Introduction

There is an ongoing debate concerning the prehospital pathway of stroke patients: should patients in remote areas with suspected stroke be admitted to the nearest primary stroke center first to receive intravenous thrombolysis (IVT), if eligible and then be secondarily transferred (ST) to a comprehensive stroke center (CSC) for endovascular thrombectomy (EVT) or should they be directly admitted (DA) to a CSC [1, 2]?

In Baden-Württemberg, a state in southwest Germany with an area of 35,751 km2 and 11 million inhabitants, the incidence of ischemic stroke was 254 per 100,000 inhabitants in 2014. Of these patients 2.4% underwent EVT at 14 centers. The rate of EVT throughout Germany was 2.3% in 2014. Approximately 140 stroke units are available in Baden-Württemberg, distributed by the Ministry of Social Affairs and therefore assuring a sufficient geographical coverage [3, 4]. Emergency physicians and paramedics are required by law to transfer patients with suspected stroke to the next primary stroke center regardless of the stroke severity. This regulation was introduced when IVT was the only treatment for acute ischemic stroke. Now that EVT has been shown to be more effective in proximal arterial occlusion, the abovementioned regulation carries the risk of delaying EVT in these patients. This retrospective study therefore aimed at analyzing the stroke registry of Baden-Württemberg regarding transfer times and clinical outcome in the DA and ST pathways.

Methods

Study Design

This was a retrospective observational cohort study based on a prospectively maintained regional stroke registry. All stroke centers in Baden-Württemberg are required to contribute data to a prospective registry without the need of informed consent. This study is therefore exempt from institutional review board approval. The stroke registry was designed in 2004 and has been maintained for quality assurance of IVT treatment; however, some parameters such as occlusion site, time of groin puncture and modified Rankin scale (mRS) score at 90 days after stroke onset are not documented.

Patient Selection

Registry data of patients treated between January 2014 and December 2017 were analyzed. Inclusion criteria were treatment with EVT, premorbid mRS 0–2 and admission to a CSC within 24 h after stroke onset. Patients with missing discharge mRS were excluded.

Outcome Measures

Primary outcome parameters were time from onset to admission at a CSC and good clinical outcome defined as a discharge mRS of 0–2. Secondary outcome parameters were discharge mRS score, mRS shift (difference between premorbid mRS and mRS on discharge), hospital mortality and occurrence of symptomatic intracranial hemorrhage (any intracranial hemorrhage associated with neurological deterioration).

Statistical Analysis

Statistical analysis was performed using R version 3.4.3 (R, Open Source). Comparisons between DA and ST were performed using Student’s t-test, Mann-Whitney U-test and χ2-test. All variables were entered into a univariate analysis to identify possible predictors of clinical outcome. Variables with P < 0.05 were then included in a multivariate logistic regression analysis to identify independent predictors of good clinical outcome. Odds ratios (OR) and 95% confidence intervals (CI) were estimated. A P-value of <0.05 was considered statistically significant.

Results

In total, 2797 patients were selected for analysis (Fig. 1), of which 59.2% were admitted directly to a CSC (DA group).

Fig. 1
figure 1

Flow chart of excluded and included patients

Full size image

Patients in the DA group were admitted significantly earlier to a CSC (median 102 vs. 210 min, P < 0.001) compared to ST and received IVT more frequently (61.9% vs. 48.0%, P < 0.001). Out of 2797 patients, 1051 (37.6%) achieved a good outcome. In the DA group the proportion of good outcome was significantly higher (DA 42.2% vs. ST 30.9%, P < 0.001) and median discharge mRS (DA 3 vs. ST 4, P < 0.001) and median mRS shift (DA 3 vs. ST 4, P < 0.001) were significantly lower (Fig. 2). The rate of hospital mortality was significantly lower in the DA group (DA 17.7% vs. ST 24.3%, P < 0.001) and the proportion of intracranial hemorrhage was similar in both groups (DA 9.3% vs. ST 7.5%, P = 0.101) (Table 1).

Table 1 Patient characteristics
Full size table

Table 2 compares the characteristics of patients with good and poor outcome. When including all variables with P < 0.05 in a multivariate logistic analysis, direct admission to a CSC was an independent predictor of good clinical outcome (adjusted OR 1.32, CI 1.09–1.60, P = 0.004), independent of the National Institutes of Health Stroke Scale score. Bridging IVT was not a predictor of outcome (P = 0.382) (Table 3).

Table 2 Univariate analysis
Full size table
Table 3 Multivariate analysis
Full size table

Discussion

This retrospective study analyzed a regional stroke registry and compared DA and ST regarding time metrics and short-term clinical outcome in 2797 patients. The data show that secondary transfer to a CSC takes more than twice as much time as direct admission, even in a small state like Baden-Württemberg. Similar results have been published by Park et al. and Prothmann et al. (Table 4; [5, 6]). Hence, interhospital transfer leads to a considerable delay of EVT in ST patients [7, 8]. Mokin et al. reported that one out of three patients becomes ineligible for thrombectomy because of unfavorable Alberta Stroke Program Early CT Score (ASPECTS) worsening following interhospital transfer [9].

Table 4 Time from onset to admission at CSC
Full size table

According to the present results, patients in the DA group have better clinical outcome and a lower rate of hospital mortality compared to ST patients. Direct admission was an independent predictor of good outcome. Hence, this study with 2797 patients confirms a recently published meta-analysis on DA vs. ST with 2068 patients [5, 6, 10,11,12,13,14,15,16]. This is an important finding, because in many German states including Baden-Württemberg emergency doctors and paramedics are required to transfer patients to the nearest hospitals certified for IVT. This regulation was issued when IVT was the only treatment option of ischemic stroke. Nowadays it is well known that IVT is unable to remove long thrombi [17], which led to the development and refinement of mechanical thrombectomy devices. Several randomized controlled trials have demonstrated that EVT with or without IVT is superior to IVT alone [18]; however, EVT requires more technical equipment and specialized staff compared to IVT and can only be provided in CSC.

These results have two important implications: first, emergency doctors and paramedics should be permitted and trained to identify those patients that are more likely to suffer from large vessel occlusion and therefore requiring direct transfer to a CSC for EVT. Simplified stroke severity scores can be helpful here, because large vessel occlusions are usually associated with a higher stroke severity [19]. Transcranial doppler ultrasound performed by emergency physicians is a potential option as well; however, it is accompanied by several obstacles, such as an insufficient temporal bone window in many patients and an extensive training, which is necessary for adequate Doppler ultrasound [20].

Fig. 2
figure 2

Scores on modified Rankin scale at discharge. Directly admitted patients had a significantly higher rate of good clinical outcome on discharge (mRS 0–2) (P < 0.001)

Full size image

Second, interhospital transfer times need to be improved. There are several reasons for this time delay, such as inefficiencies in triaging and availability of patient transport [21, 22]. Further studies are necessary to determine where exactly time is unnecessarily lost but this can be different in every hospital. Admitting stroke patients with suspected large vessel occlusion directly to a CSC may put primary stroke centers under financial pressure; however, less than 3% of stroke patients in Germany underwent EVT in 2014 [3]. Therefore, this should not be a matter of concern. Besides, primary stroke centers are still important for a nationwide stroke unit coverage. Besides DA and ST, there is a novel triage concept in which neurointerventionalists are transferred to primary stroke centers for EVT [23,24,25,26]. Further trials including this triaging option should be considered as well when reforming triage pathways in a stroke network.

The major strength of this study is its size with 2797 patients. Although its size exceeds those of previous studies on DA and ST, it has several limitations. A significant weakness is the retrospective and nonrandomized nature of this study. It is a potential bias that patients admitted to a primary stroke center were only selected for transfer to a CSC when they were severely affected, and therefore a priori had a lower chance of good outcome, whereas patients that were less severely affected were nevertheless included in the MS pathway. This is reflected by the higher stroke severity at baseline in the DS group. Out of 4312 patients 1199 (27.8%) had to be excluded due to missing discharge mRS, mostly in patients who were transferred to another hospital after thrombectomy, which is a potential bias but the results are in accordance with previous studies [10]. Furthermore, parameters such as time from stroke onset to groin puncture, occlusion site and recanalization success were not available, which might also cause a bias. Moreover, discharge mRS but not 90-day mRS was available in this stroke registry. Nonetheless, early mRS has been reported to strongly correlate with 90-day mRS scores [27].

Conclusion

The results confirm previous studies and show that time from onset to admission at a CSC in ST patients is more than twice as much compared to DA patients and associated with worse outcome. Emergency physicians should be allowed to transport stroke patients to any institution they think is most appropriate and should not be restricted by any regulations.