Clinical outcome of rural in-hospital-stroke patients after interhospital transfer for endovascular therapy within a telemedical stroke network in Germany: a registry-based observational study

Study design

This was a registry-based multicentre cohort study with a blinded endpoint conducted according to Strengthening the Reporting of Observational Studies in Epidemiology guidelines. We analysed prospectively collected data on consecutive patients receiving EVT after interhospital transfer from 14 rural hospitals to eight referral stroke centres of the Telemedical Stroke Network in Southeast Bavaria (TEMPiS), Germany, between February 2018 and July 2020. The geographical area in which TEMPiS operates is illustrated in figure 1. TEMPiS has set up stroke units in several rural hospitals and provides them with 24/7 telemedical consultation by stroke specialists from two comprehensive stroke centres in the diagnostic process and therapy decision-making. The data originate from a study registry for quality management and evaluation of the Flying Intervention Team (FIT) project, which explores the effect of circumventing interhospital transfer by flying in specialised neuroradiologists to perform EVT on-site in rural hospitals.17 The data are collected by the attending neurologists and neuroradiologists, as well as trained documentalists with expertise in the medical field who conduct structured clinical interviews 3 months after stroke. The keeping of the study registry complies with the guidelines of the Declaration of Helsinki and is approved by the ethics committee of the Bavarian State Medical Association (ClinicalTrials registration number for FIT study: NCT04270513).

Figure 1
Figure 1

Map of the operational area of TEMPiS in a study on the effect of stroke onset inside versus outside the hospital (in-hospital-stroke vs out-of-hospital-stroke) on clinical outcomes after interhospital transfer for endovascular therapy. Map of Southeast Bavaria, illustrating the operational area of the TEMPiS network. The population density of each county is represented using different shades of blue. Primary stroke centres are indicated by blue dots; and comprehensive stroke centres are represented by red dots. The larger red dot signifies the presence of two comprehensive stroke centres in Munich. The distance from Munich is depicted by black half circles. Blue arrows denote the typical transfer routes. TEMPiS, Telemedical Stroke Network in Southeast Bavaria.

Study population

We defined the study cohorts by symptom onset in the rural hospital (IHS) versus symptom onset out of the hospital and admission via the emergency department (OHS). Patients who were treated by the flying intervention team (FIT) on-site were excluded from the analysis. At the time of the study, FIT interventions took place during 26 weeks of the year, from 8am to 10pm. Thus, the study population examined in this research project suffered the stroke during the 26 weeks of the year without FIT deployments, at night during the period from 10pm to 8am, or during the FIT period when a FIT deployment was not possible due to other factors such as unsafe weather conditions, no available helicopter or a lack of anaesthesia personnel at the admission hospitals.

Also, we excluded patients who did not receive EVT after interhospital transfer or were lost to follow-up at 3 months (see figure 2).

Figure 2
Figure 2

Flow of patients in a study on the effect of stroke onset inside versus outside the hospital (in-hospital-stroke vs out-of-hospital-stroke) on clinical outcomes after interhospital transfer for EVT. EVT, endovascular therapy.

Baseline characteristics, procedures and outcomes

We compared baseline characteristics including age, gender, modified Rankin Scale (mRS) before stroke (pre-mRS), pre-existing conditions, such as hypertension, dyslipidaemia, atrial fibrillation, myocardial infarction, transient ischaemic attack, prior ischaemic stroke or intracerebral haemorrhage and diabetes, as well as premedication, in particular antihypertensive drugs, oral anticoagulants, statins and platelet aggregation inhibitors, between IHS and OHS patients.

Furthermore, we examined the severity and aetiology of stroke for IHS and OHS using the National Institutes of Health and Stroke Scale (NIHSS), the Alberta Stroke Programme Early CT-Score (ASPECTS) and the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification system of subtypes of ischaemic stroke.18–20

Additionally, we compared the median treatment times from both symptom onset and hospital presentation to first brain imaging, therapy decision and groin puncture between the two study groups. This fragmentation of time intervals facilitates a better identification of possible delays in the treatment process. The time of symptom onset was defined as the time of the observed onset of symptoms or the time when the patient had last been seen well. Time of hospital presentation was defined as arrival at the emergency department for OHS patients and time of symptom discovery within the hospital for IHS patients. Therefore, ‘hospital presentation’ marks the first time at which treatment processes might be initiated and prevents distortion to faster treatment times for IHS.13 21

We examined recanalisation success and complications during EVT such as dissection, embolisation to other territories, vessel perforation, contrast medium reaction, groin puncture complications or other complications. Recanalisation was assessed by the modified Thrombolysis in Cerebral Infarction (mTICI) score, which distinguishes between no reperfusion after intervention (score 0), minimal reperfusion (score 1), partial reperfusion of less than 50% (score 2a), partial reperfusion of more than 50% (score 2b) and complete reperfusion (score 3).22

The primary clinical endpoint was the mRS after 3 months. The mRS is frequently used to measure the degree of disability after suffering a stroke. It ranges from scores 0–6 and differentiates between no symptoms (score 0), no significant disability despite symptoms (score 1), slight disability with independence in daily life but inability to pursue previous activities (score 2), moderate disability with the need of assistance in daily life (score 3), moderately severe disability with need of assistance in daily body care or inability to walk (score 4), severe disability that requires constant care in bedridden patients (score 5) or death (score 6).23 We defined moderate clinical outcomes as mRS scores 0–3 and poor clinical outcomes as mRS scores 4–6, respectively. Trained documentalists collected the mRS via structured clinical interviews with patients or family members over the telephone. The collection of the mRS in structured interviews shows high test–retest reliability, interrater-reliability and construct validity.24 Telephone interviews are not inferior to a face-to-face survey.25 26 Another primary endpoint was the overall mortality (mRS score 6) 3 months after stroke. Next to the overall mortality, we differentiated between mortality during hospitalisation and mortality after hospital discharge and analysed the time of death. Furthermore, we also examined whether the cause of death was directly related to the stroke or not.

Furthermore, we collected the initial admission diagnoses of the IHS patients by surveying the medical letters of the rural admission hospitals. We divided the initial admission diagnoses into the categories ‘infectious diseases’, ‘surgical and endovascular procedures’, ‘non-interventional cardiology’, ‘neurology’ and ‘others’.

Statistical analysis

We used the software ‘R’ for our statistical analysis.27 For descriptive statistics, we calculated relative frequencies, mean values and medians. The statistical analysis comprised χ2 tests for nominally scaled data as well as Student’s t-tests for interval-scaled data. Tests were conducted at the 5% significance level. No imputational methods for missing data were used.

We performed a multivariable ordinal logistic regression to determine the conditional association of IHS with clinical outcome 3 months after stroke (3-month mRS) as well as a multivariable binary logistic regression to determine the conditional association of IHS with mortality risk. We report adjusted ORs, controlling for possible confounding variables, including age, gender, pre-existing conditions and NIHSS Score. All effect estimates were reported with 95% CIs.

This post was originally published on