Prolonged Intubation in Patients With Prior Cerebrovascular Disease and COVID-19

Objectives: Patients with comorbidities are at increased risk for poor outcomes in COVID-19, yet data on patients with prior neurological disease remains limited. Our objective was to determine the odds of critical illness and duration of mechanical ventilation in patients with prior cerebrovascular disease and COVID-19. Methods: A observational study of 1,128 consecutive adult patients admitted to an academic center in Boston, Massachusetts, and diagnosed with laboratory-confirmed COVID-19. We tested the association between prior cerebrovascular disease and critical illness, defined as mechanical ventilation (MV) or death by day 28, using logistic regression with inverse probability weighting of the propensity score. Among intubated patients, we estimated the cumulative incidence of successful extubation without death over 45 days using competing risk analysis. Results: Of the 1,128 adults with COVID-19, 350 (36%) were critically ill by day 28. The median age of patients was 59 years (SD: 18 years) and 640 (57%) were men. As of June 2nd, 2020, 127 (11%) patients had died. A total of 177 patients (16%) had a prior cerebrovascular disease. Prior cerebrovascular disease was significantly associated with critical illness (OR = 1.54, 95% CI = 1.14–2.07), lower rate of successful extubation (cause-specific HR = 0.57, 95% CI = 0.33–0.98), and increased duration of intubation (restricted mean time difference = 4.02 days, 95% CI = 0.34–10.92) compared to patients without cerebrovascular disease. Interpretation: Prior cerebrovascular disease adversely affects COVID-19 outcomes in hospitalized patients. Further study is required to determine if this subpopulation requires closer monitoring for disease progression during COVID-19.


INTRODUCTION
Disease outcomes associated with the novel coronavirus 2019  are heterogeneous and include asymptomatic disease, mild respiratory tract illness, severe pneumonia with respiratory failure and acute respiratory distress syndrome, and death (1). It is estimated that one in four patients infected with SARS-CoV-2 requires supplemental oxygen or invasive mechanical ventilation (1)(2)(3)(4)(5). To date, survival has been correlated with multiple factors including age, medical comorbidities, and host response to the virus that may lead to multiorgan dysfunction, coagulopathy, and elevated inflammatory markers (1,(4)(5)(6)(7).
The prevalence of hospitalized patients with neurological comorbidities COVID-19 widely varies between 1 and 12% depending on the cohort and comorbidities studied (8)(9)(10)(11)(12)(13)(14)(15). Our understanding of the risk of COVID-19 critical illness due to chronic neurological conditions remains limited, with cohorts from Asia and Europe suggesting a history of ischemic or hemorrhagic stroke as risk factors for severe events such as mechanical ventilation (MV) (8) and death (8,10). Recently, two meta-analyses reviewed the relationship between the prior cerebrovascular disease and in-hospital outcomes in COVID-19, using data derived from Chinese and European cohorts, and suggested an increase in risk for critical illness (16) and mortality (17) among patients with prior cerebrovascular disease.
Studies prior to the pandemic suggest that patients with neurological comorbidities are at increased risk for critical illness compared to similarly matched older patients (18,19). Cerebrovascular comorbidities are common among older adults in the United States (U.S.) where an estimated 3% of adults have had a prior ischemic stroke (7.8 million) (20). A recent study from a U.S. cohort of 3,248 patients suggested an increased odds of in-hospital death among individuals with stroke and COVID-19, however, detailed data on other inhospital adverse outcomes is largely unknown (21). Given that prior cerebrovascular disease is one of the most common neurological comorbidities in hospitalized patients with COVID-19, information regarding severe outcomes in this population would be valuable for prioritizing prevention strategies in the outpatient neurology setting, providing prognostic information for patients and families, and assisting hospital projections as countries experience increasing numbers of SARS-CoV-2 infections.
In this study, we examined the relationship between prior cerebrovascular diseases and critical illness in the first 28 days of admission and determined the likelihood of successful extubation over a 45-day in-hospital follow-up among adults with positive SARS-CoV-2 RNA admitted to an academic hospital in Boston, Massachusetts, during the first 2 months of the city's outbreak. We hypothesized that prior cerebrovascular disease was a risk factor for critical illness in COVID-19 and a comorbidity associated with increased duration of mechanical ventilation (MV).

Study Setting and Population
This is an observational study of 1,128 consecutive patients with laboratory-confirmed SARS-CoV-2 infection hospitalized at Massachusetts General Hospital (MGH), a single-center tertiary care facility in Boston, MA. Laboratory confirmation of SARS-CoV-2 infection was obtained using real-time reverse transcription-polymerase chain reaction assays with Food and Drug Administration emergency use authorization. We identified 1,216 patients seen in the Emergency Department (ED) or hospitalized between March 1st and May 5th, 2020, given that the first reported case of SARS-CoV-2 infection in Boston was March 2nd, 2020 (Figure 1). Patients < 18 years old and patients classified as being seen in the ED using electronic data collection but were seen in outpatient clinics after manual review of cases were excluded from analyses. The institutional review board approved this study (Protocol #: 2013P001024) with a waiver of consent for retrospective analyses.

Data Collection
Engineers from the MGH Clinical Data AI Center extracted COVID-19-related data from the Partners Healthcare Systems Enterprise Data Warehouse, which comprises electronic medical record data from the Mass General Brigham network (formerly Partners Healthcare). Data queried for this study included demographics, admission, discharge, intubation and extubation events, diagnosis and International Classification of Diseases, Tenth Revision, Clinical Modification codes (ICD-10-CM), vital signs, laboratory values within 48 h of admission, and other data. Diagnostic categories were created by grouping ICD-10-CM diagnosis codes using a combination of groupings by SNOMED CT, an ontology-based terminology owned and maintained by the standards development organization SNOMED International and licensed through the National Library of Medicine. Four clinicians performed a manual chart review of identified cases of prior cerebrovascular disease up to 7 days before COVID-19 hospital admission (HA, SM, LB, and AN); a new diagnosis of cerebrovascular disease was recorded if the event occurred within 7 days prior to or after a positive SARS-CoV-2 PCR result.
Of the patients with COVID-19 included in this study, 289 underwent diagnostic neuroimaging after admission (computed tomography or magnetic resonance imaging of the brain), with 484 unique studies performed during the study period. Two radiologists (MDL and MLa) identified evidence of prior intracerebral pathology by reviewing neuroradiographic reports and categorizing intracranial pathology into seven clinical findings: acute or subacute ischemic infarct, chronic ischemic infarct, acute or subacute intracranial hemorrhage, chronic intracranial hemorrhage, post-surgical change (limited to any intracranial surgery), intracranial mass (metastases or primary malignancies), and traumatic brain injury. These measures were used to provide ancillary information on cerebrovascular disease history in addition to ICD-10-CM codes; all positive imaging data not identified using ICD-10-CM codes were manually reviewed to confirm the clinical suspicion of diagnosis (HA and SM), and added 14/289 additional cases.
The date of intubation was obtained using ventilator flowsheets, marking positive end-expiratory pressure (PEEP), and FiO2. Date of extubation was flagged if there were no PEEP and FiO2 readings after 48 h of continuous intubation markings on ventilator flowsheets. Two critical care physicians (HA and LB) manually confirmed intubation in all cases and extubation status and date in 83% of cases (n = 243/294). Death and death date was extracted from the EHR. All patients with a date of death were manually reviewed by study team members (AN, LB, SC, and KK) to determine if they transitioned to comfort-measuresonly (CMO) and if so, date of transition was recorded.

Exposure and Outcome Measures
The binary variable for cerebrovascular history included at least one diagnosis of ischemic stroke, intracerebral hemorrhage (ICH), venous sinus thrombosis (VST), subarachnoid hemorrhage (SAH), or subdural hemorrhage (SDH) recorded at least 7 days prior to admission. A composite outcome, critical illness (not to be confused with critical illness myopathy), was defined as invasive MV or death within 28 days of admission (4), and used for the primary analysis. A secondary outcome was the probability of successful extubation, defined as liberation from mechanical ventilation irrespective of the mode of ventilation delivery (i.e., endotracheal or tracheostomy tube). Follow-up time was right-censored on June 20th, 2020, to allow 45 days of observation for intubated patients given long-durations of intubation in COVID-19 patients. All positive prior neurological diagnoses and outcomes were confirmed manually by chart review (HB, SM, AN, LB, SC, and KK).

Statistical Analysis
Continuous and categorical variables were presented as median [interquartile range, (IQR)] and n (%), respectively. Mann-Whitney U tests, χ² tests, or Fisher's exact tests were used to compare differences between critical and non-critical patients where appropriate.
The control for confounding was done using inverse probability weighting (IPW) method. The propensity scores for cerebrovascular disease were estimated with a multivariable logistic regression model that included age, sex, Latinx ethnicity, and Black or African American race. The predicted probabilities from this propensity-score model were used to calculate the stabilized IPW scores (22). Logistic regression models using IPW are reported, and models estimated odds ratios (OR) [95% confidence intervals (CI)] for the association between history of cerebrovascular disease and critical illness by day 28. In sensitivity analyses, a matching strategy was used to match patients with and without cerebrovascular disease in a 1:2 ratio by age, sex, Latinx, and Black or African American race. The smallest average absolute distance was used to match across all pairs (R 4.0.0 MatchIt package).
We examined the effect of cerebrovascular disease history on the duration of intubation, with death as a competing event. Time was measured in days from intubation to successful extubation or death (two mutually exclusive events) and censored at 45 days if no event was observed. If a patient was extubated and died during the observation period, the event recorded was death. We estimated the cause-specific hazard of transitioning from intubation to successful extubation and calculated the hazard ratio (HR) for patients with cerebrovascular history compared to those without using a propensity score weighted Cox proportional hazards model. The cumulative incidence curves (CICs) for extubation and death were computed using the R package causalCmprsk with both Cox PH and Aalen-Johansen's non-parametric estimators (23,24). Propensity score weights were used for generating the CIC curves to account for confounding by age and sex. We estimated exposure effect as the restricted mean time difference, which is the area under the CIC curve and provides a more clinically meaningful measure than HR (25).

Odds of Critical Illness in Patients With Prior Cerebrovascular Disease
A history of ischemic stroke (n = 112; 9.9%), ICH (n = 22; 2.0%), or SDH (n = 60; 5.3%) was frequent among COVID-19 patients with a total of 177 (16%) patients having at least one prior cerebrovascular disease diagnosis. Patients with a prior cerebrovascular disease were more likely to be critically ill compared to non-critically ill individuals [ (Figures 2B,C), and across all ages, in-hospital 28-day mortality was higher between patients with vs. without cerebrovascular disease [

Cumulative Incidence of Successful Extubation in Patients With Cerebrovascular Disease
To further understand the relationship between prior cerebrovascular disease and COVID19 severe outcomes, we used a competing risk analysis framework to determine the relationship between cerebrovascular disease and duration of intubation without subsequent death. The cumulative incidence of successful extubation without death in patients with cerebrovascular disease was lower compared to those without cerebrovascular history (Figure 3), and there was a significant association between prior cerebrovascular disease and likelihood of successful extubation (adjusted cause-specific HR, 0.57; 95% CI, 0.33-0.98). Over a 45-day observation window, patients with cerebrovascular disease had a longer intubation time with a restricted mean time difference of  4.02 days [0. 34, 9.32] added time on mechanical ventilation compared to patients without cerebrovascular disease using a Cox-PH model. The time difference was modeled using non-parametric cumulative incidence functions in sensitivity analyses and showed an additional 5.65 days [2.40, 10.85] of intubation time in patients with cerebrovascular disease. There were no differences in cause-specific HR for death between patients with and without cerebrovascular history after adjustments.

DISCUSSION
In this observational study, we present one of the largest analyses with extended follow-up among US hospitalized patients with prior cerebrovascular disease and COVID-19. Of the 1,128 hospitalized patients, 16% (177/1,128) had at least one diagnosis of cerebrovascular disease prior to COVID-19, and odds of critical illness in this subpopulation were 1.5-times higher compared to those without cerebrovascular disease. Additionally, these data show that patients with cerebrovascular disease were less likely to achieve successful extubation and estimated to be ventilated for 4-5 days longer than patients without a prior cerebrovascular disease. Given that available data on cerebrovascular disease comorbidity and in-hospital outcomes are limited, we anticipate these findings to be relevant for outpatient prevention strategies and prognostic discussions with patients and families, especially as countries experience resurges of SARS-CoV-2 infections. In this cohort, the majority of patients with a prior cerebrovascular disease had a prior acute ischemic stroke (9.9%; 11/1,128), a prevalence which was higher than expected from U.S. 2013-2016 stroke estimates of 2.5% for adults ages > 20 years old (14). These results are consistent with published cohorts suggesting a greater number of hospitalized COVID-19 patients with cerebrovascular disease than prevalence estimates and increased likelihood of negative outcomes (4, 8, 10-13, 16, 17, 21, 28-31). The findings presented here extends our understanding of comorbidities that may contribute to increased risk of critical illness in COVID-19, and additionally suggests that critically ill COVID-19 patients with cerebrovascular disease may be prone to longer mechanical ventilation time than patients of similar ages and demographics without cerebrovascular disease. It is possible that patients with prior history of cerebrovascular disease are vulnerable to prolonged ventilation in COVID-19 given a propensity for lower levels of premorbid function, increased probability of cerebrovascular events after diagnosis or increased risk of frailty, a clinical state which is a strong predictor of adverse health effects including hospitalization, disability, and mortality (32). Recent data also suggests that COVID-19 patients with delirium are more likely to have longer duration of intubation (33). Given that predictors of delirium include medical comorbidities such as cerebrovascular disease, it may be challenging to distinguish which factors predominate in conferring risk of longer intubation times and requires large scale studies. Irrespective of cause, the impact of longer duration of intubation can be extrapolated from other critical care studies prior to the COVID-19 pandemic, which show that increased ventilation times are associated with a greater need for sedation and analgesics and higher rates of ventilator-associated pneumonia, line infections, urinary tract infections, delirium, ileus, and decubitus ulcers (34)(35)(36). Further, additional days on ventilators are likely to lead to greater rehabilitation needs, reduced cognitive function, and could be uniquely detrimental for patients with prior neurological deficits and their family members (37,38). A prior study suggested that patients with cerebrovascular disease may have more aggressive inflammatory responses on admission for COVID-19 (8). While our data did not show increases in CRP or ferritin or evidence of lymphopenia among patients with a prior cerebrovascular disease, they had higher levels of D-dimer and troponin levels, consistent with a prior study of stroke patients (8). Additionally, admission RDW was elevated in patients with cerebrovascular disease in this study. Elevated RDW has been shown to be a marker of allcause mortality, a predictor of complicated hospitalizations that included the need for MV from infectious causes such as influenza (26,39). and recent data suggests an association with increased mortality risk in COVID-19 (40). Given that COVID-19 is associated with diffuse coagulopathy and thrombotic events (41)(42)(43). further study is required to determine if COVID-19 infection exacerbates vascular pathology present in patients with cerebrovascular disease and if this subpopulation requires closer diagnostic monitoring for coagulopathy and disease progression during COVID-19.
Our work has several limitations worth noting. It is a singlecenter observational study and relies on the EHR, which may not capture full medical histories; thus, some misclassification of prior medical diagnosis is possible. To minimize misclassification bias, our group manually validated data relying on expertise from clinicians in multiple disciplines. We allowed for ancillary data regarding history of cerebrovascular disease based on radiographic imaging data to be introduced, and while we added only a small fraction of patients to the overall cohort with history of cerebrovascular disease (14 patients), this may have led to residual confounding. Data on premorbid level of functioning and details of prior lung function were unavailable and could impact the likelihood of successful extubation; neither could be adjusted for in our analyses. Long-term follow-up that includes cognitive assessments for patients with prior cerebrovascular disease will be critical to understand the longitudinal impact of COVID-19 in this subgroup. Finally, we used SARS-CoV-2 RT-PCR positive results as an indicator of COVID-19 disease, however, PCR results may be an incidental finding in some cases. At the time of admission for this cohort, asymptomatic patients were not routinely tested using SARS-COV-2 RT-PCR, and the majority of cases presented were clinically considered as having COVID-19.
In summary, our findings show that patients with cerebrovascular disease and COVID-19 have higher odds of critical illness, and a lower incidence of successful extubations. This subpopulation is estimated to have longer mechanical ventilation times compared to patients of similar ages without cerebrovascular disease. In aggregate, these data suggest there are important opportunities for proactive outpatient neurological care and open discussion regarding vaccine allocation priorities and for the management and expectations of duration of mechanical ventilation and critical disease in patients with cerebrovascular disease.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Mass General Brigham (MGB) institutional review board (IRB). Written informed consent for participation was not required for this study in accordance with the IRB for retrospective analyses.