The incidence of neurological complications in mechanically ventilated COVID-19 ICU patients: an observational single-center cohort study in three COVID-19 periods

BACKGROUND Neurological complications in COVID-19 patients admitted to an intensive care unit (ICU) have been previously reported. As the pandemic progressed, therapeutic strategies were tailored to new insights. This study describes the incidence, outcome, and types of reported neurological complications in invasively mechanically ventilated (IMV) COVID-19 patients in relation to three periods during the pandemic. METHODS IMV COVID-19 ICU patients from the Dutch Maastricht Intensive Care COVID (MaastrICCht) cohort were included in a single-center study (March 2020 – October 2021). Demographic, clinical, and follow-up data were collected. Electronic medical records were screened for neurological complications during hospitalization. Three distinct periods (P1, P2, P3) were defined, corresponding to periods with high hospitalization rates. ICU survivors with and without reported neurological complications were compared in an exploratory analysis.

Understanding the neurological complications in ICU patients with COVID-19 is crucial for providing appropriate care and management.Although direct neuroinvasion of the SARS-CoV-2 virus (variants) J o u r n a l P r e -p r o o f is only anecdotally described (14), it remains unclear whether neurological complications are (i) are attributable to systemic COVID-19 effects (e.g., hypoxia, inflammation, coagulopathy), and/or (ii) the adverse effects of prolonged, intensive medical treatment.Assessment of cerebrospinal fluid (CSF) and plasma samples in COVID-19 patients with neurological complications has provided evidence for cerebral hypoxia, blood-brain-barrier disruption, and overstimulation of microglia, possibly related to a cytokine storm (15).Based on novel scientific insights in COVID-19, management of critically ill COVID-19 patients however evolved over the course of the pandemic.(16,17).Especially changes in inflammatory treatment have affected the disease course of critically ill COVID-19 patients, in terms of duration of invasive mechanical ventilation (IMV), length of ICU stay, and mortality (17).It remains unclear how changes in therapeutic strategies and viral variants affect the incidence and types of neurological complications in critically ill COVID-19 patients.
The primary aim of this single-center cohort study was to describe the reported neurological complications in IMV COVID-19 ICU patients across three distinct periods during the pandemic.The secondary aim was to compare (1) clinical and treatment-related variables during the ICU stay and (2) functional outcome in ICU survivors with and without neurological complications.

Study Design
This single-center observational study was conducted in the department of Intensive Care Medicine, at the academic Maastricht University Medical Center + (MUMC+), Maastricht, the Netherlands.

Study results are reported in accordance with the Strengthening of Reporting of Observational
Studies in Epidemiology (STROBE) guidelines for observational studies.Data of patients admitted to the ICU during the COVID-19 pandemic between March 2020 and October 2021 was used.During this time frame, three periods (P) were defined in our cohort based on increased numbers (peaks) of hospital admissions: from March 11 th 2020, to May 31 st 2020 (P1), from June 1 st 2020 to March 23 rd 2021 (P2), and from March 24 th 2021 to October 1 st 2021 (P3) (Supplementary Figure 1).The characterization of P2 and P3 was accompanied by changes in therapeutic strategies (e.g., to treat inflammation).All ICU survivors were invited for a post-ICU follow-up visit at least three months after ICU discharge.

Study Population
All eligible patients were included in the prospective Maastricht Intensive Care COVID cohort (MaastrICCht) study (18).Patients eligible for this study were aged ≥18 years, critically ill and

Neurological complications
Electronic medical records were retrospectively and systematically screened (by FdV and MA) for neurological symptoms and complications during the total hospitalization period.A pragmatic inhouse scoring protocol was developed (see Supplementary Document 1), which included categories based on previously published neurological complications and symptoms in COVID-19 patients (9,20).Neurological symptoms (not classified as neurological complications) included anosmia and dysgeusia, myalgia, headache, and any expressed cognitive problems.The following neurological complications were distinguished: acute encephalopathy/delirium, stroke, seizures, persistent coma, (suspected) meningoencephalitis, cerebral sinus thrombosis, Guillain-Barré Syndrome, acute disseminated encephalomyelitis on brain imaging, (iatrogenic) peripheral nerve injury, and ICUacquired weakness.Neurological symptoms and complications were scored as present when noted in the patients' medical reports or ICU/ward discharge letters.Also, the number of neurological consultations and diagnostics (e.g., computed tomography (CT), magnetic resonance imaging (MRI), electroencephalography (EEG), and lumbar puncture (LP)) during hospitalization were also noted if these were reported in the hospital records.During the screening process, the number and types of extracranial vascular complications (e.g., pulmonary embolism and myocardial infarction) were collected as especially venous thrombosis is a common complication in COVID-19 ICU patients.For the purpose of this study, only the reported neurological, vascular and diagnostic events during the ICU stay were analyzed.

Demographic and clinical data
Demographic and clinical data were extracted from the electronic patient files and collected as part of the MaastrICCht cohort data collection.Registered demographic variables were age (years), gender, and BMI (kg/m 2 ), the Acute Physiology and Chronic Health Evaluation version II (APACHE II) score and the sequential organ failure assessment (SOFA) score, were registered.Other clinical variables of interest were duration of IMV (days), incidence and duration of prone position (% and days, respectively), length of ICU stay (days), length of hospital stay (days), and ICU mortality (%).

J o u r n a l P r e -p r o o f
The following supportive or therapeutic agents were collected: neuromuscular blocking agents (rocuronium), sedation regime (propofol, midazolam, dexmedetomidine), steroids (dexamethasone), and interleukin-6 (IL6) receptor inhibitors (tocilizumab).The frequency of use (%), duration of administration (days), and dosage (unit depending on the therapeutic agent) were reported for each agent.

Follow-up data
All ICU survivors were invited to the post-ICU outpatient clinic for a follow-up visit at least three months after hospital discharge (21,22).Health-related quality of life was assessed using the Euro-QoL-5D-5-level questionnaire and presented as a country-specific health utility score (HUS), as previously described (23).In the Netherlands, HUS ranges from −0.446 (worse than death) through 0 (death or as bad as death) to 1 (perfect health).In addition, Euro-QoL visual analogue scale (VAS)ranging from 0 (the worst imaginable healthy state) to 100 (the best imaginable health state) -was used to assess self-rated health status.

Statistical analysis
Data were analyzed using MATLAB (version R2021b; The Mathworks Inc., Natick, Massachusetts).All variables were compared across P1, P2, and P3 using an ANOVA (or a non-parametric counterpart, i.e., Kruskal-Wallis test) with post-hoc comparisons using the Tukey method.An additional exploratory analysis was performed to compare the characteristics of ICU survivors with and without neurological complications.For this purpose, patients across the three COVID-19 periods were grouped based on the event of one or more neurological complications during ICU admission.The Independent Samples T-test (or non-parametric counterpart, i.e., Mann-Whitney U test) was used to compare ICU survivors with and without neurological complications.Multivariable logistic regression models were defined to assess the relationship between patient demographics, COVID-19 periods, and specific treatment strategies between ICU survivors with and without neurological complications.Categorical variables were compared across the COVID-19 periods using a general linear model.Additional multivariable logistic regression models were used to explore if the COVID-19 period or certain treatment strategies are independently associated with the occurrence of neurological complications.A p-value < 0.050 was considered statistically significant.
J o u r n a l P r e -p r o o f 6

Reported neurological complications during ICU admission
Neurological complications were reported in 79/324 (24.4%) patients, of which 11 patients had more than one neurological complication (Table 2).The proportion of patients with reported neurological complications significantly decreased over the periods (P1 to P3: p = .032).The most commonly reported types of neurological complications during ICU admission were acute encephalopathy/ J o u r n a l P r e -p r o o f delirium (60.8%) and ICU-acquired weakness (40.5%) (Table 2).Neurological symptoms were mostly reported during hospital admission, with only a few notable neurological symptoms during the ICU admission period (Supplementary Table 1).The number of neurological consultations and diagnostic assessments during ICU admission are described in Supplementary Table 2 and 3, respectively.

COVID-19 ICU survivors with and without neurological complications during ICU admission
Out of 197 COVID-19 ICU survivors, 61 had experienced one or more neurological complication(s) (31.0%).For 86.9% of these patients this involved encephalopathy/delirium or ICU-acquired weakness.Eight (13.1%) survivors with any other neurological complication (predominantly stroke) were excluded to prevent bias due to specific treatment strategies for these complications.(Figure 1).Comparing demographic and clinical variables revealed that ICU survivors with a reported neurological complication were older (p = .003),had a longer duration of IMV (p < .001),and a longer ICU stay (p < .001)(Table 3).However, only age was independently associated with the occurrence of neurological complications in a multivariable analysis (OR adj = 1.054; 95% CI = 1.017 -1.093; p = .004)(Supplementary Table 4).For the neurological complications group, the median duration of IMV was longer than the median length of ICU stay in our hospital because of the higher number of patients who were intubated before referral to our hospital.ICU survivors with neurological complications were treated for a significantly longer period of time with rocuronium (p = .032),propofol (p = .002),midazolam (p = .022),dexmedetomidine (p = .006),and dexamethasone (p = .006)(Table 3).Only the use of dexmedetomidine was significantly higher in ICU survivors with a neurological complication (p < .001)(Table 3).Correcting for COVID-19 period revealed that only age was independently associated with the occurrence of neurological complications (OR adj = 1.053; 95% CI = 1.015 -1.091; p = .005)(Supplementary Table 5).Similarly, correcting for treatment strategies revealed that only age was independently associated with the occurrence of neurological complications (OR adj = 1.053; 95% CI = 1.016 -1.092; p = .005)(Supplementary Table 6).

Follow-up assessment
Follow-up data was available for 47.2% of survivors with and for 43.2% of those without a neurological complication (Figure 1; Table 4).The time interval between the post-ICU follow-up visit and ICU and hospital discharge was not significantly different between the groups (p = .542and p = .562,respectively) (Table 4).No significant difference was observed for the reported EQ-5D HUS score between patients with and without neurological complications (p = .054)(Table 4).Moreover, a multivariable analysis revealed neither age, nor gender, nor length of IMV and ICU stay, nor the occurrence of neurological complications were independently associated with the EQ-5D HUS score J o u r n a l P r e -p r o o f (Supplementary Table 7).Evaluation of individual domains of the EQ-5D or EQ-5D VAS score also revealed no significant differences between patients with and without neurological complications, except for the Anxiety and Depression domain (p = 0.046) (Table 4).

DISCUSSION
The overall reported incidence of neurological complications in this cohort of 324 IMV COVID-19 ICU patients admitted to a tertiary care hospital in the Netherlands was 24.4%.This retrospective study has several interesting findings.First, ICU mortality was high in this cohort and did not significantly change across the three periods.Second, the incidence of reported neurological complications during ICU admission decreased significantly over time.The most frequently reported neurological complications were encephalopathy/delirium (14.8%) and ICU-acquired weakness (9.9%).Third, ICU survivors with reported neurological complications were significantly older, more often male, required a longer duration of IMV, and had a longer stay in the ICU.This was accompanied by longer administration of sedatives and neuromuscular blocking agents.Reported health-related quality-oflife at follow-up was not significantly different between survivors with or without neurological complications.

Comparison of defined COVID-19 periods
The proportion of patients with reported neurological complications significantly decreased across the defined COVID-19 periods.A higher admission SOFA score in P2 and P3 compared to P1 suggests that patients had more (multi) organ failure at ICU admission later in the pandemic, although this was not associated with an increase in reported neurological complications or mortality.Although the length of ICU stay and ICU mortality did not significantly decrease over the three periods, the duration of IMV and length of hospital stay did significantly decrease over time.This could suggest that patients were more effectively treated for their pneumonia and/or associated inflammatory response.Indeed, the decrease in duration of IMV over time coincided with a decrease in the occurrence of neurological complications, particularly for encephalopathy/delirium.However, this relation could be confounded by the increase of certain treatment strategies over time.More patients were treated with dexamethasone and tocilizumab in P3 than in P1, which have shown to be effective in reducing the length of hospital stay in large clinical studies (24,25).Together, the most striking differences were observed between P1 and the other two periods.These differences may be (partially) caused by different virus strains.Where P1 was primarily dominated by the Alpha variant (i.e., B.1.1.7),the Delta variant of the virus (i.e., B.1.617.2) was predominant in P2 and P3 in the Netherlands (26).However, other factors, such as vaccination status or evolving ICU admission J o u r n a l P r e -p r o o f criteria, may also have contributed to the observed differences between the periods which complicate the interpretation of the results.

Neurological complications during ICU admission in perspective
The reported incidence of neurological complications in another cohort of IMV COVID-19 ICU patients was 50.0%, with acute encephalopathy/delirium being the most frequently reported complication (38.3%) (20).Similar numbers have been reported in a COVID-19 ICU cohort with half of the patients requiring IMV: a total incidence of neurological complications of 44.2%, with delirium being the most frequent neurological complication (38.5%) (27).These COVID-19 ICU cohort studies adopted a similar design (i.e., retrospective screening of electronic medical records), but the patient admission period of these two studies did not extend beyond P1 as defined in our cohort.A COVID-19 ICU cohort study covering a similar time frame as our study -although without a distinction between different admission periods -reported a lower incidence of neurological complications of 14.3%, again with encephalopathy/delirium being the most frequently reported neurological complication (6.6%) (4).Notably, the incidence of neurological complications in reports restricted to the first COVID-19 period is higher than in reports also including COVID-19 ICU admissions after June 2020, suggesting a general decrease in the incidence of neurological complications (4,20,27), which is in line with our results.However, the overall incidence of reported neurological complications in our cohort was lower compared some other cohort studies (20,(27)(28)(29), which could be due to underreporting of neurological complications in our retrospective medical file study.Nevertheless, a prospective study monitoring neurological complications in COVID-19 ICU patients reported an incidence of only 12.7% (4).

COVID-19 ICU survivors with and without neurological complications during ICU admission
To avoid a mortality bias, ICU survivors with and without neurological complications were compared to study demographic and therapeutic differences between these two patient groups.ICU survivors who experienced acute encephalopathy/delirium and/or ICU-acquired weakness were older and required longer IMV and admission to the ICU than those without any reported neurological complications.Older age has previously been identified as a risk factor for delirium in COVID-19 ICU patients (30).Longer duration of IMV and longer length of ICU stay have previously been associated with the development of encephalopathy/delirium (29,30) and ICU-acquired weakness (31,32).
Patients with neurological complications also received prolonged treatment with a combination of neuromuscular blockers, sedatives, and steroids.These agents have been associated with neurological complications in other COVID-19 ICU cohorts (29,31,32).However, multivariable J o u r n a l P r e -p r o o f analyses revealed that only age was independently associated with the occurrence of neurological complications.

Specificity of neurological complications in COVID-19 ICU patients
Although it is tempting to attribute reported neurological complications to the SARS-CoV-2 virus itself, the category and the incidence of the reported complications resembles neurological complications observed in critically ill IMV patients with a diversity of admission diagnosis.Texeira-Vaz et al. (33) compared neurological complications in IMV ICU patients with COVID-19 acute respiratory distress syndrome (ARDS) with non-COVID-19 ARDS patients and reported encephalopathy/delirium as the most frequent complication in both groups (33% in COVID-19 versus 26% in non-COVID-19 ARDS patients).Neurological complications have also been reported in ICU patients with influenza, albeit with a lower incidence of 4-6% in adult patients (34,35).ICU-acquired weakness has been frequently reported in sepsis patients (incidence 25-31%) (36) and is associated with prolonged IMV (37).Encephalopathy/delirium has been reported with an incidence of 48-67% in sepsis patients (38)(39)(40) and in 84% of critically ill patients requiring IMV (41).Westphal et al. (42) compared the incidence of delirium in critically ill patients with and without COVID-19, showing a higher incidence of encephalopathy/delirium in COVID-19 ICU patients.However, when only IMV patients (duration ≥ two days) were included in the comparison, the incidence of encephalopathy/delirium did not significantly differ between the two groups.The prevalence of neurological complications in different ICU populations may suggest treatment effects or similar pathophysiological mechanisms underlying neurological complications.For instance, Etter et al. (15) reported evidence of blood-brain-barrier disruption and microglia activations in COVID-19 patients with neurological complications, reflecting a cytokine storm.Similarly, the cytokine storm has been associated with in patients with influenza, ARDS, and sepsis (43).

Long-term implications of neurological complications in COVID-19 ICU survivors
No significant difference in health-related quality-of-life at the time of the post-ICU follow-up visit was observed in the group of survivors with neurological complications compared to those without, except for the Anxiety and Depression domain.This suggests that the most common neurological complications might not be essential for quality-of-life after hospitalization for COVID-19.It is important to note that neurological disabilities were not specifically addressed during the clinical follow-up.Due to prolonged ICU stay and the high incidence of IMV, COVID-19 ICU patients (and non-COVID-19 patients alike) are vulnerable to the development of long-term symptoms, both physically and mentally.Long-term symptoms are commonly reported in COVID-19 ICU survivors and J o u r n a l P r e -p r o o f in that respect resemble the 'post-intensive care syndrome' (PICS) as described in the general ICU population (44)(45)(46)(47)(48)(49).

Study limitations
The presented study expands on previous knowledge as the incidence of neurological complications is described in an IMV COVID-19 ICU cohort comparing three periods during the COVID-19 pandemic.
However, this study has several limitations.Due to the reliance on the documentation of neurological complications during ICU stay, some neurological complications may have been missed because of information bias.There was no standard neurological consultation and examination in our IMV patients and brain imaging, and other diagnostics were only performed on clinical indication.In addition, there was limited use of standard diagnostic tools -such as the Confusion Assessment Method Intensive Care Unit (CAM-ICU) -in our ICU during the pandemic and the severity and duration of the symptoms were not taken into account.Moreover, the high mortality rates contributed to the likely underreporting of neurological complications.However, we consider the presence of these limitations to be comparable in all periods.Therefore, the possibly introduced bias would not have affected the comparison between COVID-19 periods, although it may explain why the incidence of neurological complications is lower than in some other cohorts (20,(27)(28)(29).
Other factors leading to underreporting of neurological complications include interhospital patient transport.For this reason, this study focused on neurological events during the ICU stay (as data during ward and rehabilitation admission were less complete).To address potential underreporting, there is a need to further educate, train and implement (new) delirium assessment tools (e.g., CAM-ICU, Intensive Care Delirium Screening Checklist (ICDSC)) at the bedside to enhance early recognition of delirium (50).Another way to reduce the risk of underreporting is through investment and implementation of novel advancements in early automated detection of delirium, especially with EEG in comatose patients or in patients waking up (see for example Williams Roberson et al. (51)).
Moreover, the duration of the three defined periods was disproportionate: 81 days (P1) versus 295 days (P2) versus 191 days (P3).Although the three periods were chosen based on ICU admission peaks during the pandemic, their disproportionate durations and number of patients may have affected our results.Furthermore, this study describes the characteristics and outcomes of patients of a single hospital and only of COVID-19 patients who required intubation and ventilation.
Restricting our cohort to only ICU and IMV patients improves the homogeneity in the studied sample.Although this study design also comes with restrictions in terms of generalizability, the main observations of this study align with and expand previous results in COVID-19 and general ICU J o u r n a l P r e -p r o o f populations, emphasizing the importance of monitoring neurological complications in critically ill patients.Moreover, although high rates of COVID-19 readmission have been reported (52), the study design (18), which was focused on serial data analysis, enabled to include all data during any ICU admission or readmission within each patient.In that way no patients were registered twice in the database if they developed neurological complications at a later date leading to readmission to the  57)).Lastly, complete follow-up data was available for a limited number of survivors (42.6%); hence, this should be considered when interpreting the follow-up results.High rates of data loss to follow-up are common in the ICU (58), emphasizing the need for investments in standardized followup practice of critically ill patients.

CONCLUSION
This single-center observational study describes the incidence of neurological complications in a cohort of IMV COVID-19 ICU patients in three periods in the pandemic.A high incidence of neurological complications during ICU admission was observed, which decreased as the pandemic progressed.The two most common neurological complications were encephalopathy/delirium and ICU-acquired weakness, and were associated with prolonged and intense ICU treatment in those who survived.However, having a neurological complication during the ICU stay did not lead to significantly worse reported health-related quality-of-life at follow-up in a subgroup of the cohort.The three COVID-19 periods were compared using a Kruskal-Wallis test (or general linear model for categorical variables).J o u r n a l P r e -p r o o f

LIST OF ABBREVIATIONS
J o u r n a l P r e -p r o o f required IMV, admitted to the ICU for severe hypoxia, and diagnosed with a SARS-CoV-2 infection confirmed by a positive PCR and/or a CT scan of the chest with a COVID-19 Reporting and Data System (CO-RADS) score of 4-5, indicative of severe COVID-19 pneumonia (19).Patients were admitted via the emergency department, hospital wards or referred from other ICUs, either for tertiary care or due to lack of bed availability in the regional ICUs.

Table 1 . Sample characteristics for each COVID-19 period.
Demographic, ICU admission, and treatment parameters are shown for the total sample (N = 324) and the three defined COVID-19 periods.