Outcomes of Patients Admitted to the ICU for Acute Stroke

A Retrospective Cohort

Thibaut Carval; Charlotte Garret; Benoôt Guillon; Jean-Baptiste Lascarrou; Maëlle Martin; Jérémie Lemarié; Julien Dupeyrat; Amélie Seguin; Olivier Zambon; Jean Reignier; Emmanuel Canet


BMC Anesthesiol. 2022;22(235) 

In This Article


This study was approved by the ethics committee of the French Intensive Care Society (CE SRLF 21–09) on January 25, 2021. In accordance with French law on retrospective studies of anonymized healthcare data, informed consent was obtained from all patients and/or their next of kin. This report complies with STROBE guidelines.

Study Design, Setting, and Population

We retrospectively identified consecutive adults (≥18 years) admitted to the intensive care unit (ICU) of the Nantes University Hospital between January 1, 2014, and December 31, 2018, by searching our hospital database for International Classification of Diseases system (ICD-10) codes indicating stroke (I60.0 to I62.0; I63.0 to I63.9; or I64.0). For patients admitted more than once during the study period, we considered only the first admission. One of us (TC) reviewed the medical chart of each patient thus identified and selected those patients with confirmed stroke. Exclusion criteria were isolated subarachnoid hemorrhage, isolated subdural hematoma, and posttraumatic cerebral hemorrhage.

Data Collection and Outcomes

The following data were extracted from the ICU electronic health records (Cerner Millennium, North Kansas City, MI): baseline patient characteristics, including demographics, comorbidities, chronic medications, baseline severity indexes, clinical parameters, admission diagnosis, and stroke characteristics; radiological findings; acute-phase stroke treatments used (thrombolysis or endovascular thrombectomy, neurosurgery, or embolization); occurrence of stroke complications during the ICU stay (hydrocephalus, intracranial hypertension, seizures, status epilepticus, and/or pneumonia); treatment-limitation decisions (withdrawal or withholding of life-supporting interventions) during the ICU stay; and use of mechanical ventilation during the ICU stay. Vital status was recorded at ICU discharge, at hospital discharge, and on days 28 and 180 after ICU admission.

The neurological outcome was assessed based on the modified Rankin Scale (mRS) score on days 28 and 180 after ICU admission, as recorded in the electronic health record. Scores on the mRS can range from 0 to 6, with higher scores indicating greater disability and a score of 6 indicating death. After hospital discharge, mRS scores were assessed by a neurologist during scheduled in-person visits.


The primary objective of the study was to describe the clinical features and outcomes of patients admitted to the ICU for the management of stroke. The secondary objective was to identify predictors of a good 6-month neurological outcome, defined as an mRS score of 0 to 2 (0, no disability; 1, no significant disability; and 2, slight disability).[17,18]

Statistical Analysis

Continuous variables are described as median [interquartile range] and compared using Wilcoxon's test. Categorical variables are shown as counts (percent) and compared using Fisher's exact test. Missing data were ignored. The 6-month neurological outcome was handled as a binary variable (good, i.e., mRS 0–2; or poor, i.e., mRS > 2). Logistic regression analyses were performed to identify variables associated with a good 6-month neurological outcome. The odds ratios (ORs) were computed, with their 95% confidence intervals (95%CIs). A multivariable logistic regression model was then built to identify factors independently associated with a 6-month mRS score of 0–2. The candidate variables were selected based on the literature[16,19,20] (age, Glasgow Coma Scale [GCS] score, and hemorrhagic vs. ischemic stroke) and clinical plausibility (persistent pupillary light reflex, body temperature, and acute-phase stroke therapy). All tests were two-sided, with P values lower than 5% taken to indicate significant associations. Statistical tests were performed using the R program, version 3.5.0 (R Foundation for Statistical Computing, Vienna, Austria; www.R-project.org/).