The use of non-invasive multimodal neuromonitoring in adult critically ill COVID-19 patients

Abstract

Introduction:

Non-invasive neuromonitoring could be a valuable option for bedside assessment of cerebral dysfunction in COVID-19 patients admitted to intensive care units (ICUs). This systematic review aims at investigating the use of non-invasive multimodal neuromonitoring in critically ill adult COVID-19 patients.

Methods:

MEDLINE/PubMed, Scopus, Cochrane, and EMBASE databases were searched for studies examining the use of non-invasive neuromonitoring (transcranial doppler, TCD; Brain4care Corp. for cerebral compliance, B4C; optic nerve sheath diameter, ONSD; near infrared spectroscopy, NIRS; pupillometry; and electroencephalography, EEG) in COVID-19 patients admitted to ICUs.

Results:

We identified 32 studies from a systematic search of 7001 articlesregarding the use of neuromonitoring in ICU COVID-19 patients: 1 study on TCD, ONSD and pupillometry; 2 studies investigating brain compliance with the B4C tool and TCD; 3 studies on NIRS and TCD; 4 studies on TCD; 1 case series on pupillometry; and 21 studies on EEG. One-hundred nineteen patients underwent TCD, 47 pupillometry, 49 ONSD, 50 B4C, and 900 EEG. We found that altered cerebral hemodynamic, brain compliance, brain oxygenation, pupillary response, and brain activity are common in patients with COVID-19 admitted to ICU but not clearly associated with worst outcome and development of new neurological complications.

Conclusions:

The use of non-invasive multimodal neuromonitoring in COVID-19 ICU patients could be considered to facilitate the detection of neurological derangements. Determining whether these findings facilitate the earlier detection of neurological complications or guide appropriate therapy requires additional studies.

Introduction

Neurological complications from coronavirus disease-2019 (COVID-19) can occur either as presenting symptoms or during hospitalization and intensive care unit (ICU) stay^1,2. However, diagnosis of these complications is often delayed, as severe COVID-19 patients may require deep sedation and the use of neuromuscular blockade to facilitate mechanical ventilation. Accurate neurological evaluations are, therefore, rarely feasible in these patients. Brain imaging may have a role in preventing diagnostic delay of neurological complications. However, brain imaging may be time-consuming and often difficult and potentially unsafe to perform in unstable, critically ill patients, and may pose risk of viral transmission to others during transfer³. For all these reasons, bedside imaging during the pandemic has undeniable pragmatic advantages.

Non-invasive neuromonitoring may be considered to detect neurological abnormalities at the bedside in COVID-19 patients³. Among the different available techniques, transcranial doppler (TCD), optic nerve sheath diameter (ONSD), automated pupillometry, near infrared spectroscopy (NIRS) and electroencephalogram (EEG) have been applied in this patient population^1,4,5. Each monitoring tool has specific advantages. For example, TCD allows the assessment of cerebral blood flow, ONSD provides information on cerebrospinal fluid pressure, NIRS allows investigating cerebral oxygenation, automated pupillometry can accurately detect pupillary abnormalities and brainstem impairment, and EEG can detect seizures and background abnormalities or generalized slowing, as in cases of encephalopathy⁶. We therefore provided a systematic review aiming at investigating the use and applications of non-invasive multimodal neuromonitoring in critically ill adult COVID-19 patients.

Methods

This systematic search was conducted in accordance with the Joanna-Briggs Institute (JBI) Reviewer’s Manual for Systematic Reviews of Literature and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines⁷. The study protocol was registered and published on July 5^th, 2021 (International Prospective Register of Systematic Reviews–PROSPERO-CRD42021265617).

Search strategy

Relevant studies were identified by searching in the, PubMed, Scopus, EMBASE, and Cochrane electronic databases for pre-prints and articles. We also identified studies by citation searching from the bibliography of each relevant study, as reported in Figure 1 “Identification of studies via other methods”. A complete list of search and MeSh terms is reported in the Supplemental Digital Content (SDC)—Item 1. No language restrictions were applied.

Figure 1. PRISMA 2020 flow diagram for new systematic reviews which include searches of databases, registers and other sources.

This figure depicts the number of records identified from each database or register searched (rather than the total number across all databases/registers). From: Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021; 372:n71. doi: 10.1136/bmj.n71. http://www.prisma-statement.org/.

Eligibility criteria and study selection

Articles were considered for inclusion from the selected databases from January 1^st, 2020, to February 14^th, 2022. Studies were considered eligible if: the study population included adult patients (≥18 y/o) diagnosed with COVID-19 by positive reverse transcription polymerase chain reaction and the study described the use of non-invasive neuromonitoring in COVID-19 ICU patients. Studies performed in pediatric-age patients were excluded. Other information about eligibility criteria can be found in the SDC –Item 2. Titles and abstracts were independently screened in duplicate by two authors (DB and LP). When necessary, the corresponding authors of selected articles were contacted to obtain missing data.

Data extraction and synthesis

Data were extracted from all eligible studies by two independent reviewers (DB and LP) in accordance with the PICO approach. The variables extracted from the selected articles and data synthesis are presented in the SDC—Item 3.

Assessment of methodological quality

Two authors (DB and CR) independently assessed methodological quality using the JBI appraisal tools. The quality of case reports and case series was assessed with the modified 8-item Newcastle Ottawa Scale (NOS)⁸, while the quality of case-control and cohort studies was rated using the COVID-19 adapted NOS⁹, as illustrated in SDC—Item 4.

Definition of neurological derangement

Definitions of neurological derangements by neuromonitoring tools are reported in SDC –Item 5.

Results

A total of 7,001 articles were extracted during our systematic search. After eliminating duplicates, records marked as ineligible and full-text reviews, 32 studies conducted in an ICU setting were included (Figure 1). The included studies were as follows: 1 retrospective study on TCD, ONSD and pupillometry; 2 prospective studies investigating brain compliance with the B4C tool and TCD; 3 prospective studies on NIRS and TCD; 4 prospective studies on TCD; 1 case series on pupillometry; and 21 studies on EEG (4 prospective, 17 retrospective observational). A total of 1,199 ICU patients were included. A significant variability was observed between different countries in the use of neuromonitoring in ICU patients (Figure 2). A total of 119 ICU patients underwent TCD monitoring, 47 pupillometry, 49 ONSD, 50 B4C tool, 25 NIRS, and 900 EEG.

Figure 2. Geographical distribution of ICU patients who underwent neuromonitoring.

This figure summarizes the geographical distribution of ICU patients with COVID-19 who were neuro-monitored. The use of neuromonitoring is highly heterogeneous among countries, in terms of the number of patients included in published reports. Data are reported as the percentage of ICU patients who underwent neuromonitoring. Blue circles represent the proportion of patients who were neuro-monitored in each country (the larger the circle, the larger the number of patients). ICU, intensive care unit.

The timing from ICU admission to the first neuromonitoring assessment was highly heterogeneous among studies (range 3–21 days). Previous neurological comorbidities were highly variable as outlined in SDC—Item 6. New neurological complications were detected in 18 studies by clinical signs, brain imaging or neuromonitoring. The most common neurological complication in patients with COVID-19 was altered mental status, delirium, seizures, meningitis/ encephalitis, stupor/ coma, ischemia/ hemorrhage, and hypoxic brain injury. Neurological complications, methodological features, and outcomes associated with neuromonitoring are summarized in SDC – Items 6 and 7. The most common indication for the use of neuromonitoring was altered mental status, followed by suspected seizures, and delayed awakening.

Critical appraisal of quality of studies

Of the 32 articles included for review, 3 were rated of low quality, 12 moderate quality, and 17 high quality.The methodological quality is listed in SDC—Item 8.

Transcranial Doppler

The median time from ICU admission to TCD varied from 3 to 15 days. The rationale for using TCD in COVID-19 was to observe cerebral hemodynamic changes in five studies; to detect microbubbles or micro-emboli in three studies; and to assess pulmonary shunting in one study.

Intracranial pressure (ICP) and cerebral perfusion pressure (CPP) with TCD

Battaglini et al.¹ found high ICP in 21 patients (40%). Brasil et al.^10,11 found high ICP in 7 (24%) patients, while abnormal CPP in 14 (48%) patients. Robba et al. found that non-invasive ICP measured by TCD (nICP_TCD) significantly increased after passive leg raising test, fluid challenge, and norepinephrine bolus in COVID-19 patients. Respiratory rescue therapies like prone position, inhaled nitric oxide, extracorporeal carbon dioxide removal (ECCO2R), and recruitment maneuvers significantly increased nICP_TCD, while decreasing CPP.

Cerebral compliance assessment with TCD

Brasil et al. assessed cerebral compliance using the B4C tool in a cohort of 50 ICU COVID-19 patients¹⁰. Altered cerebral compliance, defined as tidal ICP/percussion or systolic ICP peak waves (P2/P1)>1, was observed in 33/50 (66%) patients^10,11. The authors found that cerebral blood flow velocity, pulsatility index (PI), nICP, CPP, and P2/P1 values of B4C tool were similar between non-survivors and survivors, and the proportion of patients with abnormal values were similar between groups.

Optic nerve sheath diameter

One study¹ found (n=49) a mean value of ONSD of around 6 mm. The nICP_ONSD was found to be higher than normal in 10/49 (20%) patients and was associated with longer ICU length of stay.

Near infrared spectroscopy

Three studies^4,5,12 investigated regional cerebral saturation of oxygen (rSO₂) before and after a hemodynamic (i.e., passive leg raising test, fluid challenge, and/or norepinephrine) or respiratory rescue maneuver (recruitment maneuver, extracorporeal carbon dioxide removal, prone positioning, etc.), and as possible surrogates of cerebral blood flow, with low baseline rSO2 detected in 19/23 (83%) patients. Passive leg raising, fluid challenge, and norepinephrine all resulted in increased mean arterial pressure and rSO₂. With a recruitment maneuver or ECCO2R, rSO₂ decreased significantly, while it increased during prone position.

Automated pupillometry

Battaglini et al.¹ found altered pupillary reactivity in 9/29 (31%) ICU patients. Another study investigated pupillary reactivity in 18 ICU patients who either were deeply/moderately sedated (6 patients) or completely awake (12 patients), in terms of the mean percentage of reactivity, finding significant differences between sedated and non-sedated patients¹³.

Electroencephalography

The time from hospital-ICU admission/onset of symptoms to EEG recording was highly heterogeneous, ranging from 3 to 18 days. Reasons for EEG monitoring included: identifying EEG features in COVID-19 ICU patients; investigating the prevalence of seizures and other EEG abnormalities; assessing EEG patterns in patients suspected of having encephalopathy; and investigating for cerebral dysfunction in COVID-19 ICU patients as summarized in SDC—Item 7. Most common abnormal EEG findings were abnormal background abnormalities (with possible slowing background delta/theta, or alpha and beta, posterior dominant alpha rhythm, burst attenuation, burst suppression); electroencephalographically confirmed seizures or, rhythmic and/or periodic patterns (periodic discharges, rhythmic delta activity, spike and wave, or sharp and wave). Association between EEG and outcome was investigated by ten studies are summarized in SDC–Item 7.

Discussion

The main finding of this systematic search is that altered cerebral hemodynamics, brain compliance, brain oxygenation, pupillary response, and brain activity are common in patients with COVID-19 admitted to ICU and are sometimes associated with worse outcomes and the development of new neurological complications. Indeed, the COVID-19 population is at particular risk for both short-and long-term neurological sequelae². This makes systemic and cerebral hemodynamic stability of particular interest. However, none of the studies were able to definitively determine whether use of neuromonitoring in this patient population altered outcome. The use of multimodal neuromonitoring in the non-brain injured population remains limited and under-investigated.

We found that, when using TCD and ONSD, an increased ICP and PI, and reduced CPP were frequently detected in COVID-19 patients. Altered neuromonitoring findings in critically ill patients with COVID-19 seem to be similar to those found in septic patients without COVID-19, in whom altered cerebral autoregulation and mechanical ventilation can be possible independent predictors of sepsis-associated brain dysfunction¹⁴. Both conditions may be associated with acute deterioration in mental status driven by cytokines and inflammation during severe infections with high likelihood of multiorgan failure¹⁵ as well as cerebral autoregulation impairment^14,16. Acute respiratory failure, that can induce hypercarbia and subsequent cerebral vasodilatation, may modulate cerebral hemodynamic impairment.

Few studies suggest that monitoring cerebral oxygenation via NIRS could be useful, for example during the use of respiratory rescue therapies, which can have specific effects on cerebral oxygenation⁵. Similarly, hemodynamic tests, such as a fluid challenge or norepinephrine bolus, can impact cerebral hemodynamics differently⁵. However, data obtained from COVID-19 patients are still very poor, limiting the possibility of developing clear conclusions regarding the use of NIRS in this patient-population.

Pupillary reactivity and diameter were reported as impaired in many COVID-19 ICU patients. This finding may be biased because of the common use of sedatives and paralytics. Alternatively, this could indicate a status of dysautonomia, such as autonomic impairment, orthostatic intolerance, and cognitive deceleration in this population¹⁷.

Lastly, EEG was widely used in ICU patients with COVID-19 and was able to reveal many different abnormal EEG phenotypes and seizures. EEG is a good beside neuromonitoring tool for investigating brain function in patients who present with new neurological findings¹⁸. COVID-19 patients admitted to the ICU often need sedative and paralytic medications, which could explain the commonly observed altered EEG findings. Indeed, in non-COVID-19 ICU patients, including those with sepsis, these alterations are very common and frequently associated with the degree of sedation¹⁹.

In summary, our results suggest that derangements in cerebral hemodynamics are common in ICU patients with COVID-19. The timing between admission to the ICU and neuromonitoring is frequently long, suggesting that COVID-19 patients requiring ICU admission could benefit from earlier neuromonitoring assessments, although no data can support their routine use yet.

Limitations

This systematic review has several limitations: studies with small sample sizes from single center that are observational and retrospective and lack control groups. The indications for neuromonitoring were highly variable and not always clear. Neuromonitoring was not standardized, and the timing of monitoring was not controlled. Finally, none of the studies included data on long term outcomes. The outcome was not clearly defined in each study with high heterogeneity, which limits our study in making any concrete statement on the impact of neuromonitoring on the outcome.

Conclusions

The use of non-invasive multimodal neuromonitoring in patients with COVID-19 is highly variable but can be able to detect a range of neurological abnormalities, despite their effect on outcome is still unknown. Non-invasive neuromonitoring tools are quick, low-cost, safe, and easy to use, and can provide relevant clinical information at the patient’s bedside. Determining whether the findings of such monitoring aids in the early detection of neurological complications and effectively guides appropriate therapy requires additional studies.