Acute ischemic stroke treatment model for Poland in the mechanical thrombectomy era – which way to go?

Transportation delays

In Poland, where drip-and-ship (DS) is the dominant model, delays are not only the result of long distances between PSCs and CSCs, but also due to the common problem of ambulance availability. Establishing a ‘standby ambulance’ service at every hospital seems to be the best ad hoc remedy, yet is unlikely to be promptly employed in Poland for economic reasons. Rapid transportation is of utmost importance in acute ischemic stroke, where 1.9 million neurons die every minute [27]. Not only is treatment delay associated with poorer functional outcomes [28], but also long distance transport seriously affects survival [29]. A more comprehensive and future-proof solution to the problem would be to build a network of Thrombectomy-Capable Stroke Centers so that the transport of patients from anywhere in the region does not exceed 1 h. This appears to be the only solution that truly addresses the problem since even in very well organized centralized systems, inter-hospital transfers prolong onset-to-arrival time by over 140 min [30].

Better prehospital LVO stroke detection

Since clinical outcomes for both IVT and MT are highly time-dependent [30, 31], it is crucial to promptly separate potential LVO patients to ensure their timely access to endovascular treatment (EVT). Given that mobile computed tomography (CT) ambulances are not widely available, multiple clinical assessment LVO scales have been developed [32]. These scales differ with respect to sensitivity and specificity, with a positive predictive value of around 50%, and with the highest predictive power to detect LVO observed in the VAN (Vision, Aphasia and Neglect), LAMS (Los Angeles Motor Scale), NIHSS (The National Institutes of Health Stroke Scale) and RACE (Rapid Arterial OCclusion Evaluation) scales [7]. Despite the lack of evidence to support the use of clinical stroke scales in routine pre-hospital triage [7], it seems reasonable to select and introduce one of them into ambulance crew training nationwide to facilitate the triage process and decision-making. Creating additional MT capable centers would reduce the risk of inappropriately classified patient accumulation in scarce CSCs.

Information flow

In the situation of acute brain ischemia, time plays a crucial role and information flow must be addressed. Establishing reliable liaisons between ambulance services and stroke centers with a prompt pre-notification process is the cornerstone of an efficient network [33]. A good system should include established alarm telephone numbers and decision paths for both ambulance services and acute stroke neurology departments. Smartphone-based telephone and tele-transmission consultations from an ambulance may help to make a timely and correct decision where to refer the patient first [34]. The ideal solution would be to create a single network of internet communication channels between PSC and MT centers in order to quickly transfer data and make swift decisions (e.g., e-stroke mobile application). Research is currently underway on modern technologies such as telemedicine, biomarkers and infrared screening devices to improve pre-hospital differentiation between ischemic stroke and stroke mimics [35]. Both centralized and decentralized systems will need to address all these issues.

Radiology services

Efficient stroke treatment cannot exist without good quality radiology services. In the MT Pilot Program, many centers reported far from ideal computed tomography/computed tomography angiography (CT/CTA) services at PSCs [12]. Delays or a lack of precise radiological assessment (particularly where teleradiology was involved) was a matter of serious concern. It has been postulated that standardized protocols for CT/CTA should be established in each stroke center to facilitate swift decision-making by consulting MT teams. The planned introduction of the POLCARD-financed advanced CTA postprocessing system called Brainomix may further improve and speed up the process. In a decentralized system, some of these issues would be bypassed, since many more patients would go directly to a Thrombectomy-Capable Stroke Center, where consultant radiologists would be accustomed to operating along MT operators and there would be fewer problems with communication.

Local stroke units crisis

It has been reported that, due to shortages of personnel, local stroke units started to disappear in some areas [12]. It would require a fundamental change in reimbursement philosophy, backed by patient lobbying groups, but it seems obvious that creating more and better equipped Thrombectomy-Capable Stroke Centers would also be a partial remedy to this unfortunate situation. TCSC would provide more comprehensive services to stroke patients providing both IVT and MT locally.

Nursing staff

We have been facing unprecedented service disruptions due to personnel shortages and a lack of proper funding [12]. Acute stroke patients require very close monitoring and it must be emphasized that nursing care in stroke units involves different staffing standards and policy than in other departments. Government agencies responsible for reimbursement must be made aware of the specific situation of stroke unit patients. Unless this issue is resolved, we are about to witness further loss of key personnel and growing instability of the system.

Treatment model selection

Today there are three main treatment models of LVO-suspected patients: mothership (MS), drip-and-ship (DS) and drive-a-doctor/drip-and-drive (DD). So far there is no strong evidence of a clear advantage of either of the transportation models (DS vs. MS) and they should be applied according to local organizational logistics and individual characteristics of the patient while other models are still in the early phase of assessment. Presently the MS model is favored in highly populated areas where patient transportation to the CSC is below 30–45 min and DS is preferred above this time limit [7].

New paradigm in operator training

Unfortunately, training MT operators is still a divisive and political issue. There is no doubt that the shortage of MT-trained operators is a great challenge [20, 26]. The growing demand for timely and effective stroke treatment in Poland was a stimulus for the extension of the spectrum of MT operators well beyond the few neuroradiologists by the Minister of Health [37]. With MT available only in scarce neuroradiology-based CSCs, it may be considered the main reason for inadequate access of patients to proper treatment [38]. There is growing understanding, also strongly supported by the World Federation for Interventional Stroke Treatment (WIST), that interventional cardiologists with previous peripheral and carotid expertise will be quick to obtain MT skills and therefore their training may be much shorter [38, 39]. For experienced interventionalists, training at neuroradiology centers of excellence along with structured supervised silicone and cadaveric model exercise may be the best and shortest way to achieve necessary MT competence without compromising patient safety [40].

The scale of the problem

The Polish National Health Fund (NFZ) monitors closely numbers of patients with acute stroke and present quality of care delivered with regard to thrombolysis and mechanical thrombectomy. We know well the number and costs of the provided therapies from the online NFZ databases. What we are completely unaware of is the cost associated with delayed or abandoned treatments resulting from untimely access to treatment. We realize that there are many patients who have been failed by the present system, but since they are not reported, we cannot see the scale of the need that is unmet [39, 41, 42]. Given the number of mechanical thrombectomy procedures performed in other healthcare systems in the world, we can only estimate that we provide good services to less than half of the eligible patients [13–17]. In 2020, approximately 90,000 patients diagnosed with acute stroke were hospitalized in Poland, with roughly 75,000 identified as ischemic. There were 2320 MT procedures performed within 12 months from November 2019, which makes around 3% of all ischemic strokes (online NFZ data for 2019 and 2020). In the UK, with a similar number of 95,000 stroke patients hospitalized yearly, 83,000 were diagnosed to be ischemic. Among those, 33,000 cases were diagnosed to be LVO strokes, of which 11,580 were thrombectomy-eligible (the sum of patients presenting with LVO within 6 h from onset and those presenting later with favorable perfusion criteria), which makes 13% of all ischemic strokes [43]. Germany with its staggering 150 MT-center-strong network treats 7.2% of ischemic strokes invasively [13], and the Czech Republic even more (7.5–8.1%) [16]. France, with almost twice as many MT centers at its disposal (2018) as Poland today, decided to implement considerable changes to improve access to endovascular treatment. Two groundbreaking postulates emerged, namely training operators outside the field of neurointervention to perform MT and creating additional, smaller and closer to the patient Thrombectomy-Capable Stroke Centers [44]. The newest European Stroke Organization (ESO) Guidelines on intravenous thrombolysis for acute ischemic stroke (2021) recommend extending the therapeutic time window for intravenous recombinant tissue plasminogen activator (IV rtPA) and MT to 4.5–9 h with CT/magnetic resonance imaging core/perfusion mismatch, so the expected number of MT-eligible patients may increase even more in future [45]. Acute ischemic stroke patient pathways in different transportation models are presented in Figure 1.

Figure 1

Acute ischemic stroke patient pathways in different transportation models

CT – computed tomography, CTA – computed tomography angiography, PSC – Primary Stroke Center, CSC – Comprehensive Stroke Center, LVO – large vessel occlusion.

What needs to be done

The detrimental effect of time passage in the acute ischemic stroke situation was clearly shown in the MR CLEAN trial, where each hour of delay resulted in a 7% drop in probability of a good clinical outcome [1]. Patients revascularized early, within 2 h of symptom onset, had a 33% absolute difference in good outcome compared to the control. This difference plummeted to 6.5% after 6 h of symptoms [46]. The selection of the best model of a stroke network and transfer system for Poland is urgent. There is no European Stroke Organisation (ESO) recommendation as to which of the two dominant transfer models (MS or DS) of LVO-suspected stroke patients should be favored; thus the choice ought to be made based on patient characteristics and regional service organization [7]. Comparison of time intervals in two dominant transportation models (MS and DS) is presented in Figure 2. The MS model may be preferred in urban areas where transport to the MT center is less than 30–45 min, while patients beyond this time limit may benefit more from DS [7]. Observational studies suggest that for suspected LVO stroke patients the MS model offers better functional outcomes [47–53]. The mothership model, compared with DS, offers shorter transfer times, 275 vs. 179.5 min, respectively [31]. Associated transfer delays result not only in worse clinical outcomes [54], but in a recent large study increased mortality as well [29]. In the ETIS registry, the mothership paradigm outperformed the drip-and-ship model with respect to 90-day functional independence (modified Rankin Score – mRS 0-2) 60% vs. 52%, respectively, as well as for excellent outcome (mRS 0-1), which was 9.6% higher in the MS model [55]. This analysis also assessed the clinical impact of both (MS and DS) models with respect to the PSC-CSC distance as well as the time interval between imaging and groin puncture. The MS paradigm was better for patients transported over longer distances (> 12.5 miles from PSC to CSC), or when the time from CT to puncture exceeded 140 min. In this analysis, despite the relatively short mean time of inter-hospital transport (43.5 min), no advantage of the drip-and-ship model was observed for patients localized further from CSC. The authors note that simple switching from DS to MS mode for LVO-suspected patients would not improve outcomes for all, since those who live in remote rural areas will always have longer travel times [55]. In another study, direct transfer to MT centers (MS) showed that thrombolysis would be delayed by only 12 min, while thrombectomy could be performed 91 min sooner compared to DS [48]. Researchers suggest that LVO-suspected patients should be treated by the MS model if the door-in-door-out (DIDO) time in PSC exceeds 40 min [56], or time from imaging to groin puncture exceeds 90 min [57]. In a recent meta-analysis of 7824 patients, the MS and DS models were compared [58]. There was no significant difference in onset-to-thrombolysis time, but in the MS model, the mean stroke onset-to-puncture time was significantly shorter, functional outcome was better, and the risk of symptomatic ICH was lower. The comparison of the DS and MS models has also been tested in the Polish healthcare environment [59]. Again, MS vs. DS comparison resulted in, as expected, significantly shorter times from onset-to-groin puncture (85-minute difference) and CT-to-groin puncture (84-minute difference), but there was no statistically significant improvement in functional outcomes (mRS 0-2) for direct transfer. Still, excellent functional outcome (mRS 0-1 at 90 days) was superior for the MS model compared to DS (33 vs. 23%, respectively). Karlinski in his commentary on this study concluded that even modest benefits observed in the MS model in the Polish healthcare system pointed to a potential advantage of this paradigm, should the MT system be better organized and reimbursed [60]. Schlemm used mathematical modeling to calculate additional-delay-to-thrombolysis thresholds associated with the greatest reduction in disability-adjusted life years and concluded that patients suspected of LVO ischemic stroke should be treated with the MS paradigm if additional delay to thrombolysis was < 30 min in urban and < 50 min in rural settings [61]. The notion that the MS paradigm trumps DS was challenged by the surprising results of a recent randomized controlled trial (RACECAT) in an exceptionally well-organized stroke network in Catalonia (unpublished). Good outcome defined as mRS 0-2, as well as mortality at 90 days, was similar in DS and MS groups. However, there is a significant caveat to this study: these results may be hard to replicate, since in this particularly well-organized stroke network, crucial time measures were, unlike anywhere else in the world, comparable in MS and DS models.

Figure 2

Comparison of time intervals in two dominant transportation models

PSC – primary stroke center, CSC – comprehensive stroke center, DS – drip-and-ship, MS – mothership.

Drip-and-ship is the preferred model in Polish reality today [12]. A glance at the location of the MT Pilot Program CSC centers on a map of Poland reveals the reason: there are many regions with long travel distances.

With only 20 CSCs providing MT, many towns lie outside a 70-kilometer circle to the nearest MT stroke center (e.g., Suwałki, Płock, Konin, Tomaszów Lubelski, Sandomierz, Jelenia Góra, Ostrów Wielkopolski, Piła, Koszalin, Słupsk, Ostrołęka, Mława, Giżycko, Sierpc, Wałcz, Szczecinek, Gorzów Wielkopolski). It may be estimated that ambulance transfer from these places to the nearest MT centers takes at least 2 h. Similar findings were reported in the HERMES meta-analysis, but it is also the real life experience of well-developed MT networks [30, 62]. Helicopter transport may be considered an option but in reality this is often unavailable on time due to other emergencies it constantly covers. Reports from Western Europe indicate that barely 4% of patients with acute stroke are transported this way [63]. It needs to be emphasized that transfer time is not transport time. The latter is around 36% of the transfer time, which also includes patient-associated delay, time to ambulance arrival on scene, on-site assessment time, telephone contact by ambulance crew with Stroke Center, and all delays associated with pre-hospital measures taken [64, 65]. When 50-kilometer circles are applied to 17 main regions on a map of Poland (Figure 3), most voivodships will accommodate at least 3 such circles. This rough estimate by no means indicates that the transport time within the circle would be limited to 1 h, since geography and road congestion are to be considered. Yet, even such a crude assessment gives us some information about transport difficulties within the present MT network.

Figure 3

‘No country for old men with stroke’ – map of Poland’s presently active 20 Comprehensive Stroke Centers (CSC), surrounded by 50-km radius areas with around 60% of country’s area potentially beyond fast access to mechanical thrombectomy (MT) services

There are sophisticated modeling programs (e.g. Destine Health) which help construct the most suitable transport models for stroke patients for specific regions and geography [66, 67]. They take into account geographical conditions, the efficiency of the local hospital system, and the best transportation system available on site. These algorithms include as specific data as different LVO screening stroke scales, crucial in-hospital and inter-hospital time intervals such as ‘door-to-needle’, ‘door-in-door-out’, ‘door-to-groin puncture’ and even take into account the proportion of patients being treated with IVT and MT in the region of interest. The modeling programs provide their user with an interactive map which helps predict the likelihood of a good outcome with different sets of transport protocols. This software has been successfully introduced in several different countries (e.g., USA, Japan, UK, Sweden and Germany) and may be useful when creating new or planning how to reorganize the existing stroke networks [https://destinehealth.com/].

Most centers taking part in the Polish Pilot MT Program were active participants of the recent 7^th Symposium on Ischemic Stroke [12] and had the opportunity to share their experiences and name problems. So far, the DS model is a default in Poland, and for obvious reasons: transfer times between PSCs and CSCs are long. It seems impossible to improve patients’ timely access to treatment in either the MS or DS models without further expansion of the MT network. We believe that the discussion which model of treatment, DS or MS, would be better for Poland in 2022, is a substitute one. Recent research suggests that in most instances, in LVO-suspected patients, the MS model should be preferred since it offers a much better chance of a good clinical outcome. With the present limited network of MT Comprehensive Stroke Centers in Poland, we are highly unlikely to improve. In fact, we do need to considerably increase the number of Thrombectomy-Capable Stroke Centers at the cost of Primary Stroke Centers and bring services closer to the patient [47, 68–70]. With approximately 75,000 ischemic strokes diagnosed in Poland, we would have to perform 5,250–7,500 interventions yearly to achieve 7–10% of strokes treated by MT. If we managed to extend the MT network to 3–5 per voivodship, the number of centers would rise to 50–85. With 50 active MT centers (CSC and TCSC), approximately 105–150 thrombectomies per center would be performed yearly, not far from the currently observed mean value of 116 MT per center. This estimated case volume would put them in the high-volume range according to the Mission Thrombectomy 2020+ Global Executive Committee report prepared in cooperation with the Society of Vascular and Interventional Neurology (available online), established operator/center volumes recommendations [69], and a recent nationwide analysis in the US [21] with clear benefits of preserved patient safety, treatment burden more equally spread and improved outcomes. Considering the present shortage of operators and the time required for necessary MT training, it seems essential for specialists from various fields of medicine to get involved [16, 24, 41, 42].