Effect of the coronavirus disease 2019 (COVID-19) pandemic on stroke research

The COVID-19 pandemic has altered the normal operation of European hospitals since March 2020, and stroke care has not been an exception. Patients, professionals as well as hospital care routines have all suffered the consequences.

Health authorities, public health professionals, doctors, nurses and other health professionals suddenly changed the way they were used to work and focused all their efforts on fighting the pandemic. The ICTUSnet project was no exception, and the work of the beneficiaries was focused on real-time data analysis, evidence assessment and, finally, safe care of COVID-19 and non-COVID-19 patients according to the resources available at any time. Multiple ICTUSnet project activities were postponed until the epidemiological situation allowed them to continue.

The stroke care chain has been altered for several reasons, firstly, at the prehospital level, due to:

  • saturation in emergency services,
  • delays in seeking medical care at the onset of symptoms because of late identification of the disease as a consequence of social distance
  • delays in calling emergency systems for fear of being infected by SARS-CoV2
  • showing stroke symptoms together with respiratory symptoms of Covid19

Secondly, at the hospital level, due to:

  • the overload of the emergency department
  • the need to adapt the action protocols
  • the need to establish a protected stroke code and to know the infectious status of the patient to place him in a safe space.

Initially, professionals  did not have all the necessary protection measures. Some of them became infected, some for long periods of time. They have also suffered stress and, in addition, the care overload has prevented attention to other non-care tasks.

On the other hand, although the main symptoms of coronavirus infection are respiratory, it has been detected that cerebrovascular disease can be a clinical manifestation of COVID because the increased risk of vascular events in these patients. and stroke COVID related could be more severe with poor outcomes.

At the Universitary Hospital Son Espases (IdISBa partner), we have compared stroke care   from March 1 to May 31, 2020 to the same period of time from  previous year  founding no differences in the number of patients seen nor in the number of cerebral reperfusion treatments performed in the ischemic stroke; we did detect an increase symptoms onset time   upon arrival at the hospital in those patients who came to the hospital on their own, and also in the time from arrival at the hospital to the start of mechanical thrombectomy (door-introducer), attributed to the established safety protocol.

Today our work has been adapted to guarantee excellent care for patients affected by cerebrovascular disease, whether or not they are infected with SARS-CoV2, and to provide the maximum safety measures for healthcare professionals and patients.

The impact of the pandemic on the care of each region has been different depending on: COVID-19 incidence, stroke incidence, and  health resources adaptation dedicated to this pathology (e.g. critical beds have been able to include stroke unit beds).

For all above mentioned, it is likely that Covid 19 has affected the ICTUSnet project not only in its execution, with the deadlines  extension , but also in its result (e.g.  WP1 takes into account the action times).

Stroke management in North Portugal

Acute stroke is an emergency that requires a fast, time-sensitive treatment. If not promptly treated, it may cause severe neurologic impairment, resulting in limited functional status and loss of quality of life.  Therefore, an organized system of care is essential, focusing on prevention primary prevention, prehospital care, acute phase treatment, rehabilitation and community reintegration.  

In Portugal, stroke is the leading cause of death and permanent disability, with an important socioeconomic impact affecting all regions. The Northern Regional Health Administration (ARS Norte) is a political identity responsible for strategic management of population’s health in the north of Portugal, supervising 10 hospital centres and 6 hospitals. These include 7 Primary Stroke Centres offering intravenous thrombolytic treatment and 4 Comprehensive Stroke Centres (CSC) where additionally mechanical thrombectomy can be performed.
A patient with a suspected stroke is transported by an emergency team to the nearest hospital with a structured stroke code. In the hospital the patient is evaluated by a stroke team according to a guideline based care. The patient then performs brain imaging exams and, if indicated, is treated with acute phase therapies, such as intravenous thrombolysis and mechanical thrombectomy (the last one in CSC). This chain sometimes implies hospital transfers that may last minutes to hours, depending on the distance and available means to take the patient from one PSC to a CSC. 

In order to improve the coordination between the several agents involved in the stroke care pathway, ARS Norte created a team with elements from the 11 hospitals. The conception of a local registry became an important initiative to revise current practices and propose new strategies to better treat patients in this region. The participation as an ICTUSnet partner will help to achieve these goals, especially in those patients treated with mechanical thrombectomy. 

Un vistazo a la heterogeneidad de los informes de alta hospitalaria de enfermos diagnosticados con ictus

Una de las tareas del proyecto ICTUSnet es el desarrollo de herramientas de minería de textos (TM) para asistir en el proceso de extracción de información relevante para la evaluación de la calidad asistencial. La información se extrae de los informes de alta hospitalaria de pacientes diagnosticados con ictus y se compila en un registro centralizado. La unidad de Text Mining del Barcelona Supercomputing Center lidera el desarrollo de estas herramientas dentro del proyecto.

Para esta tarea disponíamos de 2696 informes de alta de AQuAS y 611 informes del Hospital Universitario Son Espases (HUSE) y lo primero que hicimos fue seleccionar el subconjunto de informes a anotar para, después, poder entrenar y evaluar los algoritmos de TM.

HUSE da cobertura a la población de las Islas Baleares mientras que los datos de AQuAS proceden de los 26 hospitales que dan cobertura a la población de Cataluña. Así que decidimos usar la proporción de la población a la que da cobertura cada una de las fuentes, con lo que el 13% de los informes que forman el corpus se seleccionaron del HUSE y el 87% de AQuAS .

Para los datos de AQuAS, quisimos mantener también la heterogeneidad (proceden de 26 hospitales distintos), y para ello separamos los documentos en clusters (ver detalles del clustering al final). Este ejercicio nos ha permitido tener una visión de la heterogeneidad de los informes. En el gráfico siguiente podemos ver todos los informes proyectados en el plano donde los colores representan los diferentes cluster (8 en total).

Al mirar con detalle los datos, observamos que los clusters se corresponden de manera bastante fiel a la procedencia de los informes. Así, el cluster verde (2) está constituido exclusivamente por los 164 informes del HUSE, mientras que el cluster rojo (3) contiene 379 informes del hospital K3 que representan el 99.74% del cluster. A su vez, el cluster lila (4) contiene 249 informes del hospital N1 representando el 81,85% del cluster. El cluster naranja (1) está repartido (casi mitad y mitad) por documentos de los hospitales H2 y Z.

El grupo formado por los clusters 0,6 y 7 recoge informes de hasta 24 hospitales diferentes, 19 de los cuales en un porcentaje superior al 90% (es decir, más del 90% de los informes del hospital pertenecen a este grupo).

Observamos una tendencia evidente a la concentración de los informes de un determinado hospital en un único cluster. Así, de media, el 89,6% de los informes de un hospital se concentran en un mismo cluster, siendo la concentración mayor la del 100% y la menor la del 52% (72% si agrupamos los clusters 0, 6 y 7 en un único grupo).

Para realizar el clustering hemos seguidos los siguientes pasos:

  • Utilizacion la implementación de Gensim de doc2vec con los embeddings médicos en Soares, Felipe, et al. “Medical word embeddings for Spanish: Development and evaluation.” Proceedings of the 2nd Clinical Natural Language Processing Workshop. 2019.
  • Una vez generados los vectores de cada documento, utilizamos UMAP supervisado para reducir la dimensionalidad, indicando las dos fuentes de datos como el label de cada ID, y usando la distancia de coseno. Establecemos random_state = 0 y el resto de parámetros por defecto.
  • Sobre este output aplicamos K-means con k=8 y random_state = 0. Todos los documentos de HUSE caen en el mismo cluster, y ninguno de AQuAS cae en este cluster.

Stroke care pathways at CHU de Montpellier

Background design of human feature lines and symbolic elements on the subject of human mind, consciousness, imagination, science and creativity

Multiple randomized controlled trials have recently proved the superiority of mechanical thrombectomy compared to the best medical therapy in patients with recent cerebral infarct and large vessel occlusion in the anterior circulation. There are 6 Stroke Units in the Occitanie-East region, formerly Languedoc-Roussillon. Only one of these units, the Comprehensive Stroke Center (CSC) of the Montpellier hospital, has a centre of interventional neuroradiology (NRI) that can perform mechanical thrombectomy.

Patients with a suspicion of stroke are referred to the nearest stroke unit, then the candidates for mechanical thrombectomy are transferred secondarily to the CSC in Montpellier. Currently, there is no evidence that initial management in a proximity stroke unit and a secondary transfer for a mechanical thrombectomy is associated with a less positive outcome, compared to patients initially treated at the regional stroke centre.

The objective of our local clinical research registry is to compare the clinical outcome of patients first admitted in the CSC of Montpellier with those transferred from the Proximity Stroke Unit. In this way, the implementation of a local database will allow us to generate information and knowledge about stroke patients’ management in our region in order to contribute to the achievement of Ictusnet objectives.

The aim of CHU Montpellier’s participation in ICTUSnet as partners is to describe the management of suspicion of large vessel occlusion in our region, compared to other participating regions. In order to improve stroke patients’ management, we need to evaluate the delay of mechanical thrombectomy and clinical outcome of patients after 3 months

What should we do to overcome barriers in Stroke care and urgent prehospital care?

The organization of acute stroke treatment services at the regional/national level is key to guaranteeing reperfusion therapies reach the whole population.

A recent paper published in the European Stroke Journal (1), the scientific journal of the European Stroke Organization (ESO), describes the results of a recent survey performed in 44 European countries. The majority of the countries have stroke care plans and most of them take responsibility for the organisation/implementation of stroke systems of care (86%), quality of care assessment (77%), and act as a liaison between emergency medical systems and stroke physicians (79%). As for stroke systems of care, the focus is mainly on prehospital and in-hospital acute stroke care (the Code Stroke systems are available in the 84%). The survey also shows that the preferred urgent transport is via non-medicalised ambulances. The paper highlights that the presence of stroke care plans, the availability of stroke registries, the transport of urgent stroke patients via non-medicalised ambulances, and the drip-and-ship routing (the route where patients are firstly assisted at the reference stroke centre, and then sent to the comprehensive stroke centres) of acute patients showed higher reperfusion treatment rates.

However, stroke is not yet a priority everywhere in Europe, which is a barrier to the spread and delivery of high-quality stroke care.

This April, ICTUSnet members will discuss Acute Stroke Care Plans among the different South-western regions in the first ICTUSnet workshop, to be held in Palma de Mallorca. We believe that ICTUSnet project will overcome different barriers by creating a permanent network that promotes collaborative research, incorporates new technologies for data analysis, and fosters mutual learning across SUDOE regions.

  • Abilleira S et al. Planning of stroke care and urgent prehospital care across Europe: Results of the ESO/ESMINT/EAN/SAFE Survey. European Stroke Journal 0(0) 1–8. (Available Online). https://doi.org/10.1177/2396987319837106