TeleStroke Units Serving as a Model of Care in Rural Areas
10-Year Experience of the TeleMedical Project for Integrative Stroke Care
Background and Purpose—Stroke Unit care improves stroke prognosis and is recommended for all patients with stroke. In rural areas, population-wide implementation of Stroke Units is challenging. Therefore, the TeleMedical Project for integrative Stroke Care (TEMPiS) was established in 2003 as a TeleStroke Unit network to overcome this barrier in Southeast Bavaria/Germany. Evaluation of its implementation between 2003 and 2005 had revealed improved process quality and clinical outcomes compared with matched hospitals without TeleStroke Units. Data on sustainability of these effects are lacking.
Methods—Effects on the stroke care of the local population were analyzed by using data from official hospital reports. Prospective registries from 2003 to 2012 describe processes and outcomes of consecutive patients with stroke and transient ischemic attack treated in TEMPiS hospitals. Quality indicators assess diagnostics, treatment, and outcome. Rates and timeliness of intravenous thrombolysis as well as data on teleconsultations and secondary interhospital transfers were reported over time.
Results—Within the covered area, network implementation increased the number of patients with stroke and transient ischemic attack treated in hospitals with (Tele-)Stroke Units substantially from 19% to 78%. Between February 2003 and December 2012, 54 804 strokes and transient ischemic attacks were treated in 15 regional hospitals, and 31 864 teleconsultations were performed. Intravenous thrombolysis was applied 3331 stroke cases with proportions increasing from 2.6% to 15.5% of all patients with ischemic stroke. Median onset-to-treatment times decreased from 150 (interquartile range, 127–163) to 120 minutes (interquartile range, 90–160) and door-to-needle times from 80 (interquartile range, 68–101) to 40 minutes (interquartile range, 29–59).
Conclusions—TeleStroke Units can provide sustained high-quality stroke care in rural areas.
Organized care in Stroke Units is considered the most effective intervention for patients with acute stroke.1,2 The World Health Organization declared the aim of treating all patients with stroke on dedicated Stroke Units.3 The short time window for recanalization therapies4 and the preference particularly of elderly patients to be treated in their familiar local hospital favor a decentralized system of care. However, implementation of Stroke Units requires multidisciplinary stroke expertise which is not sufficiently available in many, often rural, areas. Therefore, delivering appropriate stroke care in such regions is a major public health challenge.5 A possible approach is to establish TeleStroke Unit networks: Stroke Units in rural areas supported by stroke centers in terms of education, quality management, and teleconsultations.
The TeleMedical Project for integrative Stroke Care (TEMPiS) in Bavaria, Germany, was one of the first TeleStroke projects. The evaluation of its implementation phase between February 2003 and December 2005 showed to be effective in providing safe and extended thrombolysis6,7 and in improving stroke outcomes compared with matched hospitals without TeleStroke Units.8,9 However, little is known about sustainability of such networks in terms of structures, quality of care provided, and outcome.
Therefore, we analyzed changes in structures as well as processes and patient outcomes within the TEMPiS network for a 10-year period regarding our aim of providing evidence-based acute care for all patients with stroke within the network area.
TEMPiS comprises 2 neurological stroke centers and 15 regional hospitals in Southeast Bavaria/Germany (Figure 1). Detailed descriptions of the concept were published previously.8 None of the regional hospitals had a Stroke Unit before network start in 2003. Local stroke care is provided by internal medicine departments in 11 hospitals and by neurological departments in 4 hospitals. The concept requires a neurologist on site at least during working days (part or full time). In the remaining times, the 11 hospitals without neurological departments have access to neurological expertise via telemedicine only. The 4 hospitals with neurological departments use the teleconsultation service for second opinion and organization of emergency interhospital transports.
The core elements of TEMPiS are as follows:
Implementation of a TeleStroke Unit in each hospital with formation of a multidisciplinary stroke team and implementation of standardized stroke care protocols. Intravenous thrombolysis (IVT) is provided as a 24/7 service with IVT routinely delivered in regional hospitals. The size of TeleStroke Units is planned to cover a volume of ≥250 patients with stroke and transient ischemic attack (TIA) per year.
Continuing training for all stroke team members provided by a multidisciplinary education team provided by the 2 stroke centers. The comprehensive quality management system also comprises audits, compulsory participation in the Bavarian stroke registry and the TEMPiS IVT registry, benchmarking, and certification procedures.
Implementation of telemedicine for 24/7 availability of acute neurological stroke expertise. Telemedicine comprises a telephone contact between the on-site physician and the vascular neurologist at the stroke center, transmission of digital brain images to the vascular neurologist (forwarded to a neuroradiologist/neurosurgeon if indicated) and, on request of the vascular neurologist, a real-time clinical examination of patients via videoconferencing. The teleconsultation team based in the stroke centers consists of 5 to 7 full-time vascular neurologists who must not be in charge of other clinical tasks during their teleconsultation shift. Teleconsultations are always performed as a physician-to-physician communication. The system runs on dedicated stationary and mobile workstations also allowing consultations from home. Mandatory indications for teleneurological assessment were predefined. They include eligibility for IVT, decreased consciousness, brain stem symptoms, intracranial hemorrhage, severe stroke (National Institute of Health Stroke Scale score >12), and progressive stroke. A telemedical follow-up can be requested at any time of the in-hospital stay. Written summaries are sent electronically to the enquiring hospital shortly after finishing teleconsultations.
Geographical Selection of Regional Hospitals
At network implementation, the target area in Southeast Bavaria comprised 16 administrative districts with 1.90 million inhabitants covering an area of 14 538 km2 (Figure 1).10 To enable IVT administration for all suitable patients, we aimed at prehospital transport distances of no longer than 30 minutes. This approach resulted in 12 TeleStroke Units with ≥250 strokes per year. By the end of 2012, 3 additional hospitals within 3 new administrative districts entered the network. The coverage rose to 2.28 million inhabitants and 17 887 km2 resulting in a population density of 127 inhabitants/km2.10 Another hospital with a high stroke volume was member from 2006 until 2011 but then left the network to run an autonomous neurological Stroke Unit (data not included in analyses).
All data for this 10-year analysis refer to the TEMPiS TeleStroke Unit hospitals; data from the 2 centers are not included. Data were collected from several sources: data from the general local health insurance (AOK), routine data for administrative purposes, data of the TEMPiS teleconsultation, and IVT registries and the Bavarian stroke registry.
Health Insurance Data
The AOK is the largest health insurance in Bavaria with a market share of ≈40% and provided data to assess the trend in proportions of patients with stroke treated in hospitals with specialized stroke care. Because we wanted to demonstrate the effect of our network in rural areas, we restricted this analysis to the rural TEMPiS region (Figure 1) excluding therefore 1 regional hospital in the city of Munich (Pasing). Annual numbers of AOK-insured patients with stroke and TIA admitted to acute hospitals with a (Tele-)Stroke Unit (regardless of TEMPiS participation) were compared with all AOK-insured patients with stroke and TIA living in this region.
The financial departments of the regional hospitals provided numbers of stroke and TIA admissions per year that were collected for administrative purposes. Cases with hospital main diagnoses ischemic, hemorrhagic, and undefined stroke and TIA (International Classification of Diseases, Tenth Revision codes G45, I61, I63, and I64) were included.
TEMPiS Teleconsultation Registry
Data from all teleconsultations were collected prospectively by the 2 stroke centers including date and time of request. Analysis of transfer recommendations to a stroke center was restricted to the years 2003 (February 1, 2003, to January 31, 2004) and 2012.
TEMPiS IVT Registry
All TeleStroke Units use a uniform IVT protocol for documentation of times (onset, admission, and treatment), National Institute of Health Stroke Scale, and formal contraindications. All protocols are sent to the network coordination center for plausibility checks before entering the data into the registry. Completeness of protocols is ascertained by random samples taken by the network coordinator and comparison with the teleconsultation registry and the Bavarian stroke registry. If onset of stroke symptoms was not observed, the last-time-seen-well was defined as onset. In-hospital strokes were excluded from analyses of timelines. Data from 2006 to 2008 were not collected and are therefore not included in the analyses. IVT rates are expressed as number of IVT compared with number of ischemic strokes (ISs) admitted to the hospitals.
Bavarian Stroke Registry
All patients with stroke and TIA of the participating hospitals were prospectively registered using anonymized data sets. The annual documentation rate compared with routine data has continuously been >95% since 2008, but lacked some completeness until 2007 (62.9%–87.1%). The data set covers information on baseline characteristics of patients with index stroke, diagnostics, treatment, and outcome. The data sets are checked for plausibility by the Bavarian Working Group for Quality Management in Inpatient Health Care. Evidence-based stroke quality (for definitions, see Table I in the online-only Data Supplement) is assessed, and target values for Stroke Units are used according to the independent German Quality Indicator Board.11
The Mann–Whitney U test was used to test differences in continuous variables, and the χ2 test was used for those in proportions. Data are presented by year for demonstrating trends over time. We calculated the P for trend adjusted for age and sex by adding year of admission as continuous variable with the respective indicator as outcome to a logistic regression model for categorical and a generalized linear model for continuous variables. All tests were 2 tailed, and statistical significance was determined at an α-level of 0.05. Analyses were performed on IBM SPSS Statistics 22 and SAS 9.2.
The TEMPiS teleconsultation and IVT registries have been approved by the institutional review board in Munich. Basis for the Bavarian stroke registry is a German law (§137 SGB V). According to German legislation, no patient consent is required for documentation in routine observational quality registries. Retrospective analysis of anonymized health insurance data was approved by the data protection officer of the AOK.
Impact of the Network on the Regional System of Stroke Care
The average number of the AOK-insured patients with stroke and TIA living in the TEMPiS region during the years 2002 to 2012 was 5095 per year. The annual proportions of patients with stroke and TIA admitted to hospitals with a dedicated Stroke Unit increased from 19% (995/5292) before network implementation (2002) to 52% (2585/4907; P<0.001) in 2003 and to 78% (4039/5151) in 2012 (P<0.001; Table). From 2003 onward, the majority of patients treated in hospitals with a Stroke Unit were admitted to TEMPiS hospitals (71% in 2003 and 72% in 2012).
Stroke Care in TEMPiS Regional Hospitals
Between February 1, 2003, and December 31, 2012, a total of 54 804 patients with stroke and TIA were documented in the TEMPiS hospitals. The number of patients with stroke and TIA increased from 4109 (in 2003) to 7207 (in 2012) (Figure 2). The mean number of patients with stroke and TIA per hospital per year increased from 342 to 480 (2003 and 2012). Table II in the online-only Data Supplement shows demographic and clinical characteristics of patients with stroke over time. Time trends show a better differentiation of stroke subtypes, increasing age, and less severe symptoms (decreasing National Institute of Health Stroke Scale) on admission over time.
By the end of 2012, the total number reached 31 864. During the past decade, numbers of teleconsultations per year have steadily increased from 1928 (in 2003) to 4513 (in 2012; Figure 2). In 2003, videoconferencing was done in 988 of 1928 teleconsultations (51%) with the rest of the teleconsultations restricted to evaluation of brain images and telephone consultation. This proportion of video-comprising teleconsultations slightly decreased to 47% (2134/4513) in 2012 (P=0.004). Recommendations of secondary transfers from the regional hospitals to stroke centers decreased from 11.5% (252/2182 teleconsultations) in 2003 to 7.0% (317/4513) in 2012 (P<0.001).
Quality of Stroke Care
The absolute numbers of IVT per year steadily increased from 10 before network implementation (2002) and 63 in the first year of the network (2003) to 685 in 2012, totaling 3331 IVT since 2003. IVT rate as the proportion of all patients with IS admitted to the hospitals increased from 2.6% (63/2466) in 2003 to 15.5% (685/4409) in 2012 (P<0.001; Figure 3). Median onset-to-treatment times decreased from 150 minutes (interquartile range [IQR], 130–165) in 2003 to 120 minutes (IQR, 87–160) in 2012 (P<0.001). Median door-to-needle times also decreased from 80 (IQR, 68–101) to 40 minutes (IQR, 29–59; P<0.001). Median onset-to-door times remained stable (60 minutes [IQR, 45–79] in 2003 and 68 minutes [IQR, 48–105] in 2012; P=0.63). The percentage of IVT treatments administered within the first hour after admission improved from 26% (15/57) in 2003 to 80% (543/675) in 2012 (P<0.001).
Data from the Bavarian stroke registry about diagnostics, treatment, and outcome procedures for the 10-year period are presented in Table II in the online-only Data Supplement. Most parameters show clear improvement over time, whereas length of in-hospital stay decreased. Seven-day in-hospital mortality in all patients with IS has decreased from 5.0% (74/1469) in 2003 to 3.1% (145/4618) in 2012, whereas 7-day mortality after IVT remained stable (mean, 6.3%; P for trend=0.122).
Table III in the online-only Data Supplement shows nationwide defined stroke quality indicators for 2012. TEMPiS hospitals achieved quality-of-care levels above target values in 9 indicators with 4 indicators slightly below target (all within 1% difference form targets).
The main findings of 10-year experience of TEMPiS showed that this type of telemedical Stroke Unit network is sustaining, offers state-of-the-art acute stroke care by increasing access to Stroke Units and improving thrombolysis service, and is associated with long-term improvement in terms of quality indicators of acute hospital care.
Many TeleStroke networks were established during the past few years worldwide. In contrast to most other networks with a focus on telemedicine-enabled stroke thrombolysis or referral to tertiary hospitals,12,13 TEMPiS integrates the Stroke Unit concept in regional hospitals.
According to local health insurance data, the proportion of patients with stroke and TIA being admitted to a hospital with a dedicated Stroke Unit increased from 19% to 78% in the source population. Apart from establishing the new TeleStroke Units, many efforts have been made to convince the population, ambulance services, and primary care physicians about the better prognosis of patients with stroke treated in a Stroke Unit. This strategy was reinforced by the positive result of the TEMPiS efficacy analysis.8 However, a Stroke Unit admission rate of 78% does still not appear satisfactory because the goal should be close to 100%. In Germany, emergency medical services are usually organized within administrative districts and seem to be less inclined to bring patients to a TeleStroke Unit of the adjacent district even if it could be reached within 30 minutes. This has led to a change of our strategy. In consensus with health authorities, in 2013, we started to set up further TeleStroke Units in administrative districts so far lacking a (Tele-)Stroke Unit.
Despite the primary focus on Stroke Unit care, the IVT rate from all patients with IS admitted to TEMPiS hospitals rose from 0.4% to 15.5%, reflecting a high rate compared with other systems of care.14 Although IVT rates in TeleStroke networks are usually reported as tissue-type plasminogen activator rates of teleconsultations only, we could not find any comparable published IVT rate (IVT per IS) other than the 9.6% reported from Spain.15 During the 10 years, onset-to-treatment times and door-to-needle times decreased remarkably. Eighty percent of all IVT in the TEMPiS Stroke Units are delivered within 60 minutes of admission, thus being clearly beyond the target of the American Stroke Association of 50%.
Decrease of secondary transfers is mainly attributable to more careful recommendations regarding surgical treatment in hemorrhagic strokes and improved local stroke care. The increase in patients with minor stroke may also contribute.
Costs for regional hospitals and centers in our TeleStroke Unit concept are substantial and need adequate reimbursement. In 2002, health policy authorities decided to support telemedicine networks. When the controlled study showed that the TEMPiS concept was both improving clinical outcomes and cost effective when compared with conventional care (higher costs during acute care were fully compensated by lower rehabilitation and long-term care costs),16 ongoing support by health insurances covering the teleconsultation service and the quality management program was decided in 2006. Meanwhile, the running costs of the TeleStroke Units for the regional hospitals are reimbursed within the German diagnosis related groups system. Teleconsultants are employed by the network and paid according to German tariffs. However, cost-effectiveness is only given, if numbers of patients with stroke and TIA per hospital and the number of teleconsultations exceed critical limits. Use of mobile (laptops or tablet) devices for teleconsultants may reduce costs and may enable networks with lower numbers of teleconsultations,17 especially beyond normal working hours.18
Participation in such a comprehensive network requires ongoing motivation. In our opinion, the most important incentives include the everyday experience of obtaining fast and clinically relevant recommendations and the conviction in the regional hospitals that participation in the network is beneficial for patients and hospitals alike. Patients benefit from improved outcome.8,9 Hospitals experience the advantages of specialization in treatment of one of the most frequent and devastating diseases.
The study has limitations. Because of nonmodifiable regional structures, the transfer of our concept to other rural regions meets limitations. The population density in the rural TEMPiS area of 127 inhabitants/km2 allows annual stroke volumes of >250 patients with stroke and TIA per TeleStroke Unit. With substantially lower population densities like, for example, in Georgia, United States (63/km2) or Alberta, Canada (6/km2), TeleStroke networks also reported benefits, mainly by increasing numbers of IVT.19,20 Our single-network data represent prospectively collected registry data, routine data, and data of 1 health insurance. Therefore, selection and coding biases as well as data collection errors cannot be excluded. However, cross-checking of the data between different sources ensured plausibility. Documentation rate for the Bavarian stroke registry compared with routine data lacked some completeness in the years 2003 to 2007 but improved in more recent years. A control group is lacking. We are therefore not able to separate network effects from general improvements of stroke care over time in the past decade.
The analysis of secondary transfers is based on the recommendations of physicians providing teleconsultation services. In some cases, transfers may not have been carried out and some patients may have been transferred against recommendation of a teleconsultation. The presented proportion of patients treated in a hospital with a (Tele-)Stroke Unit in relation to all hospitalized patients with stroke and TIA living in the TEMPiS region might have several biases. First, the analysis is restricted to patients of 1 health insurance. Although the AOK is the largest health insurance in Bavaria with a proportion of ≈40%, this may produce a selection bias. Second, all patients with stroke and TIA registered in the TEMPiS region are included but not all of them may have experienced their stroke while staying in this area. Third, to be treated in a hospital with a Stroke Unit does not inevitably mean to be treated in a Stroke Unit. A minority may have been treated on other wards for comorbidities or complications (eg, intensive care units), but each TEMPiS hospital has medical and financial incentives to treat patients with stroke preferably on their TeleStroke Unit.
The TeleStroke Unit network TEMPiS is an example of how the challenges of area-wide implementation of Stroke Units in rural areas can be met. Although network structures have to be adapted to regional features, for example, population densities, geographical and administrative borders, and hospital infrastructures, we think that many of the experiences gained in the TEMPiS concept might be used as a role model for other rural areas.
We thank all nurses, therapists, and physicians of the participating hospitals and all teleconsultants for their exceptional support of the network. Participating stroke centers were as follows: Department of Neurology, Klinikum Harlaching, Städtisches Klinikum München, and Department of Neurology, University of Regensburg. Regional hospitals were as follows: Since 2003: Bad Tölz, Burglengenfeld, Cham, Dachau, Ebersberg, Eggenfelden, Freising, Kelheim, Mühldorf, Pasing, Rosenheim, Straubing. Additional hospitals were as follows: Traunstein since October 2006, Zwiesel since January 2008, Bad Reichenhall since July 2011. We thank P. Hermanek (BAQ) as well as W. Langenecker and C. Bachhuber (AOK). We also thank S. Gutmann-Patchett for careful language editing.
Sources of Funding
TEMPiS is continuously supported by the Bavarian health insurance companies. Initial support was also given by the Bavarian State Ministry for Health and the German Stroke Foundation (Stiftung Deutsche Schlaganfall-Hilfe). Parts of the project and the data analysis were supported by the German Federal Ministry of Education and Research (BMBF) within the Competence Net Stroke.
Drs Müller-Barna, Hubert, and Boy are involved in the management of TEMPiS. Dr Müller-Barna received speaker honoraria from Covidien. Dr Heuschmann received research support from German Federal Ministry of Education and Research (BMBF), European Union, German Stroke Foundation, Charité-Berlin, Berlin Chamber of Physicians and University Hospital Würzburg; Dr Wiedmann received research support from BMBF; and Dr Hubert received research support from Bavarian Ministry of Health. Dr Audebert received speaker honoraria from Boehringer-Ingelheim, Lundbeck A/S, Pfizer, BMS, Takeda, and Sanofi and consultancy honoraria from Lundbeck A/S, Roche, and Bayer Vital. The other author reports no conflicts.
Presented in part at the 9th World Stroke Congress, Istanbul, Turkey, October 22–25, 2014.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.006141/-/DC1.
- Received May 15, 2014.
- Revision received June 12, 2014.
- Accepted June 30, 2014.
- © 2014 American Heart Association, Inc.
- Indredavik B
- Lees KR,
- Bluhmki E,
- von Kummer R,
- Brott TG,
- Toni D,
- Grotta JC,
- et al
- Joubert J,
- Prentice LF,
- Moulin T,
- Liaw ST,
- Joubert LB,
- Preux PM,
- et al
- Audebert HJ,
- Kukla C,
- Clarmann von Claranau S,
- Kühn J,
- Vatankhah B,
- Schenkel J,
- et al
- Audebert HJ,
- Schenkel J,
- Heuschmann PU,
- Bogdahn U,
- Haberl RL
- Audebert HJ,
- Schultes K,
- Tietz V,
- Heuschmann PU,
- Bogdahn U,
- Haberl RL,
- et al
- 10.↵Regional Database Germany. Federal Statistical Office and the Statistical Offices of the Länder. https://www.regionalstatistik.de/genesis/online (Codes 173-01-4 for population and 171-01-4 for area). Accessed April 3, 2014.
- Heuschmann PU,
- Biegler MK,
- Busse O,
- Elsner S,
- Grau A,
- Hasenbein U,
- et al
- Schwamm LH,
- Holloway RG,
- Amarenco P,
- Audebert HJ,
- Bakas T,
- Chumbler NR,
- et al
- Silva GS,
- Farrell S,
- Shandra E,
- Viswanathan A,
- Schwamm LH
- Saver JL,
- Smith EE,
- Fonarow GC,
- Reeves MJ,
- Zhao X,
- Olson DM,
- et al
- Pedragosa A,
- Alvarez-Sabin J,
- Molina CA,
- Sanclemente C,
- Martín MC,
- Alonso F,
- et al
- Audebert HJ,
- Boy S,
- Jankovits R,
- Pilz P,
- Klucken J,
- Fehm NP,
- et al