Inpatient Teleneurology Follow-up Has Comparable Outcomes to In-Person Neurology Follow-up

Abstract

Background and Objectives

Community emergency departments often transfer patients for lack of neurology coverage, potentially burdening patients and accepting facilities. Telestroke improves access to acute stroke care, but there is a lack of data on inpatient teleneurology and telestroke care.

Methods

From our prospective telestroke registry, we retrospectively reviewed 3702 consecutive patients who were seen via telestroke between September 2015 and December 2018. Patients who required transfer after initial telestroke evaluation or who were kept at hospitals without consistent neurology coverage were excluded from analysis. We compared baseline demographics, clinical characteristics, and hospital outcomes in patients who were subsequently followed remotely by a teleneurology neurohospitalist and those followed in person by a neurohospitalist.

Results

There were 447 (23%) patients followed by a teleneurology neurohospitalist and 1459 (77%) patients followed in person by a neurohospitalist. Both groups presented with similar stroke severity. In multivariate analysis, there were no significant differences in discharge disposition, stroke readmission rates, or 90-day modified Rankin Scale (mRS) scores. Length of stay was shorter with teleneurology follow-up. In the subgroup of patients who received tissue plasminogen activator, patients showed no differences in outcomes and had similar complication rates. Teleneurology follow-up resulted in a 3% transfer rate for higher level of care after admission. There remained no difference in outcomes in a subanalysis without Comprehensive Stroke Centers. A higher proportion of non-Hispanic Black patients and a lower proportion of Hispanic patients in the teleneurology follow-up group were possibly due to spoke location demographics.

Discussion

Teleneurology follow-up resulted in comparable outcomes to in-person neurology follow-up, with few transfers after admission. For select neurology and ischemic stroke patients, teleneurology follow-up provides an alternative to transfer for hospitals lacking neurology coverage.

Geographic disparity in acute stroke care in the United States has gradually improved as the result of systematic implementation of telestroke networks. These telemedicine stroke evaluations have expanded access to stroke specialists, increased utilization of IV tissue plasminogen activator (tPA), and improved door-to-needle times (DTNTs) among patients with acute ischemic stroke (IS).^1,2 However, when patients have received or are no longer candidates for acute therapies, the quality of their remaining inpatient care is still limited at many community hospitals by the lack of staff neurologists. Patients presenting to community hospital emergency departments (EDs) are often transferred due to lack of inpatient neurology coverage for follow-up.³ Although escalating care for patients needing the additional services of a comprehensive stroke center (CSC) is important, transferring patients with milder IS for a stroke consultation can burden both the patients and families and strain the limited resources of hub hospitals.

Many telestroke networks have begun keeping patients with milder stroke at resource-appropriate centers followed by community neurologists. Previous studies have shown conflicting data regarding the drip-and-stay model—in which patients are retained at the spoke hospital after tPA—and the drip-and-ship model—in which patients are transferred to a hub hospital after receiving tPA.^4,5 More recent studies have demonstrated shorter lengths of stay,⁶ improved time to consultation, and comparable functional outcomes in patients with teleneurology follow-up compared with community neurology–based follow-up. However, there are limited data comparing neurohospitalist in-person follow-up with inpatient teleneurology follow-up care.

To meet the needs of our spoke community hospitals, our telestroke network provides inpatient consultations for all neurology patients who can appropriately be kept at a spoke hospital. We wanted to evaluate the real-world practice of inpatient teleneurology follow-up consultations vs in-person neurohospitalist follow-up at spoke hospitals and compare hospital metrics, tPA complications, and functional outcomes for all patients initially seen by our telestroke network. We hypothesized that patients subsequently managed with teleneurology follow-up would have similar outcomes to those seen by in-person neurology follow-up after admission.

Methods

In this retrospective observational study of prospectively gathered data, we identified 3702 consecutive patients who were seen via telestroke activation between September 2015 and December 2018. We compared baseline demographics, clinical characteristics, and outcomes between patients followed up with in-person neurology or teleneurology. Our primary hospital-based outcomes were length of stay (LOS), discharge disposition, and subsequent transfer for higher level of care after admission. We also assessed patient-level outcomes such as readmission rates, 24-hour NIHSS scores for patients with stroke, and modified Rankin Scale (mRS) scores at 90 days.

Patients with suspected stroke who presented to spoke EDs within 24 hours of symptom onset and were remotely and emergently evaluated by a telestroke neurologist were included in our cohort. Telestroke consultants include board-certified and board-eligible Neurology and Vascular Neurology physicians. Patients transferred before being admitted to the hospital and in-hospital stroke codes were excluded. Some spoke centers had no or partial neurology coverage. For this study, we included only patients admitted at spoke sites with full-time routine neurology coverage either via in-person or teleneurology consultants. Patients seen for acute IS via telestroke were followed up as routine consults by the same group of telestroke consultants who completed their acute evaluations in the ED. Patients had final diagnoses including but not limited to stroke, seizure, neuromuscular emergencies, headache, and encephalopathy. Figure 1 shows the distribution of patients evaluated on follow-up and characteristics of spoke hospitals. The Lone Star Stroke Consortium Telemedicine Stroke Registry (LeSteR) is a Texas telemedicine registry that prospectively collects data from our teleneurology network clinical hub and spoke sites. Baseline variables with more than 30% missing data were not included in the analysis. Patients with missing outcome variables were removed from analyses. Details for the protocol followed by LeSteR are available in a prior publication.⁷ At the time of this study, the network included 17 spoke sites. Of these 17, 7 were primary stroke centers, and 3 are integrated CSCs.

Figure 1. Patient Distribution.

Patients who were evaluated over telestroke in the ED from 2015 to 2018 were included in our cohort. Of these, patients who were transferred to another center after the initial evaluation were excluded—only patients who stayed at their evaluated spoke site were included. Those with no or sporadic neurology follow-up were also excluded. We compared patients with stroke who were seen by an in-person neurologist with a teleneurologist. ED = emergency department.

Spoke Sites

All spoke sites supported by teleneurology had certain minimum requirements. These included access to MRIs, lumbar punctures, and ability to obtain routine EEGs. Patients requiring neurointensive care, neurosurgical evaluation, or who would benefit from an in-person neurologic examination were screened in the ED as often as possible and transferred from the ED. At non–teleneurology-supported sites with neurology support, patients with strokes were triaged by telestroke consultants in the ED. If nonstroke neurologic emergencies were present, initial recommendations were made to stabilize the patient, and the remaining triage was left to the site neurologist. There was no way to standardize practice among spoke sites not supported by teleneurology, but all 7 of these sites were at least primary stroke centers, so there is some uniformity with respect to stroke care. However, general neurology consults would be cared for at the discretion of the site neurologist.

Standard Protocol Approvals, Registrations, and Patient Consents

The Institutional Review Board of the University of Texas and the Memorial Hermann Hospital at Texas Medical Center ethics committee approved the study. Waiver of consent was obtained as this is a retrospective chart review study.

Statistics

Descriptive statistics including mean, SD, median, interquartile ranges, frequency, and percentage were provided for demographic information, baseline characteristics, and clinical metrics and outcomes. For group comparison, the 2-sample t test and Wilcoxon rank-sum (Mann-Whitney test) were conducted for continuous variables, and the χ² test and Fisher exact test were conducted for categorical variables (Table 1). The linear regression model was used for modeling continuous outcomes such as 24-hour NIHSS score and LOS. Logistic regression was used for modeling binary outcome variables such as tPA complications and stroke readmission. Ordinal logistic regression was used for modeling ordinal outcome such as the 90-day mRS score. Multinomial logistic regression was used for modeling nominal outcome such as disposition. All models were adjusted for age, sex, race and ethnicity, and initial NIHSS score (Table 2). All analyses were performed in SAS 9.4 (Cary, NC).

Table 1.

Baseline Characteristics of Patients at Spoke Sites

Table 2.

Outcomes of Patients With Stroke Followed Up at Spoke Sites

Power Calculation

We consider the clinically meaningful effect size for continuous outcome as 0.5 and the clinically meaningful group difference for binary outcome as 10%. With a sample size of 447 in the teleneurology follow-up (TNF) group and 1459 in the in-person neurology follow-up (IPF) group, we have a power over 90% to detect these group differences by a 2-sample t test or 2-sample proportion test with a type I error rate of 0.05. When limited to tPA-only patients, with a sample size of 88 in the TNF group and 403 in the IPF group, we still have over 90% power to detect the effect size of 0.5 and 75% power to detect a group difference of 10%, with a type I error rate of 0.05.

Data Availability

Anonymized data not published within this article will be made available on reasonable request to the corresponding author from any qualified investigator.

Results

We identified 3,702 patients seen via telestroke at spoke EDs in the teleneurology network from September 2015 to December/2018. Patients were excluded if they were transferred after initial evaluation (n = 376) or were kept at spoke hospitals with sporadic or no neurology inpatient follow-up (n = 1420). Figure 1 demonstrates the breakdown of our patient populations: 447 (23%) patients were managed with TNF, and 1459 (77%) were managed with IPF. There were no differences between TNF and IPF groups with respect to age, male sex, stroke severity, or premorbid mRS score (Table 1). Non-Hispanic Black patients were more likely to have TNF (42% vs 21%), and Hispanic patients were more likely to have IPF (6% vs 19%). Patients admitted to centers with TNF were less likely to have received tPA than patients seen at centers with IPF (20% vs 28%, p < 0.0001).

In the multivariate analysis, there were no significant differences in discharge disposition, stroke readmission rates, or 90-day mRS scores (Figure 2) between TNF and IPF groups. The LOS was shorter for TNF (p = 0.01), and patients were discharged to home and rehab at similar rates. After admission to the spoke hospital, only 3% of the TNF patients were subsequently transferred for higher level of care; none seen by IPF were transferred (Table 2).

Figure 2. Functional Outcomes on Telestroke-Evaluated Patients.

(A) Ninety-day mRS scores for all teleneurology-evaluated patients. TNF and IPF patients had comparable good functional outcomes, 90-day mRS scores 0–2 (p = 0.54; OR 0.90, 95% CI 0.64–1.26). There was no difference in the ordinal shift (p = 0.79; OR 1.04, 95% CI 0.80–1.35). (B) Ninety-day mRS scores of patients who received tPA but stayed at their local spoke sites. TNF and IPF patients had comparable good functional outcomes, 90-day mRS scores 0–2 (p = 0.57; OR 0.81, 95% CI 0.40–1.65). There was no difference in the ordinal shift (p = 0.77; OR 0.92, 95% CI 0.52–1.62). mRS = modified Rankin Scale; TNF = teleneurology follow-up; tPA = tissue plasminogen activator.

In a subanalysis of patients with confirmed strokes (n = 1089), there were no differences in readmission rates (4.2% vs 3.1%, p = 0.20), discharge disposition (p = 0.24), or 90-day mRS scores 0–2 (p = 0.93; eFigure 1, links.lww.com/CPJ/A382). Although the LOS was shorter between patients with stroke with TNF vs IPF (2 vs 3 days, p = 0.0006), TNF patients were more likely to require transfer after admission (3.7% vs 0, p < 0.0001). When evaluating outcomes in tPA recipients only, we identified 491 total patients who received tPA. Of these, 88 (18%) were subsequently managed at a center with TNF, and 403 (82%) were managed at a center with IPF. There was no difference in DTNTs for patients who received tPA at either type of center (eTable 1, links.lww.com/CPJ/A382). Few tPA complications (angioedema, symptomatic and nonsymptomatic intracerebral hemorrhage, and systemic hemorrhage) occurred in both groups (Table 2). There were no significant differences in discharge disposition (eTable 2, links.lww.com/CPJ/A382), stroke readmission rates, or 90-day mRS scores (Figure 2) among patients who received tPA.

Discussion

Among our cohort of patients who stayed at our telestroke network's spoke centers, 23% were subsequently managed with TNF. Among our full cohort of patients receiving TNF, the subset of patients with a diagnosis of IS, and the subset who received tPA, there were no differences in discharge disposition or 90-day functional outcomes, and there was decreased variation in LOS. In the overall cohort of patients, there was no significant difference in transfer after admission, patients with a diagnosis of stroke were more likely to be transferred after admission for higher level of care.

The influence of teleneurology on patient care has been well documented during the last 2 years under the influence of the COVID-19 pandemic; however, many centers instituted inpatient teleneurology/telestroke consultations even before the pandemic to address the growing demand for neurology coverage in our rural communities^6,8,9 Two recent articles compared metrics from telestroke or teleneurology inpatient consultations before and after implementation and found comparable functional outcomes⁹ and improved times to consultation.⁸ Although our LOS did was not significantly lower in all patients with TNF, LOS among patients with stroke did decrease as it did in a prior study of patients with stroke. The lower LOS may reflect the lower acuity of patients with stroke who stayed at TNF sites as patients with large strokes or requiring endovascular evaluation were transferred to a CSC. DTNTs were longer in 1 study after implementation of their telestroke follow-up⁸; however, they remained consistent in our study. Although there is not enough information to conjecture why our DTNT remained consistent, our telestroke network ensured that each rounding site had 2 telemedicine cameras available to minimize any effect on acute stroke care metrics.

Implementation of inpatient teleneurology follow-up reduces the initial transfer rate of patients with IS to hospitals with higher level of care, which can lead to a significant cost reduction.¹⁰ Only 3% of the patients were transferred after admission to our spoke sites; none evaluated by IPF and 13 patients evaluated over TNF, of which 9 had a diagnosis of stroke. Patients with stroke, however, did have a significant increase in transfers after admission with TNF. Reasons for transfer included status epilepticus, symptomatic intracranial hemorrhage after tPA, close monitoring due to fluctuating examination, and for reasons not directly related to the primary neurologic diagnosis (see eTable 3 for a full list, links.lww.com/CPJ/A382). Transfer for higher level of care because of clinical deterioration or additional testing required during a hospital stay is a complication that could delay care, affect outcomes, or increase LOS. For our patients, the need for higher level of care did not lead to significant differences in discharge disposition, increased mortality, or, for our patients with stroke, differences in 90-day mRS scores—although this may have been from the low number of patients. Although there was no difference in the 90-day mRS score among our patients with IS, the increased number of transfers after admission demonstrates the need for caution. Careful evaluation of imaging and clinical status before admission and an assessment of a spoke sites' capabilities for managing patients with stroke are necessary to ensure safety and limit these types of transfers.

We found a higher proportion of non-Hispanic Black patients and a lower proportion of Hispanic patients in the TNF group. Although this disparity may reflect the local spoke demographics, further evaluation into transfer patterns, the hospital's catchment region, and local population data is necessary to assess whether there is any inherent bias in how patients are triaged in spokes with IPF vs TNF. The spoke sites that we provide TNF to are sites that would otherwise have no neurology coverage available for their hospitals, which represents a significant disparity. That outcomes are not affected by this disparity is reassuring and suggests that our model may be deployed by other centers to address local disparities. We see that in the overall comparison of patients with IS seen at centers with TNF, patients were less likely to have received tPA than patients seen at centers with IPF follow-up.¹¹

There are several limitations to this study. Although many of our data were collected prospectively, our analysis is a retrospective one, and therefore, conclusions can only be associative. Not all our spoke sites were uniform, and as such, the level of morbidity each site can accommodate could vary. Three of our IPF spokes supported neurointensive care units and could care for sicker patients. Although the remaining 4 IPF sites all had similar resources as our 3 TNF sites, the ability to generalize outcomes is slightly limited by these differences. Our analysis also found differences in race and ethnic demographics between our TNF and IPF patient populations. As research into disparities was not a prespecified aim of this study, we are limited in our ability to discuss the demographics of the spoke site, evaluate for biases in transfer patterns, or provide additional information into socioeconomic determinants of health. However, these disparities should be included in future studies into the deployment of inpatient telemedicine services. We also cannot account for changes in the rounding styles of our neurologists as a result of being on telemedicine. However, our group created dedicated rounding service lines at each of our TNF spoke sites to mimic in-person neurohospitalist rounding as closely as possible. Although our teleneurology providers were part of the same practice group, the neurohospitalists at our IPF sites were not, and comparisons across different practice groups may oversimplify nuances despite our large sample size. Although the 90-day mRS score outcomes were not our primary outcome, they represent an important end point, and we were only able to gather data on 60% of the patients in our total cohort. Thirty-day readmission rates are also an important end point, of which we only had data on 47% of our total cohort. Prospective studies should be conducted to look more at these and other patient-centered outcomes.

In our teleneurology network, of 10 spokes, patients receiving TNF care had comparable outcomes when evaluated against patients receiving IPF. For patients with IS in particular, there were no differences in discharge disposition, LOS, or 90-day functional outcomes. Our results suggest that for select patients with stable neurologic conditions, telemedicine-based follow-up in the hospital can be a safe alternative to in-person neurology follow-up. Some patients did require subsequent transfer from teleneurology supported sites, which speaks to the importance of strict selection of patients appropriate for teleneurology evaluation before admission. This research expands on the growing evidence that hospitals lacking in-person neurology coverage can safely use telestroke coverage. Limiting transfers in the ED for patients who can safely stay in their communities alleviates the burden on hub facilities and helps address disparities in access to care. Although telemedicine follow-up should not replace in-person follow-up where such resources are available, it provides an option in resource-limited settings. Further studies are needed to characterize the patient's perspective on their care in the inpatient setting on telemedicine and replicate these findings in other teleneurology networks.

Appendix. Authors

Study Funding

This research is made possible by funding from the Lone Star Stroke Clinical Research Consortium. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the State of Texas.

Disclosure

The authors report no disclosures relevant to the manuscript. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.

TAKE-HOME POINTS

• → In our telestroke network, comparing patients seen after their initial telestroke consultation by either in-person (IPF) vs teleneurology follow-ups (TNFs), there were no differences in length of stay or discharge disposition.

• → However 3% of patients did transfer to hub sites for higher level of care.

• → Among only patients with a final diagnosis of stroke, there was no difference in discharge disposition or 90-day modified Rankin Scale scores.

• → Teleneurology may be a safe option for patients in institutions where in-person neurology follow-up is not available. This should be limited to select patients with stable neurologic conditions to avoid delays in care.