Delirium is an important public health problem. It is independently associated with worse clinical outcomes, including persistent cognitive impairment, increased mortality, and greater risk of institutionalization [1]. The prevalence of delirium is high in the intensive care unit (ICU), occurring in up to 70% of the sickest patients requiring mechanical ventilation [1]. Early studies in hospitalized patients with coronavirus disease 2019 (COVID-19) report delirium rates of 20–30%, which increase to 60–70% in severe illness [1]. An international multicenter cohort study that included 69 adult ICUs across 14 countries of 2088 COVID-19 patients reported that over a 21-day period, delirium had a prevalence of 55% and lasted a median of 3 days (IQR, 2–6 days) [2], which is more common and prolonged than that in non-COVID cohorts.

Despite its frequency, the pathophysiology of delirium remains poorly understood. Multiple factors are associated with delirium, many of which are coincident with critical illness (e.g., mechanical ventilation), and therefore clinical-pathological correlation has been difficult to prove as it is challenging to disentangle delirium attributable risk from the impact of general critical illness itself. Current evidence suggests that medications (e.g., benzodiazepines), systemic inflammation and/or the acute stress response may contribute to cerebral metabolic insufficiency, neuroinflammation (e.g., glial activation), neurotransmitter imbalances and deficiencies in neurological substrate and/or failure of network connectivity [3], ultimately leading to the development of delirium.

It was initially presumed that higher rates of delirium might be the result of direct neuronal SARS-CoV-2 infection. SARS-CoV-2 is a neurotropic virus that has the potential to enter the central nervous system (CNS) via angiotensin converting enzyme 2 (ACE2) receptors in the olfactory bulb [4]. Herpes simplex virus 1 infects the olfactory bulb and then the brain to cause encephalitis. Animal models have shown that some coronaviruses, including SARS-CoV, can do the same. However, in a single-cell RNA sequencing gene expression analysis of human biopsy samples, later confirmed in mouse models where deeper olfactory bulb tissue could be examined, ACE2 receptors were found in vascular cells (e.g., pericytes and immune cells of the macrophage/monocyte lineage) and not neurons [5]. Further, the SARS-CoV-2 virus has rarely been isolated from samples of cerebral spinal fluid (CSF), suggesting that viral replication within neurons is of less importance than other potential mechanisms such as immune-mediated damage within the CNS.

Indeed, the high proportion of delirium in patients with COVID-19 associated critical illness is likely due to microvascular disease and inflammatory mechanisms. Severe coagulopathy and vascular endothelial dysfunction leading to small vessel occlusions and microhemorrhages, evident on magnetic resonance imaging, contributed to delirium rates of greater than 80% in 150 patients with COVID-19-related critical illness admitted to two centres in France [6]. In mouse models, SARS-CoV-2 infection induces a hypermetabolic state and resultant hypoxic local environment, suggesting an underlying association between viral infection and ischemic infarcts [7]. Findings of territorial ischemic lesions (n = 6; 14%), activated microglia (n = 37; 86%) and infiltration by cytotoxic T-lymphocytes (n = 34; 79%) were seen in the post-mortem evaluation of 43 patients who died with SARS-CoV-2, 12 of whom died in ICUs [8]. A distinct microbleed phenomenon in cerebral white matter has been previously described in patients with general critical illness [9], the pathogenesis of which was attributed to hypoxemia from acute respiratory failure. Further work is needed to determine the SARS-CoV-2 dependent and independent factors or conditions responsible for microvascular and inflammatory changes seen in delirium.

Delirium remains a key independent predictor of cognitive impairment and dementia incidence at least 3 months following hospitalization [10]. Certain ICU-specific delirium phenotypes (e.g., hypoxic and septic) have been associated with greater risk of long-term cognitive impairment [11]. These states may perpetuate chronic neuroinflammation and neurotoxicity [12]. Work also is underway to examine how SARSCoV2 might contribute to prolonged cognitive impairment post hospitalization. Similar neurotropic viruses such as Middle East respiratory syndrome (MERS) and severe acute respiratory syndrome (SARS-CoV-1) coronaviruses have been shown to trigger the formation of Lewy bodies which are present in a range of neurologic diseases including Parkinson’s dementia. In a post-mortem analysis of the brains of rhesus macaques infected with SARS-CoV-2 virus, in the midbrain region was infiltrated by T-lymphocytes, activated microglia and intracellular Lewy bodies [13]. Such observations might herald a higher proportion or severity of critical illness induced long-term cognitive impairment or accelerated dementia (Fig. 1).

Fig. 1
figure 1

Risk factors for COVID-19-associated delirium

Apart from exploring the biologic mechanisms of SARS-COV-2 related delirium, it is also crucial to consider the importance of changes to standard ICU clinical practices during COVID-19. Resources have been strained (e.g., higher nurse: patient ratios) and there has been widespread reduction in family visitations. There have been reports of deeper levels of sedation, fewer spontaneous breathing trials, and limited mobility sessions in patients with COVID-19 as compared to other critically ill patients. In a worldwide two-day prevalence study of ABCDEF bundle practices across 212 ICUs in 38 countries amidst the first wave of the pandemic, rates of adherence to individual bundle items ranged from 16% for family engagement to 62% for sedation assessment [14]. This is in comparison to pre-pandemic rates of 67% and 89%, for family engagement and sedation assessment respectively [15]. Difficulties in reaching standards of care during times of strain may contribute both to increase rate of delirium and worsened long-term sequelae.

Long-term post-ICU follow-up will be instrumental to understand the full spectrum of health consequences from COVID-19-associated critical illness. Although outpatient clinics dedicated to follow-up are opening at many academic institutions, especially where large numbers of SARS-CoV-2 outbreaks have occurred, there is no standardized follow-up for survivors and their families. Creating such programs will enable standardized assessment of long-term cognitive outcomes in a representative population of critical illness survivors. Further, COVID-19 critical illness survivors potentially represent a relatively homogenous sub population with common mechanisms and offer an incredible opportunity to map out survivor biology and epidemiology, whilst generating evidence using trials to inform clinical care.