Delirium and COVID-19: a narrative review of emerging evidence
Summary
Delirium is a common condition affecting hospital inpatients, including those having surgery and on the intensive care unit. Delirium is also common in patients with COVID-19 in hospital settings, and the occurrence is higher than expected for similar infections. The short-term outcomes of those with COVID-19 delirium are similar to that of classical delirium and include increased length of stay and increased mortality. Management of delirium in COVID-19 in the context of a global pandemic is limited by the severity of the syndrome and compounded by the environmental constraints. Practical management includes effective screening, early identification and appropriate treatment aimed at minimising complications and timely escalation decisions. The pandemic has played out on the national stage and the effect of delirium on patients, relatives and healthcare workers remains unknown but evidence from the previous SARS outbreak suggests there may be long-lasting psychological damage.
Introduction
Delirium is a common serious neuropsychiatric syndrome. It refers to an acute, global disturbance of cognition that is seen in acute illness [1]. Delirium can affect people of all ages, but it predominantly affects hospitalised older adults and those in critical care. It is associated with increased morbidity and mortality [2]. Historically, the nomenclature of delirium has been complicated by use of terms such as acute confusional state, acute brain dysfunction, acute brain failure and altered mental status. Recent consensus has been reached, defining delirium as the clinical manifestation of an acute encephalopathy – a rapidly developing (< 4 weeks) pathobiological brain process [3]. Delirium is a heterogeneous and fluctuating syndrome resulting from varied precipitants [4, 5]. Peripheral causes may be multifactorial, but for the purpose of this review of delirium in COVID-19, we will focus on infection as the precipitant.
Respiratory infections are a common precipitant of delirium [6]. SARS-CoV-2 is the novel coronavirus that causes COVID-19. COVID-19 was declared a pandemic on 11 March 2020 by the World Health Organization. Although it poses a risk to all ages, adults aged ≥ 65 y are at greatest risk of severe disease, hospitalisation, intensive care admission and death [7]. Classical symptoms are a high temperature; a new, continuous cough; and a loss or change to the sense of smell or taste. However, presentation in older adults may be atypical. Guidance from the Regional Geriatric Program of Toronto includes the following advice: ‘typical’ symptoms of COVID-19 such as fever, cough and dyspnoea may be absent in the elderly despite respiratory disease. Only 20–30% of geriatric patients with infection present with fever; atypical COVID-19 symptoms (in older adults) include: delirium; falls; generalised weakness; malaise; and functional decline [8]. As the pandemic has developed, it has become clear to clinicians that the prevalence of delirium was greater, and the clinical course more complicated, than would be expected from historically more common respiratory viruses such as influenza. The frequency and intensity of complications associated with COVID-19 have precipitated interest and research in the area of COVID-19-associated encephalopathy [9].
Recognising, diagnosing, managing and preventing delirium in these novel and adverse circumstances has come with unique challenges. These include a lack of staff and a paucity of space, limiting the ability to utilise classical non-pharmacological treatments. Use of personal protective equipment, which restricts classical visual cues, enforced isolation and the lack of visiting has compounded the challenges and potentially promoted an excess use of pharmacological drugs. Currently, there are no published data on in-hospital use of benzodiazepines or antipsychotics during the pandemic, but there has been cause to question the usual ‘low and slow’ approach to antipsychotic prescriptions [10]. Alongside this, community use of benzodiazepines has risen in many countries [11] as has the use of antipsychotic prescriptions in older people with dementia [12]. While these are unique and difficult challenges, the pandemic has propelled delirium to the scientific and public consciousness. From a research viewpoint, it has provided the opportunity to study delirium precipitated by a single insult in a relatively homogeneous population.
Our review aims to summarise the current evidence available on the clinical presentations, epidemiology, outcomes and pathophysiology of COVID-19 delirium, and how this differs from what we term ‘classical delirium’. The review will report only on in-hospital studies, covering both critical care and general ward admissions. We will build on our experience of working in ward and critical care environments during the pandemic where evidence is sparse.
Delirium phenomenology
Delirium is a syndrome with a wide range of clinical expressions, in part due to the variety of precipitants and underlying vulnerabilities in patients with delirium. The clinical signs observed reflect impairment in several cognitive domains, but deficit in attention is viewed as a core feature. Other signs include impairment in short-term working memory, orientation, comprehension, vigilance, visuospatial ability and executive function [13]. Disorientation to time, place and identity is common and is used as a core symptom to screen for delirium in clinical settings. However, this is insensitive due to the fluctuating nature of delirium [14, 15].
There are many subtypes of delirium described and delirium may lie on a diagnostic spectrum including subsyndromal delirium and severe delirium. Subsyndromal delirium was first described by Lipowski in 1983 [16], and has subsequently been defined by the presence of any core delirium symptoms or severity scores on rating scales that are below the diagnostic threshold [17].
Motor subtypes present as hypoactive delirium (with psychomotor retardation and reduced consciousness) and its opposite, hyperactive delirium (with psychomotor excitation, distress and hyper-alertness). Further attempts to describe sub-phenotypes by, for example, aetiology, mechanism or underlying dementia, may improve our understanding going forward. This is particularly the case in ICU populations, with the added complexity of delirium occurring in patients receiving sedative medications [18].
Delirium phenomenology in COVID-19 has been studied as part of a cluster of observed neurological manifestations and these have been expertly collated and explored [19]. However, there is little in the literature about delirium phenomenology per se in COVID-19. Anecdotally, the severity and duration of delirium appeared greater and, in our experience, there was a need for higher doses of antipsychotics to treat the symptoms of extremely distressing hyperactive delirium. There was also a marked hyperactive phenotype seen in older people at the end of life alongside trials of less typical sedative drugs such as dexmedetomidine [20]. We observed a number of patients with prolonged delirium (i.e. > 1 week), especially in the post-intensive care cohort.
Prevalence and incidence
The prevalence of delirium in hospitalised older adults is estimated at 23% (95%CI 19–26%) [21]. In critical care settings, prevalence ranges from 31% (95%CI 24–41%) in all patients to 50% in mechanically ventilated patients [22], and 34% in critically ill children [23]. Non-COVID-19-related delirium remains underdiagnosed in up to three-quarters of hospitalised adult patients [24]. Incomplete understanding of delirium and resultant lack of education among healthcare professionals, alongside avoidant behaviours towards a clinically challenging patient group, contribute to this. Guidelines in critical care and ward areas recommend regular screening for delirium in the unwell patient [15, 25] to improve the current rate of delirium detection, which remains low. Only about one-third of delirium is recognised by routine clinical teams, across settings [26, 27]. Formal diagnosis should be made using the Diagnostic and Statistical Manual on Mental Disorders, Fifth Edition [15, 28, 29].
Studies of delirium to date in patients with COVID-19 have suggested prevalence rates from 12% to 84% [30-32]. This disparity reflects the challenges in recognition and the diversity of settings. In a large systematic review and meta-analysis of 48 studies with 11,553 COVID-19 patients from 13 countries, delirium prevalence was 24.3% (95%CI 19.4–29.6%). There was a higher prevalence in those aged > 65 y (where it developed in almost one out of every three individuals). Studies were mainly in ward settings. However, in those aged < 65 y, delirium was present on arrival in 15.7% (95%CI 9.2–23.6%) and delirium acquired in hospital was 71.4% (95%CI 58.5–82.7%), suggesting delirium had a greater chance of developing during treatment in younger people who were more likely to be in critical care settings [33].
The prevalence and incidence of COVID-19-associated delirium in the ICU setting is even higher than in general adult hospital admissions. In a large, multicentre cohort study including 69 adult ICUs, across 14 countries, 54.9% of people had delirium for a median duration of 3 days. Acute brain dysfunction (coma or delirium) affected patients for a median of 12 days (IQR 7.0–18.0 days). However, this study underlines some of the problems with delirium prevalence research. Of the 4530 patients admitted to the included ICUs, 1008 were not screened because they were admitted to an ICU that did not monitor for delirium; a further 1036 were admitted to sites that did not have capacity to screen eligible patients or collect data during the study period. This highlights both the lack of standardised international protocols for recognising and diagnosing delirium and the additional pressures on staff and resources experienced during the pandemic [34].
Like non-COVID-19 delirium, what we know about the incidence and prevalence of COVID-19 delirium is limited by the difficulties of screening and diagnosis. However, the prevalence of delirium associated with COVID-19 appears greater than that seen in delirium from other causes [32, 34]. In addition, longer-term consequences (while currently relatively unknown) may result in significant individual, societal and financial burden [21].
Difference in outcomes
Historically, there has been a significant burden associated with a diagnosis of delirium, even when controlling for other factors, and this is consistent across ICU and non-ICU settings [35, 36]. Delirium is associated with prolonged hospital stay, days on a ventilator, incidence of later-life dementia, admission to long-term institutional care and mortality [37].
The association between baseline vulnerability and pre-morbid frailty may not be as simple as delirium being an expression of baseline frailty. Dani et al. [38] reported that the excess mortality in delirium may be greater when it occurs in previously fitter individuals, perhaps due to the precipitant of delirium in a ‘fit’ individual representing more severe disease. Most studies have identified delirium as an independent predictor of death in the ICU, although some smaller studies have found no association [39, 40]. These inconsistencies have been explained by differences in case-mix, the tools used for the assessment of delirium and the study design. This reiterates the problem with a lack of standard screening and in-depth specialist assessment.
The mechanisms by which delirium independently increases the risk of death (after adjusting for illness severity) remain unclear. Delirium complicates and impairs recovery. Patients with delirium are more likely to receive psychotropic drugs, more likely to fall and less likely to mobilise effectively during and after their illness, all of which may have an adverse impact on survival [41]. In addition, patients with delirium are less likely to maintain adequate hydration and nutrition and be less compliant with healthcare interventions. Of particular concern might be hypoactive delirium, where reduced arousal might lead to avoidable complications such as aspiration pneumonia and acute kidney injury further spiralling a downwards trajectory of worsening delirium. All of these factors will have been heightened during the pandemic.
We are beginning to see some of the short-term outcomes of delirium following COVID-19. As with classical delirium, COVID-19 delirium is associated with increased mortality [33, 37]. In a meta-analysis of 17 studies reporting the difference in mortality between people with COVID-19 with and without delirium, the pooled mortality rate was 44.5% (95%CI 36.1–53.0%) in the delirium groups and 21.2% (95%CI 15.4–27.6%) in those aged < 65 y. This meta-analysis reports an unadjusted odds ratio of 3.12 (95%CI 2.1–4.8) for short-term mortality in those with delirium compared with those without. A large UK study demonstrated the contribution to mortality of delirium in those with COVID-19 was lost when controlling for frailty and age [42]. More data are needed to be clear about the impact of disease severity but it is plausible that this will be the same as non-COVID-19 delirium. In the medium-term, delirium in COVID-19 appears to be associated with adverse outcomes in survivors. In a small UK cohort, COVID-19 delirium was associated with worse functional performance at 4 weeks – although cognition and death rates were no worse in those with and without delirium [41]. Delirium is a recognised risk factor for dementia and cognitive decline [43, 44], but it remains to be seen if the same applies for COVID-19 associated delirium [41].
Overall, these findings emphasise the emergency nature of delirium and the short-term consequences. In the longer term, the sequelae of delirium associated with COVID-19 are still to be determined, particularly any possible association with long COVID [45].
Pathophysiology and potential mechanisms: how does it happen?
The fundamental key to understanding the pathophysiology of delirium is to describe the mechanism of the precipitant (e.g. the immune-inflammatory response to infection), and how it interacts with delirium vulnerability (ageing and neurodegenerative disease). Precipitants of delirium are varied and include indirect (or peripheral) precipitants such as surgery, infection, pain, as well as direct insults to the central nervous system. There are several distinct neurobiological mechanisms (or combinations thereof) involved in delirium pathogenesis. Unravelling these mechanisms in heterogeneous populations has been a huge challenge to delirium research [1]. COVID-19 has allowed delirium to be studied in a single disease and in a relatively homogenous population. The flurry of rapid research in this area is constantly changing our understanding. In an attempt to identify the individual aetiologies associated with delirium (described as toxic metabolic encephalopathy) in COVID-19-positive patients [46], the most common aetiologies were septic encephalopathy (n = 347/559 [62%]), hypoxic-ischaemic encephalopathy (n = 331/559 [59%]) and uraemia (n = 156/559 [28%]). Multiple aetiologies were present in 435 (78%) patients. A short summary of the common pathophysiological mechanisms involved in delirium, and how they may be highlighted in people with delirium and COVID-19 as a precipitant follows.
Neuronal cell dysfunction
Serum neurofilament light chains are a peripheral marker of neuroaxonal injury that have been shown to be associated with postoperative delirium and long-term delirium sequelae [46]. Neurofilament light chains have been reported to be associated with the severity of COVID-19 [47, 48] and neurological complications. Similarly, glial fibrillary acidic protein (a marker of blood–brain barrier disruption) has been associated with disease severity and neurological complications [Hay et al. preprint, https://doi.org/10.1101/2021.04.28.21256277]. This confirms that COVID-19 is broadly neurotoxic.
Peripheral inflammation
Peripheral inflammation is a well-established trigger of delirium. Currently, the pathophysiological mechanisms by which it disrupts brain function are not fully understood. People with delirium would appear to have an amplified peripheral inflammatory response, perhaps orchestrated by changes in immune cell function. In the case of COVID-19, both innate and adaptive immune responses to SARS-CoV-2 infection precipitate an amplified immune to brain communication and hence a neuroinflammatory response with encephalopathy – seen clinically as delirium [49]. In a study comparing inflammation between patients with COVID-19 delirium and delirium in other medical disorders, C-reactive protein concentrations were significantly higher in the COVID-19 group (mean (SD) 81.7 (80.0) vs. 58.8 (87.7) mg.ml-1, p = 0.04), and white blood cell counts were significantly lower (7.44 (3.42) vs. 9.71 (5.45) 109.ml-1, p = 0.04). Other studies have replicated the predictive effect of C-reactive protein in delirium in COVID-19, further supporting an inflammatory basis for delirium in COVID-19 [50]. Neuropathological studies have demonstrated blood–brain barrier breakdown, with inflammation of the endothelium and fibrinogen extravasation – all hallmarks of increased immune to brain communication [51]. Impaired blood–brain barrier mechanisms have been demonstrated in animal models of COVID-19 [52].
Direct CNS effects
There is surprisingly little evidence to support direct neurotropism of SARS-CoV-2 infection, with conflicting studies mostly reporting no evidence of SARS-CoV-2 within the central nervous system [53, 54]. As with other viral infections causing encephalitis, such as herpes simplex, neuro-inflammation may be secondary to increased leucocyte transmigration across the blood–brain barrier [55].
The brain requires a large amount of energy to function and oxygen or glucose deficiency can markedly reduce brain function. The long-standing ‘cerebral metabolic insufficiency’ hypothesis proposes that delirium is caused by the failure to meet the energy requirements of the brain [56] and cerebral glucose hypometabolism has been demonstrated in people with COVID-19 and cognitive impairments [57]. Delirium affects one in four people with stroke, and stroke was reported as a common complication in severe COVID-19 [58, 59]. Microvascular endotheliopathy causing cerebral small vessel ischaemia may also contribute [60]. Severe hypoxia and hypotension in severe disease will also precipitate delirium. Drug use and medication changes can precipitate delirium, sometimes by direct effects on brain neurotransmitter systems, such as acetylcholine and dopamine. In addition, polypharmacy and renal and/or hepatic dysfunction may lead to inappropriately high or prolonged blood drug concentrations. COVID-19 was also associated with hyponatraemia, which can precipitate delirium [61], alongside uraemia, in the context of acute renal dysfunction [1].
While the underlying pathophysiological processes in COVID-19 are likely to be similar to non-COVID-19 delirium, the evidence suggests that a peripheral hyper-inflammation might be more common, coupled by extreme physiological changes driven by COVID-19 severity. It is worth noting pre-clinical animal studies have reported this may be due to increased senescent cells (a hallmark of biological ageing) causing an accelerated inflammatory response to coronavirus, which is rescued by drugs that reduce senescent cell numbers [62, 63]. This may go some way to explaining the increased disease severity, delirium and mortality seen in older people with COVID-19 (Fig. 1).

Management
How to treat COVID-19 delirium
The COVID-19 pandemic provided a perfect storm of increased delirium prevalence, increased severity, coupled with resource limitations. Although the basis of management of COVID-19-associated delirium should not be considered different to that of standard delirium as discussed in the review by Swarbrick and Partridge in this issue [64], the practicalities of delivering this were highly challenging. Some of the key elements such as visual cues, spatial orientation, a calm environment and presence of family and friends may be limited due to staff shortages and restricted environments but units should still be striving to achieve these non-pharmacological measures.

Screening, assessment, identification and accurate diagnosis remain essential to the management of delirium. The avoidance of ‘easy to fix’ triggers such as dehydration, constipation and electrolyte imbalances apply. Ongoing close monitoring is indicated for individuals who show initial signs of subsyndromal disease, for example the hypoactive motor phenotype. The avoidance of secondary brain injury by early initiation of active management aimed at preventing or treating organ failure, restoring metabolic disturbances and hypoxia, and consideration of artificial (including parenteral) nutrition is paramount. Fever should be managed to avoid hyperpyrexia. Complications such as seizures and non-convulsive status epilepticus should be identified promptly and managed with standard local and national protocols (e.g. [65]). Corticosteroids should be initiated where they are indicated in those with severe systemic COVID-19 [66]; dexamethasone may precipitate [67] or paradoxically improve delirium [68, 69].
Pharmacological measures may be indicated as per local protocols and will include the use of haloperidol [70]. Severe hyperactive delirium was associated with specific risks in terms of infection control earlier in the pandemic, and caused difficulties with prolonged interventions such as facemasks, high-flow nasal oxygen and tracheostomies. Delirium also complicated acute dysphagia and tracheostomy decannulation [71]. Clinicians should not be afraid of high doses of antipsychotics for severe intractable delirium at the end of life [20]. Enrolment in research studies and/or auditing of time course and management should be used to aid improved outcomes [72]. Opinions should be sought from local palliative care teams in situations of difficult to manage agitation. Early treatment escalation decisions should be made, with engagement from individuals and next of kin strongly encouraged.
Delirium in COVID-19 potentially provides a unique opportunity and framework to conduct mechanistic studies and clinical trials, to explore novel treatment options and gain a greater understanding of the pathophysiology of this condition. The focus of early trials was inevitably on major endpoints such as death and hospitalisation, with a lack of data on delirium, but it is hoped that future trials will be able to explore this.
Patient, relative and staff experiences: what is it actually like?
The experiences of COVID-19 delirium for patients, relatives and staff have played out on a national and international scale, with harrowing scenes seen on social and mainstream media at times. For those working in care homes, hospices and acute hospitals, these experiences may not be new, but due to the prevalence of delirium in COVID-19 these experiences have reached a wider audience.
The experiences of delirium are mostly negative for all involved, although a small number of studies have reported no effect (Fig. 2). A large 2011 study of non-COVID-19 delirium used 24 reports based on delirium experience accounts from 483 patients [73]. This included 1097 findings that were extracted and abstracted into 92 meta-findings. The authors grouped these into five emerging themes: ‘perception’; ‘emotions’; ‘interaction with others’; ‘dealing with delirium’; and ‘influence on further life’. Perception was found to be the most common theme across all included articles, and it was also the category with the highest number of meta-findings (n = 32). Emotions was the second biggest category (n = 15). A wide range of positive and negative emotions were reported by the patients interviewed. The most common emotions were fear and anxiety. Other negative emotions reported were feeling threatened; insecurity; panic; anger; restlessness: powerlessness; helplessness; frustration; loneliness and hopelessness; vulnerability; guilt; shame; ‘going mad’; and emotional chaos [73]. Similar themes are observed in the ICU [74] and after transcatheter or surgical aortic valve replacement [75].
Namba et al. performed content analysis of interviews with 20 bereaved family members of cancer patients who developed delirium during the last 2 weeks before death [76]. Many families were distressed by delirium and felt it reflected pain, mental distress or anxiety about death in their loved ones; others felt the delirium was a relief from real suffering. The same group re-examined this issue in a multicentre questionnaire survey of 560 bereaved family members of cancer patients who had had delirium before death. More than two-thirds of the 242 respondents perceived all delirium symptoms (other than somnolence) as ‘distressing’ or ‘very distressing’. Many felt there were deficits in the medical care provided with the need for more explanation about delirium and for medical staff to be less distant and to show greater respect for the patient's subjective world [77].
Breitbart et al. [78] reported that 76% of caregivers (usually a spouse) and 73% of nurses reported severe distress related to delirium. Mean distress levels were significantly higher for caregivers than for nurses caring for patients and even for the patients themselves. Hyperactive delirium and poor functional status were major predictors of family distress, while severe delirium and perceptual disturbances were the strongest predictors of nursing distress. Lou and Dai emphasise the increased work-load on staff and the conflicts between providing adequate care for delirious patients while not neglecting other patients, and between comforting patients while also maintaining their safety [79]. This is something that will have been compounded by the staff shortages seen during the COVID-19 pandemic.
In the context of COVID-19 delirium, there is limited published research about lived experiences. What we can do is look at the experiences of individuals from the SARS epidemic from 2003. A study looking at perceived stress scores-10 (PSS-10) in SARS survivors demonstrated that they had higher stress levels during the outbreak, compared with control subjects (PSS-10 scores 19.8 vs. 17.9/40, p < 0.01), and this persisted one year later (PSS-10 scores 19.9 vs. 17.3, p < 0.01) without signs of decrease. On re-survey a year later, SARS survivors also showed worrying levels of depression, anxiety and post-traumatic symptoms. An alarming proportion (64%) scored above the survey cut-off that suggests psychiatric morbidity. During the outbreak, healthcare worker SARS survivors had stress levels similar to those of non-healthcare workers, but healthcare workers showed significantly higher stress levels a year later (healthcare workers PSS-10 score = 22.8, compared with non-healthcare workers PSS-10 score = 18.4; p < 0.05) and had higher depression, anxiety and post-traumatic symptoms [80].
The experiences of COVID-19-associated delirium are only now coming to light but, based on experiences from non-COVID-19 delirium and the SARS epidemic, the effects on patients, relatives and staff will be negative and will, for some, undoubtedly require psychosocial intervention.
Is COVID-19-associated delirium different to normal and if so what can we do?
COVID-19-associated delirium shares phenomenology and pathophysiological pathways with classical delirium; however, evidence suggests that the prevalence is greater and the outcomes are worse. The prevalence of COVID-19 delirium in acute hospital admissions approaches 40% and this rises to 80% in the ICU population [81]. While the management remains the same, non-pharmacological management has been difficult to achieve in stressed situations with limited staff and restricted visiting. Anecdotal clinician experience is that we have been using more pharmacological measures to manage COVID-19 delirium and this reflects the severity of the distress and agitation, particularly at the very end of life. These factors mean there is a high risk of moral injury for staff and relatives. Visiting for people with COVID-19 delirium should be considered beneficial, while balancing the local infection risk. Given the situational stressors of the pandemic, the authors suggest the following practical approach: COVID-ST (Box 1).
Box 1. Practical approaches to delirium associated with COVID-19.
- Complications: Complications such as seizures and non-convulsive status epilepticus should be identified promptly and managed with standard protocols. Robust screening and identification of other potential central nervous system complications such as posterior reversible encephalopathy syndrome, encephalitis and stroke.
- Organ damage: Avoidance of secondary brain injury by earliest possible initiation of intensive care management aimed at preventing or treating organ failure, restoring metabolic disturbances and hypoxia.
- Visiting: Visiting where possible and support for visiting families.
- Imbalances: Avoidance of precipitants such as dehydration, constipation and electrolyte imbalances.
- Diet: Early consideration of artificial nutrition to supplement reduced enteral intake is paramount.
- Screening: Early screening, assessment and identification – with input from specialist teams including experienced geriatricians.
- Treatment escalation: Early discussion of resuscitation and treatment escalation plans.
While the pandemic continues, COVID-19 delirium offers the potential to look for more therapeutics, engage older adults in research activity and aid the understanding of this complex condition of ‘acute brain dysfunction’ that requires investment and further funding. Future COVID waves with younger people and the obstetric population will bring new challenges to managing delirium. And there remains a cohort of clinically vulnerable people; those on heavy immunosuppression and those unable or unwilling to have the vaccine that will be even more challenging.
Conclusions
COVID-19-associated delirium is different to classical delirium. It is more prevalent, longer lasting and associated with worse outcomes. The management remains the same, with the exception of distressing end of life agitation where the need for higher-than-normal doses of sedatives may be required. It poses a unique opportunity to discover more about this complex condition. For an up-to-date summary of the recent research studies, we encourage readers to go to the Neurology and Neuropsychiatry of COVID-19 blog [82].
Acknowledgements
No competing interests declared.