Materials and Methods
Study Design
We conducted a retrospective analysis of a multicentre, international observational dataset to ascertain the prevalence and characteristics of neurological manifestations at hospital admission and the occurrence of neurological complications during hospitalization. Data were collected according to the ISARIC-WHO Clinical Characterization Protocol (CCP), a prospective study of hospitalized patients that aims to characterize emerging infections.[14] Study sites aimed to enroll as many hospitalized individuals with COVID-19 as possible, according to locally available resources. Individuals with laboratory-confirmed severe acute respiratory distress syndrome coronavirus-2 (SARS-CoV-2) infection and hospitalized [or admitted to the intensive care unit (ICU) according to site implementation] were enrolled. A small number of sites recruited only patients admitted to ICU (Supplementary Table 1). Where resource constraints limited recruitment, sites were advised to utilize recruitment strategies to minimize bias.
Of these individuals, 261 161 were evaluated, as of 25 May 2021, for neurological manifestations and in-hospital neurological complications by clinical teams at study sites. The study was approved by the World Health Organization Ethics Review Committee (RPC571 and RPC572). Local ethics approval was obtained for each participating country and site according to local requirements. Informed consent was taken in most settings, according to locally approved procedures, or waivers were granted. De-identified data were submitted to the ISARIC database by direct entry to Research Electronic Data Capture (REDCap, version 8.11.11, Vanderbilt University, Nashville, TN) hosted by the University of Oxford or by secure file transfer when locally managed data collection systems were used. All data submitted to the ISARIC data platform were harmonized to the CDISC SDTM standard (Study Data Tabulation Model; version 1.7, Clinical Data Interchange Standards Consortium, Austin, TX). Available data included demographics, comorbidities, signs and symptoms, clinical assessments, laboratory data, medications, procedures and outcomes. Glasgow Coma Score (GCS) was collected as part of the neurological baseline variable at admission. The study protocol and case report forms (CRFs) are available online https://isaric.org/research/covid-19-clinical-research-resources/ (ISARIC CCP and ISARIC CRF, respectively).
Cohorts
The study cohort for analysis included all patients of any age enrolled in the ISARIC/WHO global database with laboratory-confirmed COVID-19 infection who were hospitalized between 30 January 2020 and 25 May 2021. Children were defined as those less than 18 years of age. Completed analyses reported outcomes for all patients, in addition to stratification by critical care, defined as admission to ICU at any time during hospitalization. We excluded patients who were missing information on hospital admission and discharge dates, ICU admission or neurological manifestations/complications (Figure 1). Availability of data on neurological variables is summarized in Supplementary Table 1. Cohort characteristics by geographic region and income classification are summarized in Supplementary Table 2. A detailed description of characteristics for all patients included in this dataset is available online (https://doi.org/10.1101/2020.07.17.20155218).
Figure 1.
Origin of study cohorts and breakdown of subgroups (ICU, non-ICU, adult and children).
Selected characteristics documented at hospital admission and during hospitalization were summarized for all patients and grouped by whether patients were admitted to the ICU. In addition to ICU admission, we assessed the severity of critical illness and outcomes when invasive mechanical ventilation or extracorporeal membrane oxygenation (ECMO) was needed for support.
Definitions
Neurological manifestations of COVID-19 at admission that were reported in CRFs included altered consciousness, fatigue, anosmia, dysgeusia, myalgia and seizure at admission. CRF-reported in-hospital neurological complications were CNS infection (meningitis/encephalitis), new seizures during hospitalization and stroke.
Outcomes
The primary outcome was the description of neurological manifestations present on admission and in-hospital neurological complications in children and adults. The secondary outcome was in-hospital mortality, accounting for associations with known risk factors, trends over time and in-hospital neurological complications.
Statistical Analysis
All continuous variables are summarized as medians with interquartile ranges (IQRs). Categorical variables are reported as frequencies with percentages. Summaries of data completeness per variable are given in Supplementary Table 3.
We analysed neurological manifestations reported at hospital admission and neurological complications during hospitalization using all available data collected as prespecified fields in study CRFs (Table 1). For analyses of neurological manifestations and neurological complications, we considered unadjusted prevalence estimates for predefined patient subgroups and adjusted estimates as a function of patient age and time of hospitalization (month/year) using generalized linear models (GLMs). All GLMs assumed a binary response (yes/no) and fixed effects for age, sex, month/year of hospitalization and contributing study site as a potential confounder (Figure 1). Age and month/year of hospitalization were treated as continuous variables and modelled via polynomial terms up to an order of 3 with model selection performed using Akaike's Information Criterion. Model estimates were summarized as marginal effects and uncertainty was reported by 95% CIs.
Unadjusted odds ratios (ORs) with 95% CIs were calculated for neurological complications when we compared ICU to non-ICU cohorts in Table 2. Adjusted ORs (aORs) from multivariable analyses accounted for prespecified variables and determined the association between covariates and neurological complications.
Secondary Analysis and Missing Data
We examined associations between neurological complications and in-hospital mortality and used unadjusted analyses to investigate the cumulative incidence of death and discharge up to 100 days from hospitalization. In multivariable analysis, we used logistic regression models to examine associations with the odds of in-hospital mortality based on recorded final disposition. For multivariable analyses, missing data on independent variables were assumed missing at random and values were imputed by Multiple Imputing using Chained Equations (MICE). To account for differences in data collection across CRFs, MICE was applied independently to each study cohort. Completeness of data included in multivariable analyses of variables is reported in Supplementary Table 3. Unadjusted cumulative incidence functions were computed for patients with reported stroke, in-hospital seizures and CNS infection. Functions were further computed for a matched subset of controls, defined as patients who did not experience any neurological complications during hospitalization. Controls were matched based on study cohort, month/year of hospitalization, geographical subregion, sex and age (5-year age bands); up to 10 matched controls per patient with a reported neurological complication.
Data Availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
The data that underpin this analysis are highly detailed clinical data on individuals hospitalised with COVID-19. Due to the sensitive nature of these data and the associated privacy concerns, they are available via a governed data access mechanism following review of a data access committee. Data can be requested via the IDDO COVID-19 Data Sharing Platform (https://www.iddo.org/covid-19). The Data Access Application, Terms of Access and details of the Data Access Committee are available on the website. Briefly, the requirements for access are a request from a qualified researcher working with a legal entity who have a health and/or research remit; a scientifically valid reason for data access which adheres to appropriate ethical principles. The full terms are at https://www.iddo.org/document/covid-19-data-access-guidelines. A small subset of sites who contributed data to this analysis have not agreed to pooled data sharing as above. In the case of requiring access to these data, please contact the corresponding author in the first instance who will look to facilitate access.
Brain. 2023;146(4):1648-1661. © 2023 Oxford University Press
