Delirium and encephalopathy in severe COVID-19: a cohort analysis of ICU patients

Patients infected with severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), also known as coronavirus disease 2019 [COVID-19], mainly develop respiratory and digestive symptoms [1, 2]. However, due to similarity of viral structure and infection pathways [3], it has been suggested early in the epidemics that SARS-CoV-2 may also invade the central nervous system and be responsible for neurological signs [4], like other coronaviruses do [5‐7]. Most β-coronaviruses, including SARS-CoV, have a neuroinvasive propensity [8]. Indeed, the SARS-CoV invades cells by using a cellular receptor angiotensin-converting enzyme 2 (ACE2), which may be expressed in ciliated upper respiratory cells and type II pneumocytes, responsible for the respiratory manifestations, but also in the central nervous system (neurons and glial cells), and endothelial cells [7, 9, 10]. Several data in humans and animals suggest that coronaviruses may have neurotropic effects and mainly affect brainstem and medullary cardiorespiratory center [10‐12]. Coexisting brain lesions were also described in piglets infected with a transmissible gastroenteritis coronavirus, responsible for non-supportive encephalitis [11]. In a recent series of autopsies [13], von Weyhern et al. also described a pronounced central nervous system involvement with lymphocytic panencephalitis, meningitis, diffuse petechial hemorrhages, and brainstem neuronal cell damage. Furthermore, presence of SARS-CoV particles was finally demonstrated in the brain of patients with SARS [14].

In a descriptive study, Chen et al. reported the epidemiological and clinical characteristics of 99 cases of 2019 COVID-19 pneumonia in Wuhan, with headache in 8 patients (8%) and confusion in 9 patients (9%). Mao et al. [15] also retrospectively described neurological manifestations of SARS-CoV-2 infection in 78 of 214 (36.4%) patients with confirmed diagnosis of COVID-19 and hospitalized in Wuhan, China. The authors showed that 53 (24.8%) of the patients suffered from central nervous system symptoms, including dizziness, headache, and impaired consciousness. The authors also reported that neurologic symptoms were more common in patients with severe infection (45.5 vs. 30.0%, p = 0.02), with more cerebrovascular diseases and impaired consciousness. However, no analysis of cerebrospinal fluid, no brain magnetic resonance imaging, and only a few brain CT were performed. We have recently shown that most patients (84%) admitted to intensive care units for acute respiratory distress syndrome (ARDS) due to COVID-19 may develop neurological features, mainly delirious manifestations [16].

ICU delirium includes fluctuating disturbances in attention and cognition developing in a short period that are not explained by pre-existing neurocognitive disorder [17]. In survivors of critical illness, delirium has been shown to be associated with worse outcomes in critically ill patients with longer hospital stays, increased risk of long-term neurocognitive sequelae and neuropsychiatric disorders, and death [18].

In COVID-19, both the neurotropism of SARS-CoV-2 and its neurological manifestations have now been confirmed [4], although the mechanisms involved in these alterations are still debated. Based on the comprehensive clinical examination, analysis of cerebrospinal fluid (CSF), electroencephalography (EEG), and brain magnetic resonance imaging (MRI) of a homogeneous prospective cohort of patients admitted in ICU for ARDS due to SARS-CoV-2 infection, we aimed at describing the incidence of delirium AND/OR abnormal neurological examination and compare the outcome of these patients to patients without delirium or any neurological symptoms.

Herein, we describe the high frequency of delirium and/or neurological symptoms (84.3%) in patients admitted to ICU for ARDS due to COVID-19 and their worse prognosis compared to patients without delirium and with normal neurological examination. Indeed, delirium and/or neurological symptoms led to sustained invasive mechanical ventilation and unusually high doses of sedations and neuroleptics, whereas patients without delirium and with normal neurological examination could be extubated and discharged earlier from ICU. The delirium also increased the risk of life-threatening accidental extubation and was associated to worsening of hypoxemia after reintubation, thus maybe contributing to increase the duration of the ventilator weaning process. Mortality rate difference between groups did not reach statistical significance (16.1versus 9.1%), although it may be due to the small number of patients in our cohort.

Because COVID-19 mostly affects men [25‐27], possibly because of a large number of ACE2-expressing cells in their lung (report not peer-reviewed) [28], most ICU patients with COVID-19 delirium were men (Table 1).

Whether the acquired acute global disturbance in cognition we describe should be referred to only as delirium or even as encephalopathy may be questioned. Delirium is mainly characterized by acute attention, awareness, and cognition disturbances, while acute encephalopathy, that may not be diagnosed at bedside, is used “to describe a rapidly developing (in less than 4 weeks) pathobiological brain process which is expressed clinically as either subsyndromal delirium, delirium or coma and may have additional features, such as seizures or extrapyramidal signs” [17]. Beyond the neurological signs we have described, the clinical diagnosis of COVID-19 encephalopathy is strengthened by paraclinical features that allowed to delineate a pathobiological process: (i) MRI findings revealing perfusion abnormalities, (ii) electroencephalographic abnormalities, (iii) hints for inflammatory process in CSF, and (iv) the lack of other identified cause of delirium beyond positive SARS-CoV2 RT-PCR in all patients. Some of our patients have therefore developed an acute COVID-19 encephalopathy. We can however not affirm that all the patients with neurological symptoms fulfilled the diagnostic criteria for encephalopathy in our cohort, considering that only part of them have been submitted to paraclinical exams.

Perfusion abnormalities have been previously described in ICU and non-ICU patients with COVID-19, describing hypoperfused areas, predominantly in the temporal lobes and to a lesser extent in the frontal lobes [16, 29].

Delirium (positive CAM-ICU) or acute encephalopathy, as defined above, may relate to systemic inflammatory reaction to SARS-CoV-2 rather than to SARS-CoV-2 itself. Knowing that we have not analyzed the CSF of all patients in our cohort, the hypothesis of an inflammatory component in the emergence of the COVID-19 delirium is supported by (i) the absence of RNA viral load in CSF at the time of the diagnosis of COVID-19 except in one patient; (ii) CSF analysis revealing mirror pattern of oligoclonal bands, elevated protein and IgG level, and elevated proinflammatory cytokine IL-6; and (iii) brain imaging showing subarachnoid contrast enhancement suggestive of abnormal permeability of the blood meningeal barrier associated with the encephalopathy. One may speculate that a systemic immune event could be responsible for such abnormal permeability in the same way that the cytokine storm is involved in the occurrence of the SARS-CoV-2 infection [1].

The high frequency and reproducibility of neurological signs in our patients reinforces the hypothesis that COVID-19 may be responsible at least for delirium and in some cases for encephalopathy. Furthermore, other causes of delirium/encephalopathy were excluded (for instance iatrogenic, alcoholic, or metabolic). There was no stroke except in 3 patients that could not explain the clinical signs. In the same way, the patients with history of epilepsy had not ictal abnormalities on electroencephalography. Iatrogenic cause was also excluded, because delirium and/or corticospinal tract signs were present in 22 (15.7%) patients on ICU admission, before administration of any treatment, and because such clinical findings are very unusual in ICU patients.

Based on CAM-ICU score, Salluh et al. [30] reported a prevalence of delirium of 32.3% in a multicenter study including 497 patients, while Khan et al. [31] more recently showed that delirium occurred in 16.5% out of the 2742 ICU patient included. Prevalence of ICU delirium is however highly variable among studies, depending of the definition and screening tools used, but also on the population studied. It was therefore suggested to reach up to 73% of the 48 patients with ARDS [32]. We have reported a very high prevalence (79.5%) of delirium in COVID-19 patients. Yet, the incidence of delirium in COVID-19 patients is still probably largely underestimated [33]. Indeed, most patients were intubated before ICU admission and already under neuromuscular blockers and delirium was diagnosed following an attempt of weaning of sedation when patients were recovering from respiratory failure.

Our results raised the hypothesis of an inflammatory component in the emergence of the COVID-19 delirium and/or neurological symptoms. Consistent with Mao et al. [15], our cohort also supports the hypothesis that severe patients were more likely to develop neurological symptoms. Li et al. [12] further suggested that the neuroinvasive potential of SARS-CoV2 may play a role in the respiratory failure of COVID-19 patients, which may explain the high prevalence of delirium in our patients who have been admitted on ICU for ARDS. Furthermore, in our cohort, there were hints for SARS-CoV2 encephalitis in 8 patients, with elevated CSF protein level, brain lesion on MRI and because CSF SARS-CoV2 RT-PCR was positive in one patient investigated.

Herein, laboratory investigations, electrophysiology, and especially brain imaging were helpful to understand the clinical findings. Seventeen patients presented with subarachnoid FLAIR and T1 contrast enhancement that were focal or diffuse. Alterations of the blood-brain barrier resulting from endothelial invasion [10] may explain the inflammatory findings in CSF analysis, as well as the negative RT-PCR in CSF. Some patients also presented cerebral, predominantly white matter abnormalities which were hemorrhagic in 7 patients. Our results are consistent with recently published data who demonstrated brain MRI abnormalities [34‐37].

The main limitation of the study is the lack of comprehensive paraclinical tests in many patients and not all patients had the same set of studies performed. Yet, we could not perform more exams, because these were (i) extremely time-consuming (brain MRI) or not available (EEG/brain MRI), (ii) impossible in many patients because of hemodynamic/respiratory instability, agitation state, (iii) or contra-indicated (e.g., anticoagulant treatment and lumbar puncture). Further larger studies are thus needed to confirm the clinical and paraclinical characteristics of the COVID-19 delirium/encephalopathy, which could greatly participate to the burden caused by the unexpected SARS-CoV2 pandemic, and to confirm the generalization of our results, our cohort coming from one country population. In the same way, the identification of patients at risk of developing delirium as well as the underlying pathophysiological mechanisms should be also elucidated. Finally, we were not sure how to classify the aforementioned symptoms, whether delirium or encephalopathy.

The main strength of our study is to highlight the urgent need for post-ICU care reorganization. Indeed, in a context of severe COVID-19 pandemic overwhelming the capacities of ICUs [2, 38], one should bear in mind that COVID-19-associated delirium may lead to unusually long ICU stay and that it is not only a consequence of CoV2-induced ARDS [39]. Furthermore, whether patients will completely recover from these neurological symptoms is uncertain and should lead to organization of appropriate post-ICU care, including respiratory and neurological rehabilitation, and long-term medical follow-up.

SARS-CoV2 infection may be frequently associated with COVID-19 delirium and/or neurological symptoms, leading to sustained sedation and mechanical ventilation thus markedly worsening the prognosis. Our study highlights the importance of the organization of adequate post-ICU care, including respiratory and neurological rehabilitation, and long-term medical follow-up, considering the high incidence of COVID-19 delirium and/or neurological symptoms, the risk of long-term neurocognitive sequelae, and neuropsychiatric disorders in survivors. Delirium and/or neurological symptoms may be due to systemic inflammatory reaction to SARS-CoV-2, although the pathophysiological mechanisms involved require further investigations.