time from ED arrival to telestroke consultation request was 43.2 min (IQR 24.0–84.0), with a median of 1 min (IQR 0.0 to 3.0) for consultation initiation and 39.0 minutes (IQR 30.0 to 54.0) from consult start to alteplase administration. Median door-to-needle time was 73.0 minutes (IQR 55.0–100.0).
We examined the directionality and significance of the timeliness of care delivery in the pre- hospital and ED phase of care during the study period (See Figure 3). All pre-hospital time metrics anchored to the LSW time significant worsened over time, except time from LSW to alteplase administration which was flat (Kτ coefficient −0.028; p-value 0.076). This was driven by the lack of improvement in the delay between LSW and ED arrival. In contrast, nearly all the time metrics in the ED phase of care improved, including telestroke consult request to callback, telestroke consult to alteplase administration, and door-to-needle time. The time from ED arrival to requesting a telestroke consult, however, worsened significantly over time (Kτ coefficient 0.023; p-value 0.009).
Figure 3.
Median pre-hospital (top) and Emergency Department (bottom) time intervals with trend lines. Kendall Tau correlation coefficients and associated p-values for each time construct are presented in the adjacent box. (Abbreviations: last seen well (LSW), emergency department (ED)
Spokes in the Network:
We modeled spoke participation in the network over time. Spokes participated for a mean of 7.9 years (range 2–9). The overall rate of growth of the network from 2003 to 2018 was an additional1.8 spokes per year (R2 0.5985). Network affiliation trend over time was not uniformly upwards.
The network grew at a predicted rate of 4.33 spokes per year through 2011 and shrank at a predicted rate of 1.46 spokes per year from 2012 to 2018 (spline model p-value: <0.0001; adjusted R2 0.9608). (Figure 4) The rate of spoke entry did not change from 2003 to 2018 (spline model p-value: 0.2001; adjusted R2 0.099). The exit rate from 2003 to 2011 was 0, but the exit rate increased by an additional 0.46 spokes/year from 2012 to 2018 (spline model p-value: 0.0095; adjusted R2 0.4367). These results demonstrate robust spoke entry into the network through 2011 and more exits from 2012 onwards.
Figure 4:
Number of spokes entering (triangles), exiting (stars), and staying within the network (circles) each calendar year with overlying fitted, one-knot spline models
Table 2 describes characteristics of participating spoke hospitals. Spoke hospitals were often medium-sized community hospitals (mean staffed beds 180, SD 153), while the hubs were major academic teaching hospitals (763 and 1,011 beds). Nearly 26% of spokes were in rural regions and half were academic. Twenty-nine spokes of 43 (67%) of the hospitals were certified PSCs in Massachusetts, New Hampshire, or Maine. An average of 39.5 consultations were called per year by each spoke and a mean of 66% of eligible patients were treated with alteplase. Over time, the percentage of spokes with a PSC designation decreased, indicating that more non-PSC spokes were being incorporated into the network. Spokes in the network tended to call consults for younger and more racially diverse patients over time. Telestroke consultations increased with time. Greater numbers of patients were treated with alteplase per year at each spoke and within the overall spoke median door-to-needle time. There was no change in the percent of eligible patients treated with alteplase over time.
Table 2.
Spoke Characteristics from 2003 to 2018
Spoke Characteristic | Mean (SD) unless otherwise specified (N = 43) | Kτ Coefficient from 2003–2018 | P-Value |
---|---|---|---|
Institution Characteristics | |||
Number of staffed bed | 180.1 (152.6) | 0.33* | 0.08 |
Rural, N (%) | 11 (25.6) | −0.08 | 0.65 |
Teaching, N (%) | 18 (50.0) | 0.24 | 0.21 |
Certified Primary Stroke Center, N (%) | 29 (67.4) | −0.51 | <0.01 |
Demographics | |||
Median age, years | 66.3 (5.4) | −0.49† | <0.01 |
Percent of female patient consultations | 50.1 (4.5) | −0.24† | 0.19 |
Percent of Caucasian patient consultations | 71.3 (11.0) | −0.76† | <0.01 |
Case Mix Indicators | |||
Median NIHSS | 5.9 (2.3) | 0.25† | 0.17 |
Percent of patients with atrial fibrillation | 14.2 (3.3) | −0.21† | 0.24 |
Stroke Metrics | |||
Telestroke consults per year | 39.5 (34.3) | 0.85† | <0.01 |
Median door-to-needle time, minutes | 75.7 (16.1) | 0.38‡ | 0.04 |
Number of patients treated with IV-tPA per year | 10.4 (15.9) | 0.91† | <0.01 |
Percent of eligible patients treated with alteplase | 66.2 (20.8) | 0.11† | 0.56 |
Abbreviations: NIHSS or National Institutes of Health Stroke Scale only recorded if video used; ED, emergency department; IV, intravenous; LSW, last seen well; N, number with complete data for each variable if incomplete
% Staffed beds > 100
% spokes with variable > than median of the variable among all spokes
% spokes with door-to-needle time < median door-to-needle time of all spokes
Univariate and Multivariable Modeling
Unadjusted associations of spoke median door-to-needle times, spoke median time from ED arrival to telestroke consultation, number of years each spoke participated in the telestroke network, and number of spoke telestroke consultation are depicted in Figure 5. The bubble plot illustrates that door-to-needle times was lower among spokes that consulted more frequently each year (p<0.01) and consulted more quickly upon ED arrival (p=0.04). There was no association between number of years in network and door-to-needle times.
Figure 5.
Bubble plot depicting spoke hospitals as a function of median time from ED arrival to consult request and median door-to-needle time (beta coefficient 0.24; p=0.04). The size of each bubble is proportional to the number of consults requested per spoke per year and is associated with median door-to-needle time (beta coefficient −0.2; p<0.01). The color of the bubble reflects the years of participation in the telestroke network (Yellow: ≤ 4 years, Blue: > 4 to ≤ 8 years, Red: > 8 to ≤ 12 years, Green: >12 years)
In Table 3, we model the relationships between the number of consults called per year by a spoke (a marker of the strength of the hub-spoke relationship) and alteplase delivery performance measures before and after adjustment for hospital-level variables. One spoke exhibited a difference in fits > 2 and was thus excluded from the analysis. In unadjusted and adjusted linear regression analyses, the number of consults called per year significantly predicted median door-to-needle time. For every 10 consults called by a spoke per year in the adjusted analysis, the median spoke door-to-needle time decreased by 1.8 minutes (p=0.02). Similarly, the number of consults called per year by a spoke was significantly associated with number treated with alteplase per year in unadjusted and adjusted analyses. For every 10 telestroke consults called per year by a spoke, an additional 1.7 of the patients presented in these consults were ultimately treated with alteplase. Finally, for every 10 consults called per year by a spoke, the percent of eligible patients treated with alteplase increased by 8.3.
Table 3.
Linear regression models of consults called per year by a spoke as a predictor of a spoke’s alteplase treatment outcomes of interest with and without adjustment for spoke-level characteristics.
Spoke variable | Median door-to-needle time (minutes) | Number treated with alteplase per year | Percent of eligible patients treated with alteplase | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Unadjusted | Adjusted | Unadjusted | Adjusted | Unadjusted | Adjusted | |||||||
β | p | β | p | β | p | β | p | β | p | β | p | |
Consults called per year (units of 10) | −1.57 | 0.02 | −1.76 | 0.02 | 2.00 | <0.01 | 1.68 | <0.01 | 13.39 | <0.01 | 8.27 | <0.01 |
Number of staffed beds (units of 10) | 0.04 | 0.80 | 0.21 | 0.17 | 0.14 | <0.01 | 0.06 | 0.05 | 0.29 | 0.75 | −0.04 | 0.94 |
Rurality | −5.84 | 0.29 | −12.12 | 0.05 | −2.57 | 0.19 | 0.10 | 0.95 | 11.30 | 0.68 | 33.92 | 0.13 |
Teaching status | 5.49 | 0.31 | 0.72 | 0.87 | 1.01 | 0.52 | 0.26 | 0.77 | 7.20 | 0.80 | −5.15 | 0.76 |
Primary Stroke Center certification | 5.74 | 0.27 | −1.20 | 0.82 | 1.39 | 0.42 | 0.50 | 0.63 | 12.38 | 0.63 | −0.61 | 0.98 |
Median age | 0.10 | 0.82 | 0.02 | 0.97 | −0.01 | 0.96 | −0.02 | 0.84 | −0.21 | 0.92 | −0.07 | 0.97 |
Percent Caucasian | 0.12 | 0.59 | −0.64 | 0.03 | 0.04 | 0.59 | 0.04 | 0.54 | −1.69 | 0.13 | −1.91 | 0.07 |
Percent female | −0.02 | 0.97 | 0.52 | 0.33 | 0.18 | 0.34 | −0.06 | 0.60 | 2.00 | 0.46 | −0.19 | 0.92 |
Percent history of atrial fibrillation | 1.15 | 0.12 | 1.24 | 0.19 | 0.00 | 1.00 | −0.13 | 0.61 | 5.14 | 0.16 | 2.59 | 0.46 |
Median NIHSS | 3.10 | 0.02 | 0.48 | 0.77 | 0.02 | 0.97 | 0.14 | 0.64 | −3.72 | 0.48 | 3.50 | 0.56 |
Abbreviations: Beta-coefficient (β); P-value (p); National Institute of Health Stroke Severity (NIHSS)
Outcomes Upon Transfer:
In a random sample of 100 patients transferred from spoke to hub, all patients had MRI- confirmed ischemic strokes. Twenty-three percent received alteplase at the spoke. Reasons for transfer included possible endovascular therapy (53%) and further management (47%). Among those transferred for thrombectomy, sixty-six percent underwent the procedure. Mean mRS was 3.2 at discharge and 2.9 at follow-up at clinic 1–3 months after stroke.
DISCUSSION:
To our knowledge, this is the largest study to date in the literature describing in detail the patient characteristics, longitudinal composition, spoke behavior, and performance metrics of a large academic telestroke network since its inception. While the pre-hospital times for patients undergoing consultations in our network increased with time, the ED time metrics decreased with time. With time, spokes participated in more consults, treated more eligible consult patients with tPA, and had lower door-to-needle times. There was a positive association between telestroke consult utilization and delivery of alteplase metrics which persisted after adjustment for various other spoke-level factors.
There are certain similarities and differences between our telestroke network and others nationwide. In a survey of programs from 27 states, telestroke programs were operational for a median of 2.4 years with an average 7.6 spokes in 2009.9 Our network is an outlier in terms of longevity and size. The trend in door-to-needle times in our network mirror the national trend observed in the Paul Coverdell National Acute Stroke Program.28 Our network’s median door- to-needle time decreased from 75 minutes in 2008 and to 63 minutes in 2017. Similarly, in the Coverdell program representing 496,336 acute ischemic stroke admissions in the U.S., this metric decreased from 79 minutes in 2008 to 51 minutes in 2017.
We observed increasing times from LSW and ED arrival, and ED arrival to telestroke consult request, in striking contrast to decreasing times to treatment upon consult request. While improvements in post-consultation times reflect years of practice and standardization of approaches to thrombolysis and thrombectomy management, the lengthening of pre-consultation time metrics deserves attention. In the pre-hospital setting, it may reflect a historical lack of public awareness of symptom recognition36 and limited Emergency Medical Services (EMS) resources.37 Since our network incorporated more non-PSC spokes over time, we performed a post-hoc analysis using the Kruskal-Wallis test that demonstrated the median time from LSW to ED arrival was nonsignificantly higher among non-PSC versus PSC spokes (1.49 versus 1.35 hours, p=0.34). The prolonged time from ED arrival to consultation could be paradoxically worse due to the broadening of inclusion criteria for consultation and a more diligent effort to identify eligible patients for alteplase by ED physician. They may be taking additional steps to validate the time LSW and carefully evaluating patients with subtle signs of stroke, resulting in greater number of patients being treated but also longer times from ED arrival to consultation.2, 38–40 There may be limited resources to screen greater numbers of patients eligible for treatment in recent years given the lengthening of the therapeutic window for both acute therapies. Further analysis of this finding in large independent cohorts is necessary.
As the number of consultations increased per year by spoke, so did the number of patients treated with alteplase and the percent of alteplase-eligible patients treated. The adjusted spoke door-to- needle time decreased. Spokes requesting more consults may have lower thresholds for suspecting ischemic stroke and thus a higher sensitivity of detection. Given the complex processes required to identify patients, complete neuroimaging, and initiate alteplase, it is likely that the training and repeated practice experienced by the spoke providers during telestroke consults improves their skills and reduces unnecessary delays. This is consistent with another study from our network, finding that more frequent contact between a spoke and hub was associated with faster alteplase delivery.19 In contrast, a study of a non-academic telestroke network in Georgia noted that participation in the telestroke network was not associated with the rate of tPA use.24 We believe that nominally belonging to a telestroke network does not heavily impact clinical processes, but rather that, in a dose-dependent manner, the degree of utilization of the telestroke service is associated with improved delivery of alteplase facilitated by bidirectional feedback through multiple interactions between hub and spoke. The finding of longer delays in ED arrival to consult request underscores the need for robust two-way feedback between hub and spoke hospitals and frequent education about the indications for and urgency of making the telestroke consultation. These delays may reflect a prioritization of specificity over sensitivity by spoke hospital personnel. One way to address this may be to decrease the amount of information required from the spoke provider to initiate a consultation. While there may be unmeasured confounders, the fact that PSC certification status alone did not account for these process improvements suggests the effect is independent of other quality improvement efforts known to be associated with shorter door-to-needle times.
There was overall growth in the network size, indicating that the network was an attractive option for hospitals without onsite stroke expertise. While spokes entered the network steadily from 2012 to 2018, the number of spokes exiting the network increased from 2012 to 2018 due to market shifts. Thirty-two percent of exiting hospitals merged with another academic center to enact broader patient retention policies and transition to a more closed system to remain competitive. Forty percent of exiting spokes switched to another remote telestroke provider. The remainder of spokes either closed or their telestroke status was unknown. The number of patients who were transferred out of the spoke while in network decreased with time, suggesting a level of confidence and competence to manage certain stroke patients developing among the spokes.
The adoption of telestroke, implementation of stroke systems of care,41 and use of thrombectomy are 3 of the most powerful forces that have shaped stroke care in the past decade and are deeply interconnected. We have previously shown that telestroke is the most frequently cited clinical application for telemedicine use (58% of U.S. ED’s use telemedicine of which, 77% use it for stroke/neurology).42 Our data suggest that deepening the strength of the hub-spoke connection improves care delivery, demonstrating that quality can be transmitted in a retrograde manner via patient-based video interactions from academic medical centers to community hospitals. These complex interactions across hospital node pairs require novel analytic tools in network science.43 As video and broadband wireless bandwidth becomes more ubiquitous, it is likely that video- enhanced 911 dispatch44 and pre-hospital medical control will become commonplace, supporting better triage to the most appropriate facility. Applications can include improved selection of cases with suspected large vessel occlusion for triage to a thrombectomy-capable site, suspected ST elevation MI to a PCI-capable site, and field triage for mass casualty events for allocating patients to a number of hospitals simultaneously. Increased use of telementoring bedside procedural interventions is emerging, as are robotic devices that can enable an operator to perform thrombectomy at a distance. Lastly, machine learning and artificial intelligence decision-support tools may provide instantaneous guidance for emergency physicians to support the use of thrombolysis in acute stroke.
This study has several limitations. The generalizability of results from a mature, academic hub- and-spoke network to other telestroke models is unknown. Nevertheless, our study is a proof-of- concept that under certain constraints, increased telestroke utilization can improve quality of care. Selection bias is possible since it is unknown if all consults were recorded in the documentation portal. There is likely variability in the threshold for calling consults by provider and site. There are presumably unmeasured confounders at the patient and hospital level. Consultation data were collected in real time by busy telestroke clinicians focused on assessing and treating patients as rapidly as safely possible, and not subject to external audit. In this environment, it is likely that some data elements will be inaccurate or incomplete. Race/ethnicity were often reported as other or unknown, limiting our ability to understand the relationship of this variable with our outcomes. We are unable to comment upon ultimate reasons for not treating with alteplase, as this was not a mandatory field. There were several variables with sizable missingness (i.e. NIHSS) and their contributions may not be fully estimated. There was no comparison group of hospitals without telestroke, but the scope of this study was to investigate the in-network trends in alteplase treatment as a function of degree of telestroke utilization. We do not have data about patients treated with alteplase at our spokes outside the consultation context. Our findings may justify a randomized clinical trial of hospitals within and outside of a telestroke network. Lastly, the Kendall tau correlation coefficient only detects the presence of monotonic temporal relationships, nevertheless, it was appropriate to utilize this test due to the lack of normality of the temporal metrics.
Supplementary Material
What is Known:
The neurologic effects of an acute ischemic stroke can be mitigated by acute, evidence-based therapies such as intravenous alteplase and mechanical thrombectomy.
Timely delivery of these therapies is challenging in locations lacking stroke expertise.
Alteplase administered by providers at hospitals without on-site stroke expertise under the guidance of a telestroke consultant is safe.
What the Study Adds:
The adoption of telestroke services in this network of academic hubs and affiliated spokes increased over time.
Treatment times at the spoke hospitals improved significantly over the years since the network’s inception.
Greater use of telestroke consultation services by spokes was associated with more eligible patients being treated with alteplase quickly.
Acknowledgments:
We are grateful to the patients and staff at the participating telestroke spoke hospitals and for our telestroke network nurse liaison (Cynthia Whitney, RN).
Sources of Funding:
There was no dedicated funding source for this study. Effort by Dr. Sharma (StrokeNet research Fellow), Dr. Viswanathan (Director of Education) and Dr. Schwamm (Principal Investigator for the Northeast Regional Coordinating Center) was supported by funding from the NINDS StrokeNet Network (NINDS U10 NS086729). Dr. Zachrison’s effort was supported by AHRQ K08HS024561. Dr. Anderson is supported by NINDS K23NS086873 and R01NS103924, and American Heart Association 18SFRN34110082.
Disclosures:
Dr. Schwamm is principal investigator of an investigator-initiated study of extended-window intravenous thrombolysis funded by the National Institutes of Neurological Disorders and Stroke (clinicaltrials.gov/show/NCT01282242) for which Genentech provided alteplase free of charge to Massachusetts General Hospital as well as supplemental per-patient payments to participating sites; serves as a stroke systems consultant to the Massachusetts Department of Public Health; serves as a scientific consultant to LifeImage regarding user interface design and usability, and trial design and conduct to Penumbra (data and safety monitoring committee, Separator 3D and MIND trials), Genentech (Steering committee TIMELESS trial) and Medtronic (Victory AF and Stroke AF trials); and serves as a continuing medical education symposium organizer or lecturer on topics in stroke reperfusion therapy [Medtronic, Boehringer Ingelheim]. Dr. Anderson provides medical consulting for ApoPharma, Inc.
REFERENCES:
- 1.National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med. 1995. [DOI] [PubMed] [Google Scholar]
- 2.Hacke W, Kaste M, Bluhmki E, Brozman M, Dávalos A, Guidetti D, Larrue V, Lees KR, Medeghri Z, Machnig T, Schneider D, von Kummer R, Wahlgren N and Toni D. Thrombolysis with Alteplase 3 to 4.5 Hours after Acute Ischemic Stroke. New England Journal of Medicine. 2008. [DOI] [PubMed] [Google Scholar]
- 3.Goyal M, Demchuk AM, Menon BK, Eesa M, Rempel JL, Thornton J, Roy D, Jovin TG, Willinsky RA, Sapkota BL, Dowlatshahi D, Frei DF, Kamal NR, Montanera WJ, Poppe AY, Ryckborst KJ, Silver FL, Shuaib A, Tampieri D, Williams D, Bang OY, Baxter BW, Burns PA, Choe H, Heo J-H, Holmstedt CA, Jankowitz B, Kelly M, Linares G, Mandzia JL, Shankar J, Sohn S-I, Swartz RH, Barber PA, Coutts SB, Smith EE, Morrish WF, Weill A, Subramaniam S, Mitha AP, Wong JH, Lowerison MW, Sajobi TT and Hill MD. Randomized Assessment of Rapid Endovascular Treatment of Ischemic Stroke. New England Journal of Medicine. 2015. [DOI] [PubMed] [Google Scholar]
- 4.Albright KC, Branas CC, Meyer BC, Matherne-Meyer DE, Zivin JA, Lyden PD and Carr BG. Acute cerebrovascular care in emergency stroke systems. Archives of Neurology. 2010. [DOI] [PubMed] [Google Scholar]
- 5.Leira EC, Kaskie B, Froehler MT and Adams HP. The growing shortage of vascular neurologists in the era of health reform: Planning is brain! Stroke. 2013. [DOI] [PubMed] [Google Scholar]
- 6.Micieli G, Marcheselli S and Tosi PA. Safety and efficacy of alteplase in the treatment of acute ischemic stroke. Vascular Health and Risk Management. 2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kleindorfer D, Xu Y, Moomaw CJ, Khatri P, Adeoye O and Hornung R. US geographic distribution of rt-PA utilization by hospital for acute ischemic stroke. Stroke. 2009. [DOI] [PubMed] [Google Scholar]
- 8.Pervez MA, Silva G, Masrur S, Betensky RA, Furie KL, Hidalgo R, Lima F, Rosenthal ES, Rost N, Viswanathan A and Schwamm LH. Remote supervision of IV-tPA for acute ischemic stroke by telemedicine or telephone before transfer to a regional stroke center is feasible and safe. Stroke. 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Silva GS, Farrell S, Shandra E, Viswanathan A and Schwamm LH. The status of telestroke in the united states: A survey of currently active stroke telemedicine programs. Stroke. 2012. [DOI] [PubMed] [Google Scholar]
- 10.Schwamm LH, Holloway RG, Amarenco P, Audebert HJ, Bakas T, Chumbler NR, Handschu R, Jauch EC, Knight IV WA, Levine SR, Mayberg M, Meyer BC, Meyers PM, Skalabrin E and Wechsler LR. A review of the evidence for the use of telemedicine within stroke systems of care: A scientific statement from the American heart association/American stroke association. Stroke. 2009. [DOI] [PubMed] [Google Scholar]
- 11.Schwamm LH, Audebert HJ, Amarenco P, Chumbler NR, Frankel MR, George MG, Gorelick PB, Horton KB, Kaste M, Lackland DT, Levine SR, Meyer BC, Meyers PM, Patterson V, Stranne SK and White CJ. Recommendations for the implementation of telemedicine within stroke systems of care: A policy statement from the American heart association. Stroke. 2009. [DOI] [PubMed] [Google Scholar]
- 12.American Hospital Directory, Inc. American Hospital Directory. 2019. www.ahd.com. Accessed January 6, 2019.
- 13.Sauser-Zachrison K. Safe and Effective Implementation of Telestroke in a US Community Hospital Setting. The Permanente Journal. 2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Schwab S, Vatankhah B, Kukla C, Hauchwitz M, Bogdahn U, Furst A, Audebert HJ, Horn M and Group TE. Long-term outcome after thrombolysis in telemedical stroke care. Neurology. 2007;69:898–903. [DOI] [PubMed] [Google Scholar]
- 15.Agarwal S, Day DJ, Sibson L, Barry PJ, Collas D, Metcalf K, Cotter PE, Guyler P, O’Brien EW, O’Brien A, O’Kane D, Owusu-Agyei P, Phillips P, Shekhar R and Warburton EA. Thrombolysis delivery by a regional telestroke network--experience from the U.K. National Health Service. Journal of the American Heart Association. 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Demaerschalk BM BE, Barrett KM, Gamble DM, Sonchik S, Comer MM, Wieser J, Hentz JG, Fitz-Patrick D, Chang YH. Comparison of Stroke Outcomes of Hub and Spoke Hospital Treated Patients in Mayo Clinic Telestroke Program. J Stroke Cerebrovasc Dis. 2018;18. [DOI] [PubMed] [Google Scholar]
- 17.Harvey J, Al Kasab S, Almallouhi E, Guerrero WR, Debenham E, Turner N, Aysse P and Holmstedt CA. Door to needle time and functional outcome for mild ischemic stroke over telestroke Journal of Telemedicine and Telecare; 2018. [DOI] [PubMed] [Google Scholar]
- 18.Baratloo A, Rahimpour L, Abushouk AI, Safari S, Lee CW and Abdalvand A. Effects of Telestroke on Thrombolysis Times and Outcomes: A Meta-analysis. Prehospital Emergency Care. 2018. [DOI] [PubMed] [Google Scholar]
- 19.Moreno A, Schwamm LH, Siddiqui KA, Viswanathan A, Whitney C, Rost N and Zachrison KS. Frequent Hub-Spoke Contact Is Associated with Improved Spoke Hospital Performance: Results from the Massachusetts General Hospital Telestroke Network. Telemedicine journal and e-health : the official journal of the American Telemedicine Association. 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Nguyen-Huynh MN, Klingman JG, Avins AL, Rao VA, Eaton A, Bhopale S, Kim AC, Morehouse JW and Flint AC. Novel telestroke program improves thrombolysis for acute stroke across 21 hospitals of an integrated healthcare system. Stroke. 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Zhao G, Huang H and Yang F. The progress of telestroke in China. Stroke and Vascular Neurology. 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Demaerschalk BM, Bobrow BJ, Raman R, Kiernan TEJ, Aguilar MI, Ingall TJ, Dodick DW, Ward MP, Richemont PC, Brazdys K, Koch TC, Miley ML, Hoffman Snyder CR, Corday DA and Meyer BC. Stroke team remote evaluation using a digital observation camera in arizona: The initial mayo clinic experience trial. Stroke. 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Muller-Barna P, Hubert GJ, Boy S, Bogdahn U, Wiedmann S, Heuschmann PU and Audebert HJ. TeleStroke units serving as a model of care in rural areas: 10-year experience of the TeleMedical project for integrative stroke care. Stroke. 2014;45:2739–44. [DOI] [PubMed] [Google Scholar]
- 24.Zhang D, Shi L, Ido MS, Green D, Li Y, Su D, Hess D. Impact of Participation in a Telestroke Network on Clinical Outcomes Evidence From the Georgia Coverdell Acute Stroke Registry. Circulation: Cardiovascular Quality and Outcomes. 2019;12:1–11. [DOI] [PubMed] [Google Scholar]
- 25.Shafqat S, Kvedar JC, Guanci MM, Chang Y and Schwamm LH. Role for telemedicine in acute stroke: Feasibility and reliability of remote administration of the NIH Stroke Scale. Stroke. 1999. [DOI] [PubMed] [Google Scholar]
- 26.Schwamm LH, Rosenthal ES, Hirshberg A, Schaefer PW, Little EA, Kvedar JC, Petkovska I, Koroshetz WJ and Levine SR. Virtual TeleStroke support for the emergency department evaluation of acute stroke. Academic Emergency Medicine. 2004. [DOI] [PubMed] [Google Scholar]
- 27.Wechsler LR, Demaerschalk BM, Schwamm LH, Adeoye OM, Audebert HJ, Fanale CV, Hess DC, Majersik JJ, Nystrom KV, Reeves MJ, Rosamond WD and Switzer JA. Telemedicine quality and outcomes in stroke: A scientific statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2017. [DOI] [PubMed] [Google Scholar]
- 28.Tong X, Wiltz JL, George MG, Odom EC, Coleman King SM, Chang T, Yin X and Merritt RK. A Decade of Improvement in Door-to-Needle Time Among Acute Ischemic Stroke Patients, 2008 to 2017. Circulation Cardiovascular quality and outcomes. 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Gonzales S, Mullen MT, Skolarus L, Thibault DP, Udoeyo U and Willis AW. Progressive rural-urban disparity in acute stroke care. Neurology. 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Messé SR, Khatri P, Reeves MJ, Smith EE, Saver JL, Bhatt DL, Grau-Sepulveda MV, Cox M, Peterson ED, Fonarow GC and Schwamm LH. Why are acute ischemic stroke patients not receiving IV tPA?: Results from a national registry. Neurology. 2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Federal Financial Institutions Examination Council. Federal Financial Institutions Examination Council Geocoding/Mapping System. 2015. https://geomap.ffiec.gov/FFIECGeocMap/GeocodeMap1.aspx. Accessed January 2, 2019.
- 32.Wang M, Wang M, Wang and Ming. Generalized Estimating Equations in Longitudinal Data Analysis: A Review and Recent Developments. Advances in Statistics. 2014. [Google Scholar]
- 33.Hirsch RM and Slack JR. A Nonparametric Trend Test for Seasonal Data With Serial Dependence. Water Resources Research. 1984. [Google Scholar]
- 34.Baltagi B. Econometrics. 2nd ed; 1999. [Google Scholar]
- 35.SAS Institute Inc. Influence Diagnostics The REG Procedure. 1999. https://www.sfu.ca/sasdoc/sashtml/stat/chap55/sect38.htm. Accessed January 9, 2019. [Google Scholar]
- 36.Kleindorfer D, Khoury J, Broderick JP, Rademacher E, Woo D, Flaherty ML, Alwell K, Moomaw CJ, Schneider A, Pancioli A, Miller R and Kissela BM. Temporal trends in public awareness of stroke :Warning signs, risk factors, and treatment. Stroke. 2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Porteous GH, Corry MD and Smith WS. Emergency medical services dispatcher identification of stroke and transient ischemic attack. Prehospital Emergency Care. 1999. [DOI] [PubMed] [Google Scholar]
- 38.Powers WJ, Rabinstein AA, Ackerson T, Adeoye OM, Bambakidis NC, Becker K, Biller J, Brown M, Demaerschalk BM, Hoh B, Jauch EC, Kidwell CS, Leslie-Mazwi TM, Ovbiagele B, Scott PA, Sheth KN, Southerland AM, Summers DV and Tirschwell DL. 2018 Guidelines for the Early Management of Patients With Acute Ischemic Stroke: A Guideline for Healthcare Professionals From the American Heart Association/American Stroke Association. Stroke. 2018. [Google Scholar]
- 39.Nogueira RG, Jadhav AP, Haussen DC, Bonafe A, Budzik RF, Bhuva P, Yavagal DR, Ribo M, Cognard C, Hanel RA, Sila CA, Hassan AE, Millan M, Levy EI, Mitchell P, Chen M, English JD, Shah QA, Silver FL, Pereira VM, Mehta BP, Baxter BW, Abraham MG, Cardona P, Veznedaroglu E, Hellinger FR, Feng L, Kirmani JF, Lopes DK, Jankowitz BT, Frankel MR, Costalat V, Vora NA, Yoo AJ, Malik AM, Furlan AJ, Rubiera M, Aghaebrahim A, Olivot J-M, Tekle WG, Shields R, Graves T, Lewis RJ, Smith WS, Liebeskind DS, Saver JL and Jovin TG. Thrombectomy 6 to 24 Hours after Stroke with a Mismatch between Deficit and Infarct. New England Journal of Medicine. 2017. [DOI] [PubMed] [Google Scholar]
- 40.Albers GW, Marks MP, Kemp S, Christensen S, Tsai JP, Ortega-Gutierrez S, McTaggart RA, Torbey MT, Kim-Tenser M, Leslie-Mazwi T, Sarraj A, Kasner SE, Ansari SA, Yeatts SD, Hamilton S, Mlynash M, Heit JJ, Zaharchuk G, Kim S, Carrozzella J, Palesch YY, Demchuk AM, Bammer R, Lavori PW, Broderick JP and Lansberg MG. Thrombectomy for Stroke at 6 to 16 Hours with Selection by Perfusion Imaging. New England Journal of Medicine. 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Adeoye O, Nystrom KV, Yavagal DR, Luciano J, Nogueira RG, Zorowitz RD, Khalessi AA, Bushnell C, Barsan WG, Panagos P, Alberts MJ, Tiner AC, Schwamm LH and Jauch EC. Recommendations for the Establishment of Stroke Systems of Care: A 2019 Update. Stroke. 2019;50:e187–e210. [DOI] [PubMed] [Google Scholar]
- 42.Zachrison KS, Boggs KM, E MH, Espinola JA and Camargo CA. A national survey of telemedicine use by US emergency departments. J Telemed Telecare. 2018:1357633X18816112. [DOI] [PubMed] [Google Scholar]
- 43.Zachrison KS, Dhand A, Schwamm LH and Onnela JP. A Network Approach to Stroke Systems of Care. Circ Cardiovasc Qual Outcomes. 2019;12:e005526. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Champagne-Langabeer T, Langabeer JR, Roberts KE, Gross JS, Gleisberg GR, Gonzalez MG and Persse D. Telehealth Impact on Primary Care Related Ambulance Transports. Prehosp Emerg Care. 2019;23:712–717. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.