Abstract
Background and Purpose:
While the successful provision of telestroke care has been well documented in the literature, studies on the impact of comprehensive teleneurology service (TN) to hospital measures are lacking. We evaluated 3 traditional health services metrics of hospital performance: time from consult request to consult completion, inpatient length of stay (LOS), and the rate of patients transferred for tertiary care.
Methods:
Medical records (n = 899) from 3 community hospitals and our TN consultation database were retrospectively reviewed during the 2 years before (n = 703, 3 hospitals) and 4 months (n = 2 hospitals) to 2 years (n = 1 hospital) after implementation (n = 196) of a TN program for routine and urgent consult requests. Consult order time, consult completion time, total length of stay and discharge disposition were compared across the pre-TN implementation group, which consisted of in-person consultations and the post-TN implementation group, which consisted of TN consultations only.
Results:
After TN implementation, median length of stay decreased 28% (3.9 vs. 2.8 days, p < 0.0001) and median time from consult order to consult completion decreased by 74% across all diagnoses (5.8 vs. 1.5 hours, p < 0.0001). There were no significant differences in the percentage of patients discharged home (52.3% vs. 56.1%, p = 0.10) or transferred to tertiary care (6.1% to 9.2%, p = 0.10).
Conclusions:
Implementation of TN program was associated with significant reductions in LOS and time to consultation completion without an increase in shunting of patients to more advanced facilities. Further research is warranted to confirm these findings in independent cohorts and other models of teleneurology delivery.
Keywords
Introduction
Telemedicine is the provision of medical care from a remote location, enabling a patient to receive care despite the consultant physician not being physically present at the bedside. Telemedicine has been applied to several clinical scenarios with variable success. 1 -3
Telestroke was the first large scale application in the field of neurology. Such programs have become well established and have proven to be successful, increasing patient access to timely tissue-plasminogen activator (t-PA) treatment, especially in communities that lack the stroke expertise commonly found at large stroke centers. 4 -6 In recent years, telemedicine has been implemented in general neurology practice as well through programs of inpatient neurological consultation, monitoring of ICU patients with neurological conditions and sub-specialty outpatient care. Moreover, the COVID-19 pandemic has highlighted the need for providing remote neurology care succesfully. 7 -10
The interest in establishing comprehensive teleneurology (TN) programs has increased as the disparity between the supply and demand for neurology care has grown. The continuing growth of the aging population and increase in disease management options have led the American Academy of Neurology (AAN) to estimate a 20% shortage in the availability of neurologists. 11
While the impact of telestroke programs on bridging this gap between vascular neurologist supply and demand of stroke care is well documented in the literature, studies on the impact of general teleneurology on hospital outcome measures are lacking. Therefore, it is essential that teleneurology programs strongly focus on the evaluation of processes of care that are linked to measurable outcomes. 12
In this study, we aimed to address whether a comprehensive TN program leads to change in key hospital process measures including inpatient length of stay (LOS), time from consult request to consult completion as well as outcomes measures of patient disposition at community hospital discharge including the number of transfers to tertiary hospitals. Such metrics of care are important to monitor as they increase the costs of care, contribute to overcrowding at tertiary hospitals, and are an important marker of the efficiency and quality of the care delivered by teleconsultation.
Methods
Data Collection
This is study was approved by the IRB. We obtained data from 3 community hospitals affiliated with our hospital and served by the Routine and Urgent Teleneurology Program (RUTN)—1 rural community hospital in western Massachusetts and 2 campuses of a large suburban community hospital for a total of 554 covered beds. All hospitals shared the same electronic medical record (EHR). We did not include consultations requested for the primary reason of acute stroke evaluation which are performed by a different faculty team through our longstanding telestroke program. An evaluation of the telestroke program outcomes have been published previously. 6
Data were collected from 2 years prior to implementation of TN program for in-person, and all available data until this research was conducted were collected from the participating hospitals. At the 2 campuses of the suburban hospital, these were available for 4 months post implementation. At the smaller, rural MA hospital, data was available for 2 years post-implementation. For the pre-TN period, only patients that were admitted as inpatients and had an in-person neurology consultation requested at the time of or after the admission were included. For this group, no exact time of consult completion was entered into the community hospital EHRs. For the consult notes, 3 time stamps were available, the “Date of Dictation,” the “Date of Transcription,” and the “Date Signed.” The “Date of Dictation” was chosen as the indicator of time of consult completion, as this was the earliest of the 3 time stamps and thus the best approximate of the time of consult. For the post-TN period, only patients that were admitted as inpatients and had a TN consultation requested at the time of or after the admission were included. Furthermore, during the post-TN period, only TN consultations were available. In-person neurology consultations at the community hospitals were no longer available. For the post-TN group, the time of consult request was retrieved. The time the consult was signed was used to approximate the time that consult was completed.
The data collection was limited to each patient’s initial consultation; follow-up consultation data were not included in this analysis. Collected data were classified by a set of defined diagnosis categories, as well as by discharge disposition categories for the comparison of disposition patterns. To address concerns that the TN consultation process would simply identify patients needing advanced care rather than facilitate onsite management, we compared the proportion of all teleconsultations that resulted in each potential discharge destination.
Teleneurology Consultation Process
The process to request TN consultation involves the following steps: (1) the physician caring for the patient identifies the need for neurology consultation; (2) a request is then placed by such physician or unit coordinator into the MGH TN internet portal. When needed, the request may include relevant radiology imaging as well as relevant patient records, such as progress notes or laboratory results; (3) the TN portal notifies (via pager and email) the covering TeleNeurologist; (4) neurologist will call back and perform a phone and/or videoconference-enabled assessment of the patient, before rendering clinical recommendations to that same consulting physician; (5) a note is written and posted to the patient EHR.
Every TN consultation provided to patients was also documented in the online Teleneurology Portal, a HIPAA compliant, secure web portal and database developed by our center to track quality and performance data. TN consult order time, time of the TN consult, consult diagnosis, and patient disposition recommendation are routinely recorded and were extracted from the Teleneurology Portal and linked to information that was gathered from each hospitals’ EHR. Information on admission and discharge times and discharge disposition was obtained from each hospital’s EHR. We evaluated the time of teleneurology consult request in relation to the time of admission, time of consult completion and time of discharge, and compared these to traditional inpatient in-person neurology consults performed by local neurologists at the community hospitals before the implementation of TN. At that time, these hospitals had an in-person, non-employed multi-physician neurology group that provided neurology consultation coverage.: Prior to the implementation of the service, the spoke hospitals general neurology consultants who were dividing their practice between outpatient clinics and inpatient consultations. In the post-teleneurology phase, the service was staffed by 8 academic general neurologists, each covering 1 week at time. Teleneurology coverage was provided 24 hours a day and 7 days per week by faculty neurologists. Neurology trainees did not participate in the consults. Consults for acute stroke and emergency cases (e.g. status epilepticus, acute cord compression) were done by a different service (Telestroke and Emergency teleneurology), staffed by MGH stroke/ICU specialists.
The data for consultations in the Pre-TN period was abstracted from each community hospital’s EHR. All hospital consults that occurred after the implementation of the teleneurology program at each hospital were performed via TN only.
To maximize efficiency, the requests are assigned to 1 of 2 categories. Urgent consults (completion in 2 hours) or routine consults that should be completed within the same business day. Teleneurology consults were available 24/7 every day of the year. Consulting providers are instructed not to request a routine consult outside business hours (8am-5 pm).
Statistical Analysis
Mann-Whitney tests were used to compare non-parametric data such as pre-TN implementation data to post-TN implementation data for LOS, time from admission to consult completion, time from consult request to consult completion, and time from consult completion to discharge. X2 Tests of Independence were used for comparing categorical data on disposition patterns, as well as for comparing differences between the pre-TN and post-TN groups in the total number of patients with each diagnosis. Results for all tests were considered significant if p < 0.05. Analyses were performed using Microsoft Excel software version 15.24 (Microsoft) with additional statistical software from QI Macros 2018 Add-In (KnowWare International, Inc.). Anonymized data will be shared by request from any qualified investigator.
Results
The first hospital to implement teleneurology consultations (2015) has a total of 140 beds with 11 ICU beds and serves a rural community in Western Massachusetts. The other 2 other hospitals belong to the second largest community hospital system in the state of Massachusetts, totaling 368 beds with 40 ICU beds. They serve a suburban population north of Boston and implemented teleneurology consultations in the spring of 2017. All 3 hospitals have multiple mobile cart-based teleneurology endpoints for bedside consultations and used wireless connectivity for video transmission. Teleneurologists were able to access all the notes, labs and images before and during the time of the consultation via the web-based teleneurology portal.
Data related to 899 unique consultations were obtained, 703 in-person, inpatient neurology consultations pre-implementation of TN and 196 inpatient teleneurology consultations. The average daily number of consults in the pre-TN 0.95/day was similar to post-TN 1.36/day
From billing data (main diagnosis), the main neurological diagnoses leading to consultations in the pre-TN implementation period were present as follows: Subacute stroke, 35.3%; encephalopathy, 16.2%; migraine/other headache, 11.2%; TIA, 5.8%; syncope/loss of consciousness, 5.4%; carotid stenosis, 1.0%; subacute intracranial hemorrhage, 0.7%; seizure/convulsions, 0.3%; meningitis, 0.1%; all other billing diagnoses that did not fit in these major categories, 23.9%. In the post-TN implementation period, data from billing records (main diagnosis) and TN portal (reason for neurology consultation): Subacute stroke, 49.5%; encephalopathy, 14.8%; TIA, 13.3%; seizure/convulsions, 12.2%; syncope/loss of consciousness, 6.1%; migraine/other headache, 3.6%; meningitis, 0.5%.
Multiple parameters differed when Pre-TN was compared to post TN implementation. The median length of stay for neurology inpatients decreased from 3.9 days (IQR 2.4,6.5) to 2.8 days (IQR 1.7,4.1, p < 0.0001). The mean time from admission to consult order was 8.6 h in the pre-TN and 20.3 h in the post-TN period (Figure 1).
Comparison of timeline of events of patient stay for patients that received an in-person inpatient neurology consult before the implementation of RUTN (Pre-TN, n = 703) to that of patients that received an inpatient TeleNeurology consult (Post-TN, n = 196).
The median time between when a consult order was placed and when a consult was performed decreased 5.8 hours (IQR 2.4, 15.6) to 1.5 hours (IQR 0.5, 2.9, p < 0.0001). These differences were significant for individual diagnoses as well: encephalopathy (5.7 hrs. difference, p < 0.0001), migraine/headache (5.2 hrs. difference, p = 0.003), seizure/convulsions, (5.2 hrs. difference, p = 0.034), stroke (5.0 hrs. difference, p < 0.0001), syncope/loss of consciousness (5.4 hrs. difference, p < 0.0001) and TIA (6.1 hrs. difference, p < 0.0001).
The percentage of neurology patients that received inpatient consults (traditional bedside in the Pre-TN period and teleneurology consults in the post TN period) who were transferred to a different facility was similar, (pre-TN 6% vs post-TN 9%, p = 0.14) (Figure 2).
Comparison in discharge disposition rates from neurology inpatient in-person consult patients (n = 703) to neurology inpatient TeleNeurology consult patients (n = 196).
Discussion
The results of our study suggest that a comprehensive TN program can help community hospitals provide access to neurologic consultations more quickly than in a traditional community practice model, and this change was associated with a shorter hospital stay when compared to the previous in-person consultation system. In addition, reassuringly, implementation of TN consults was not associated with an increase in the rate of transfers to other facilities. It also resulted in a similar rate of discharges directly to home in this population studied. We found that the time to request a consult was larger post TN and while specific reasons were not studied, we think the unfamiliarity with a new process may explain, at least in part, such delay. It is also possible that restricting placement of routine consults in the post-TN period to 8 am to 5 pm, could have contributed to such finding as in the pre-TN period there was no such rule.
ED consults were removed from the analysis for 2 reasons. First, in the pre-TN period there was no well-defined system for such consultations in both hospitals. Often, for patients with less severe presentation, there was a curbside phone call to neurology rather than a formal evaluation. Second, the focus of the study was to evaluate hospital outcome measures for inpatient admissions. The analysis of metrics for ED teleneurology consultations is of key importance for future studies.
As a measure of performance of the teleneurology service, we compared the timeliness of consultations, patient length of stay and rate of patient transfers out of community hospitals for more advanced care. These are hospital performance indicators that affect the quality of care, patient satisfaction, bed availability, and healthcare expenditures. 13,14 While there have been no similar studies in the field of neurology, implementation of telemedicine visits in intensive care units (ICU) as well as stepdown and subspecialty care units has shown to improve outcomes. In a prospective study of 6290 adults admitted to ICU, telehealth support was associated with reduced mortality from 13.6% to 11.8%, higher rates of adherence to best clinical practices, and shorter hospital length of stay (9.8 vs 13.3 days) with comparable results for medical, surgical and cardiovascular ICUs. 15 In a large prospective study comparing ICU patients with (n = 8091) and without telemedicine intervention (n = 8000), after adjusting for demographics and disease severity, mean length of stay was lower among patients served by telemedicine (2.6vs3.2d; p < 0.0001). 16 Moreover, recent studies in orthopedics, pediatrics, and emergency departments have shown that the access to telemedicine consultations was associated with shorter times to consultation and improved triage. 17 -19 The differences in efficiency seen in our study are largely explained by the shortage of neurologists who can carve out dedicated time to attend to inpatient consult requests at community hospitals. It is a common practice in community hospitals for a neurologist to first finish their outpatient scheduled visits before traveling to the hospital to begin inpatient consultations. The same factors also result in delays or lack of bedside availability during nights, weekends, and holidays or in the vent of unexpected illness or vacations. This situation often leads to delays in initiation of consultations which can have significant downstream consequences for the timely disposition of patients and implementation of therapies. While we did not study the issue of subspecialist expertise directly, access to a large academic neurology department likely increases access to subspecialty opinions for complex cases or when determining if transfer to a tertiary care hospital is medically appropriate.
All the teleneurologists were faculty physicians at Massachusetts General Hospital and serve as general neurology or stroke ward attendings and/or in the hospital neurology consult service. While all of these neurologists have subspecialties training and work in outpatient clinics they have several years of experience managing neurohospitalist issues. In addition, for complex cases, given the larger faculty of our institution, a sub-specialist was available to discuss cases with the primary teleneurologist. Before starting on service, the teleneurologists undergo telehealth training, in-service onboarding. We also conduct monthly faculty meetings to discuss clinical and operational challenges.
Obtaining expert answers faster through telemedicine may increase confidence of local referring healthcare teams, assuming there is no perception of a compromise in clinical care quality over telemedicine. With TN implementation, the response time by a consultant to a consult request decreased significantly, leading to more prompt diagnostic workup and likely to new treatment initiation. Teleneurology may have other ancillary benefits. For hospitals who lack on site neurology consultation availability altogether, TN can help retain a larger proportion of patients with neurologic symptoms in the community if expert consultation deems them safe to remain. This reduces costs and prevents dislocation of patients from their families and support networks. By decreasing the volume of inpatient consultations and the ensuing demand on neurologist’s time, a TN program may reduce community physician burden and burnout and promote retention of community neurologists in practices focusing on outpatient work.
The retrospective nature of this study is a main limitation. It is possible that differences observed in length of stay after the TN implementation might reflect unmeasured confounding that led to improvements in timeliness of care. For instance, the implementation of a new program may have indirectly led to greater attention to process delays and triggered more appropriate and timely consultations. The teleneurology implementation includes a quality improvement intervention with training of the community hospitalist staff to care for patients with neurological diseases. The contribution and duration of these elements were not studied separately from the teleneurology consult process itself, so we cannot say whether or which of these components is driving the improvements. Other factors that may impact length of stay, such as severity of illness on admission, the day of admission, patient comorbidities, social and insurance issues, hospital bed occupancy may also have differed pre-and post-implementation, and these data elements were not available for analysis. Another limitation of our study is that differently from important telestroke and teleICU studies, we did not measure adherence to best practices or clinical outcomes. Finally, our findings are from the teleneurology provision from a single academic institution to local hospitals on a spoke-hub model and should not be generalizable to services provided on different models, eg. telemedicine private companies. Finally, we did not capture the transfer rate from the ED to tertiary hospitals pre and post-TN, but it is likely that such numbers did not change significantly since the average daily number of new inpatients consults in the pre-TN was similar to post-TN.
Nonetheless, from an operational standpoint, our findings demonstrate that implementation of our TN programs was associated with increased hospital capacity and improved hospital margins for DRG-based payments by achieving a shorter length of stay for these patients. Our study did not specifically look at costs or performed a financial analysis related to TN coverage.
It is important to stress that we believe a successful TN program is one that provides fulltime coverage with well-trained specialists based on reliable and high-quality communication technologies that support high frame rates and adjustable remote-control cameras. It must ensure adequate privacy of healthcare information transmission and conform to appropriate FDA standards, and enable remote viewing of important medical information such as imaging, electrophysiology and other neurodiagnostic testing. It is important to highlight that our findings are from the teleneurology provision from a single academic institution to local hospitals on a spoke-hub model and should not be generalizable to services provided on different models.
In summary, our study suggests that an academic-based teleneurology program can meet an important and ever-growing need in the community for access to specialty care. It does so by providing the care in a timelier manner than currently exists for many patients hospitalized in the community today. Further research is warranted to confirm if these findings can be replicated in other programs and other environments.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Robert McCormick, Juan Estrada MBA, Cynthia Whitney RN, Adam Cohen MD and Marcelo Matiello MD have nothing to disclose. Lee Schwamm MD serves as a consultant on usability to LifeImage (a privately held teleradiology company) and is the Director of the Mass General Center for TeleHealth. Dr. Matiello is supported by the clinician teacher development award by the Mass General Hospital Center for Diversity and Inclusion.
