Delirium and Post-Discharge Neuropsychological Outcomes in Critically Ill Patients with COVID-19: an Institutional Case Series

Introduction

The outbreak of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), the virus that causes Coronavirus Disease (COVID-19), emerged as a public health threat in December 2019 and was declared a pandemic by World Health Organization in March 2020. Several studies have reported the serious nature of complications associated with this disease. Major neurological complications, such as encephalopathy, delirium, strokes, seizures, and ataxia, have all been observed.^1–5 Delirium appears to be a common complication, with previous investigations demonstrating an incidence of approximately 65-80% in the intensive care unit (ICU).^1,4 Delirium may occur due to direct coronavirus invasion into central nervous system, ⁶ and systemic inflammatory responses may further exacerbate neurocognitive impairment. In the ICU, synergistic factors such as sedation regimen, social isolation, and deviation from standard care protocols may further increase risk. Delirium is also associated with prolonged hospitalization, long-term cognitive and functional impairment, and increased mortality.^7–9 As such, there is a critical need to advance understanding of this syndrome in patients with COVID-19.

While a high incidence of delirium has been previously reported in COVID-19 patients, fundamental questions persist. The clinical course of delirium, including average duration and post-discharge cognitive trajectory, remains unknown. Pathophysiologic drivers of delirium (e.g., polypharmacy, inflammation, cerebrovascular events) are incompletely understood, and the extent to which standard prevention protocols are implemented is unclear. Such detailed understanding will contribute to delirium phenotyping of COVID-19 patients and provide insight into the clinical and neurocognitive burden associated with COVID-19. In this context, the objective of this study was to determine granular details associated with delirium in ICU patients with COVID-19. Specifically, the clinical course of delirium, presence of exacerbating factors, nature of prevention strategy implementation, and post-discharge cognitive outcomes were all characterized.

Results

Baseline characteristics are presented in Table 1. The majority of patients were African-American and non-Hispanic, and the most common comorbidities were hypertension, Diabetes mellitus, and obesity. Overall, age, demographics, and comorbidity profiles were similar between patients with delirium and those who did not experience delirium during hospitalization.

Neuroradiological Findings

In total, 47 patients underwent neuroimaging during hospitalization. The majority of imaging results were unremarkable or demonstrated incidental findings unrelated to COVID-19. However, some notable findings were present. A brain MRI was ordered for a 59-year-old female with COVID-19 pneumonia and recent ECMO decannulation due to severe encephalopathy (i.e., no response to commands or noxious stimulus) with preservation of brainstem reflexes. Imaging revealed abnormal fluid attenuated inversion recovery (FLAIR) hyperintensity affecting the occipital and temporal lobes (Figures 1A, 1B), microhemorrhage in the splenium of the corpus callosum (Figures 1B, 1C) and posterior leptomeningeal enhancement Figures 1C, 1D), suggestive of encephalitis. A 29-year-old female with history of ovarian malignancy presented with new-onset seizures, which for an MRI was ordered. Results revealed diffuse dural thickening and enhancement (Figure 2A) one day prior to positive COVID testing. The differential diagnosis included intracranial hypotension (recent lumbar puncture), inflammation, infection, and neoplastic processes. No definitive diagnosis was reached, though this enhancement resolved approximately one month later (Figure 2B). Lastly, one patient demonstrated diffuse parenchymal abnormalities on MRI suggestive of bilateral hypoxic-ischemic injury (Figure 3). This patient experienced two separate arrests (pulseless electrical activity) within the preceding three weeks. A non-contrast head CT two weeks later demonstrated poor sulcation bilaterally, suggesting global hypoxic-ischemic injury (Figure 4).

• Download figure
• Open in new tab

Figure 1.

Axial fluid-attenuated inversion recovery (FLAIR) (a, b) images at the level of the basal ganglia show abnormal FLAIR hyperintense signal (arrows) affecting the bilateral occipital, temporal lobes. This appears almost sulcal suggesting a higher protein component within the cerebrospinal fluid. Note the elevated FLAIR signal in the splenium of the corpus callosum (arrow) suggesting parenchymal insult. Axial susceptibility weighted imaging (SWI) (c) at the level of the splenium of the corpus callosum shows small areas of susceptibility (arrow) in the splenium, likely related to microhemorrhage. Axial T1 (d) post-contrast with fat suppression at the level of the basal ganglia shows subtle, though true, enhancement (arrows) in the posterior sulci, arachnoid pial (leptomeningeal) pattern suggesting a degree of encephalitis.

• Download figure
• Open in new tab

Figure 2.

Axial T1 post-contrast with fat suppression images at the level of the mid lateral ventricles demonstrate smooth dural enhancement (a) (arrows) along the bilateral cerebral convexities. This feature resolves one month later (b) (arrows). The overall pattern of dura-arachnoid/pachymeningeal enhancement is non-specific, can be seen with intracranial hypotension, in the procedural setting (e.g., lumbar puncture), and other scenarios (e.g., infection, inflammation).

• Download figure
• Open in new tab

Figure 3.

After initial non-contrast head CT (which was unremarkable), the patient had diffuse parenchymal abnormalities on MRI examination. Axial diffusion-weighted imaging (a) shows hyperintense signal (arrows) at the bilateral basal ganglia, thalami, and posterior cortices, regions are hypointense (arrows) on corresponding apparent diffusion coefficient map (b). These same locations are hyperintense on T2 (c) and FLAIR (d), especially at the basal ganglia and thalami.

• Download figure
• Open in new tab

Figure 4.

Non-contrast head CT performed approximately two weeks after cardiopulmonary arrest shows poor sulcation (arrows) bilaterally, suggesting global insult, most likely hypoxic-ischemic in etiology.

Discussion

In a cohort of ICU patients with COVID-19, delirium was a common complication, affecting more than 70% of patients. Furthermore, delirium was associated with prolonged hospitalization, increased length of ICU stay, discharge to skilled care facilities, and positive screens for neuropsychological impairment up to two months after discharge. Delirium occurred in the setting of multiple sedative-hypnotic agents, acute inflammatory responses, deviation from delirium prevention protocols, and cerebrovascular events, which are all factors that could have further catalyzed delirium precipitation. While ICU liberation activities were conducted, many of the interventions were infrequently implemented, particularly those that reduce delirium risk (e.g., spontaneous awakening trials). Overall, the burden of cognitive impairment was high in patients with COVID-19, as was the risk of related complications.

These results align with previous data demonstrating a high incidence of delirium in critically ill patients with COVID-19.^1–4 Cognitive dysfunction may occur as a result of direct coronavirus invasion of the central nervous system⁶ or other indirect mechanisms, such as polypharmacy, systemic inflammatory responses, or cerebrovascular events. Indeed, benzodiazepine sedation was common in this patient cohort, with nearly 60% of patients receiving midazolam sedation at one point during ICU admission. Lorazepam was a common sedation agent as well, and benzodiazepine use is associated with delirium in critically ill patients.^12–14 Inflammation may have also contributed to delirium risk. Inflammatory markers (e.g., c-reactive protein, ferritin, interleukin-6, lactate dehydrogenase) were considerably elevated in this patient cohort. In fact, serum levels observed in this study aligned with – or exceeded – previously reported values in patients with severe COVID-19,^2,5 and there was MRI evidence of neuroinflammation for at least two patients in this series. Specific inflammatory markers (e.g., c-reactive protein, lactate dehydrogenase) were elevated in delirium patients and may reflect a specific in neuroinflammatory process. Indeed, c-reactive protein increases blood-brain barrier permeability in basic science models,¹⁵ and lactate dehydrogenase has been associated with neurocognitive disorders.^16,17 However, these were unadjusted, bivariable analyses, and further covariate adjustment may have revealed a reduced – or absent – association.⁵ Lastly, cerebral ischemia may also contribute to delirium risk in patients with COVID-19. Severe hypoxic-ischemic injury occurred in a patient who experienced multiple cardiopulmonary arrests during the course of illness. Stroke has previously been reported in patients with COVID-19,¹⁸ as thromboembolic phenomena and cerebral malperfusion may both occur during the clinical course of COVID-19. As such, multiple processes likely contribute to delirium in patients with COVID-19. Targeted case-control studies can address some of these potential risk factors, such sedative-hypnotic regimens and inflammatory profiles.

Delirium prevention and management are inherently challenging for COVID-19 patients in the ICU. While delirium prevention bundles have been consistently demonstrated to reduce risk,^19,20 unique challenges posed by COVID-19 hinder the implementation of standard prevention practices. Spontaneous awakening and breathing trials, for example, were often not possible due to illness severity and associated ventilator requirements. Indeed, the median number of total spontaneous awakening trials charted was two across the entire cohort. Clinicians may have also been limited in terms of sedation regimen. Agitation was commonly observed, and nearly one-third of patients required antipsychotics. Agitation and hyperactive delirium likely prompted additional sedation and prolonged use of physical restraints. Early mobility was limited given illness severity, and family engagement was often not possible due to visitor policy restrictions. In-person interactions with clinicians were also limited given the intent of reducing virus transmission. As such, the culmination of disease severity, limited face-to-face time spent with patients, and visitor restriction policies likely hindered ICU liberation bundle implementation. Indeed, given that the median length of ICU stay was 15 days across the cohort, most interventions were charted less than once daily (Table 2). Limited implementation of delirium prevention bundles could have increased risk for delirium incidence, duration, and severity. Interestingly, prevention bundle activities were charted more commonly in delirious patients. This may have reflected clinician awareness of delirium and attempts to reduce risk. Novel strategies for implementing delirium prevention bundles in this patient population may help to further mitigate risk and should be tested in prospective trials.

Neuropsychological impairment after discharge was also present for some patients based on subjective reporting, caretaker assessment, and objective testing for depression and cognitive impairment. Furthermore, all patients that screened positive for possible impairment also experienced delirium in the hospital. These estimates may have been even higher, given that approximately 70% patients that were called for post-discharge cognitive assessment were still admitted to skilled care facilities, refused participation, or were unable to be reached. Whether post-discharge cognitive impairment was related specifically to COVID-19 or critical illness more broadly is unclear. Indeed, cognitive impairment is common at discharge for patients who experienced delirium while in the ICU, and delirium is present for nearly 20% of patients newly admitted to acute care facilities.^7,21 Moreover, cognitive impairment can be present for months-to-years after acute respiratory distress syndrome and sepsis,^22–24 and symptoms of depression and post-traumatic stress disorder are commonly reported among ICU survivors.²⁵ Neuropsychological impairment after discharge may, in part, reflect critical illness, rather than pathophysiologic insults specific to COVID-19. Nonetheless, ICU patients with COVID-19 experience considerable neuropsychological burden, and related complications, both during and after hospitalization.

The strengths of this study include granular data with respect to delirium, potential risk factors, nature and frequency of prevention strategies, related complications, and post-discharge outcomes. Data were representative of an academic tertiary care center with nearly 150 patients. A validated chart review method was used to identify delirium,²⁶ and the study measures used to characterize cognitive function, such as the FAM-CAM, Short Blessed Test, and PROMIS assessments, are validated measures that increase confidence in the results. In terms of limitations, this is a this was a single center analysis, and the results are restricted to the institution studied. The study was not conducted with a matched control cohort, as the nature of this study was descriptive. Lastly, data were limited for post-discharge cognitive outcomes, as more than half of patients called were unavailable to complete assessments.

In summary, delirium is common complication of COVID-19 with multiple contributing factors. Furthermore, neuropsychological impairment may persist in some patients after discharge. Further research should aim to identify independent risk factors in this population and novel, effective prevention strategies.

Methods

Author Contributions

The study was originally conceived by J.R., A.M., M.Z., and P.E.V. Data acquisition was conducted by J.R., A.M., M.Z., J.B., M.H.C., Y.A., and P.E.V. Data analysis was conducted by J.R., A.M., M.Z., J.B., M.H.C., M.I., Y.A., and P.E.V. Final statistical analysis was conducted by P.E.V. Neuroimaging studies were acquired and analyzed by R.L., and the final neuroimaging figures were generated by R.L. and P.E.V. All authors contributed to the manuscript writing, critically reviewed the manuscript for intellectual content, and approved the final manuscript.

Competing Interests Statement

The authors declare no competing interests.