Impact of COVID-19 Infection on the Outcome of Patients With Ischemic Stroke

Joan Martí-Fàbregas, MD, PhD; Daniel Guisado-Alonso, MD; Raquel Delgado-Mederos, MD, PhD; Alejandro Martínez-Domeño, MD; Luis Prats-Sánchez, MD, PhD; Marina Guasch-Jiménez, MD; Pere Cardona, MD; Ana Núñez-Guillén, MD; Manuel Requena, MD; Marta Rubiera, MD, PhD; Marta Olivé, MD; Alejandro Bustamante, MD; Meritxell Gomis, MD, PhD; Sergio Amaro, MD, PhD; Laura Llull, MD, PhD; Xavier Ustrell, MD; Gislaine Castilho de Oliveira, RN; Laia Seró, MD; Manuel Gomez-Choco, MD, PhD; Luis Mena, MD; Joaquín Serena, MD, PhD; Saima Bashir Viturro, MD; Francisco Purroy, MD, PhD; Mikel Vicente, MD; Ana Rodríguez-Campello, MD, PhD; Angel Ois, MD, PhD; Esther Catena, MD; Maria Carmen Garcia-Carreira, MD; Oriol Barrachina, MD; Ernest Palomeras, MD, PhD; Jerzky Krupinski, MD, PhD; Marta Almeria, MD; Josep Zaragoza, MD; Patricia Esteve, MD; Dolores Cocho, MD, PhD; Antia Moreira, MD; Cecile van Eendenburg, MD; Javier Emilio Codas, MD; Natalia Pérez de la Ossa, MD, PhD; Mercè Salvat, RN; Pol Camps-Renom, MD, PhD


Stroke. 2021;52(12):3908-3917. 

In This Article


The data that support the findings of this study are available from the corresponding author upon reasonable request.

Study Design

We conducted a prospective, observational, multicenter cohort study with 19 Hospitals in Catalonia, Spain. The recruitment of consecutive patients began between mid-March and early April 2020 depending on the Hospital, and ended on May 15th. The study was approved by the Ethics Committee of Hospital de la Santa Creu i Sant Pau as the study coordinator and thereafter, by the local Ethics Committee at each recruiting site. Patients or a legal representative gave written or verbal consent to participate. Consent for participation was waived in some centers, because the study did not include any diagnostic or therapeutic intervention and the outcomes planned are routinely registered in patients with stroke.

Study Population

Patients were eligible if they had had an acute ischemic stroke within 48 hours from the inclusion and presented a previous modified Rankin Scale (mRS) score of 0 to 3. We excluded patients presenting with (1) hemorrhagic stroke; (2) transient ischemic attack (defined as a new sudden onset of neurological deficit of ischemic origin with complete clinical recovery and absence of an acute ischemic lesion on neuroimaging); or (3) prior mRS score of >3.

Patients were classified according to the nucleic acid amplification tests (severe acute respiratory syndrome-coronavirus 2 [SARS-CoV-2] PCR from a throat swab) as COVID-19 or non-COVID-19. Patients without clinical, radiological, and epidemiological suspicion of COVID-19, who were not tested for COVID-19, were also classified as non-COVID-19. Clinical suspicion included dry cough, fever, myalgia, and hyposmia; radiological suspicion included the presence of bilateral patchy or confluent, bandlike ground-glass opacities, either on chest x-ray or chest CT scan; and epidemiological suspicion included any recent at-risk contact with a COVID-19 confirmed patient (defined by contact of 15 minutes or longer at a distance of <1.5 meters).

The following variables were recorded for all of the patients: demographic data: age and sex; vascular risk factors: previous ischemic stroke, previous intracerebral hemorrhage, arterial hypertension, diabetes, hypercholesterolemia, smoking habit, alcohol abuse, coronary artery disease, peripheral vascular disease, atrial fibrillation; drug treatment at admission (statins, antiplatelet, anticoagulants, angiotensin-converting enzyme inhibitors); prior mRS score; clinical data: National Institutes of Health Stroke Scale (NIHSS) score at admission and at 72 hours, neurological worsening defined as an increase of 4 or more points on the NIHSS score at 72 hours; imaging data: abnormal findings suggesting viral pneumonia in the chest CT and chest X-ray; Reperfusion therapies: intravenous thrombolysis, mechanical thrombectomy (MT), Thrombolysis In Cerebral Infarction scale score after MT; logistics and metrics: time from onset to admission at the Emergency Room, admission to the Stroke Unit, admission to the intensive care unit (ICU), Stroke Code activation, evaluation by a neurologist at admission, door-to-needle time (if intravenous thrombolysis), door-to-groin puncture time (if MT), days of hospitalization; Etiologic classification of stroke by the TOAST criteria (Trial of ORG 10172 in Acute Stroke Treatment).[12]

Clinical Outcomes

The primary end point was functional outcome at 3 months (±15 days), as measured by the mRS score and evaluated through a structured telephone-based interview performed by a central assessor who was unaware to whether the patient belonged or not to the COVID-19 infection group. A favourable outcome was defined depending on the previous mRS score: for patients with a previous mRS score of 0 to 2, the outcome was considered favourable when the score at discharge was 0, 1, or 2; for patients with a previous mRS score of 3, the outcome was favourable when the score at discharge was 3. The secondary outcome was mortality at 3 months. The local investigator recorded the most likely cause of death.

Statistical Analysis

First, we described the study population according to the COVID-19 status. Continuous variables were reported as means and standard deviations or medians and interquartile range (IQR) if they were not-normally distributed, as tested by the Shapiro-Wilk normality test. Categorical variables were expressed as counts and percentages. Bivariate analyses between both groups were performed using the Student t test or the Wilcoxon rank-sum test (when a nonparametric test was required) for continuous variables, and the χ 2 test for categorical variables.

Second, we performed a shift analysis using the unpaired Wilcoxon rank-sum test to demonstrate differences on the mRS score distribution at 3 months of follow-up between patients presenting or not a confirmed COVID-19 infection. We calculated also the common odds ratio of worsening of 1 point on the mRS on the presence of COVID-19 infection using ordinal logistic regression (unadjusted and adjusted for age and baseline NIHSS).

Third, we divided the study population between patients presenting favourable versus unfavourable functional outcome, as previously defined in clinical outcomes. Bivariate analyses were performed between these groups using the Student t test or the Wilcoxon rank-sum test (when a nonparametric test was required) for continuous variables, and the χ 2 test for categorical variables. Thereafter, we performed a multivariable logistic regression analysis to predict good functional outcome in our population. From an initial model including all the variables with P<0.1 in the bivariate analysis, we performed a stepwise backward regression modeling to select variables independently associated with the outcome. The final model was adjusted for potential confounders. A confounding effect was defined as an absolute change >10% in the regression coefficients when introducing the variable into the model.

Finally, we conducted a survival analysis for the secondary outcome (mortality) using Cox regression. Only variables showing a P<0.1 in the bivariate analysis were entered in the multivariable model and backward eliminated to a significance level of 0.05.

Statistical significance for all the analyses was set at 0.05 (2-sided). All the analyses were performed using Stata v.15 (TX).