Delirium severity in critical patients with COVID-19 from an Infectious Disease Intensive Care Unit.

BACKGROUND: COVID-19 is mainly characterized by respiratory manifestations. Nevertheless, neurologic complications have been described, including delirium, which appears to be frequent, prolonged, and severe.
METHODS: We conducted a retrospective analysis of demographic, clinical, and laboratory data of two cohorts: patients with COVID-19 admitted to the infectious disease intensive care unit (ID-ICU) and patients admitted to the ID-ICU with other respiratory infections in 2018-2019. Outcomes were defined as the presence, duration, and severity of delirium. Doses of antipsychotics used to control delirium were converted to equivalents and used as delirium severity. Logistics regression models were used to correlate COVID-19 with the outcomes.
RESULTS: Ninety-nine patients with COVID-19 and 40 patients without COVID-19 were included. The mean age of the COVID-19 cohort was 63 years, with a male predominance. Delirium developed in 42%, with a median duration of 3 days and an equivalent dose of olanzapine use of 10 mg/day. In univariate analysis, COVID-19 was not associated with the development or different duration of delirium when compared with patients without COVID-19. There was an association between COVID-19 and severity of delirium in a binary logistic regression model controlled to confounding variables.
CONCLUSION: COVID-19 is not associated with a higher prevalence or duration of delirium than in cohorts without COVID-19. However, it is associated with more severe forms of delirium.

Introduction

SARS-CoV-2 causes potentially serious respiratory, gastrointestinal, neurologic and/or liver disorders. The prevalence of neurologic complications associated with SARS-CoV-2 infection varies according to the different studies, with the most common manifestations being headache, anosmia, and ageusia (Andreia Costa, 2020). Other complications include stroke, epilepsy, neuromuscular symptoms, and delirium (Mao et al., 2020; Zubair et al., 2020).

The mechanism by which the SARS-CoV2 virus produces nervous system symptoms is still unknown. However, several mechanisms have been proposed, including transsynaptic transfer between infected neurons, entry by the olfactory nerve, infection of the vascular endothelium, or migration of infected leukocytes through the blood-brain barrier (Zubair et al., 2020). More recently, another hypothesis postulates indirect damage by SARS-CoV2 to the central nervous system because of low oxygen levels, coagulopathy, exposure to sedative and analgesic drugs, isolation, and immobility (Pun et al., 2021).

It is estimated that about 20-40% of patients admitted to the intensive care unit (ICU) develop delirium and the incidence may be higher in mechanically ventilated patients. In addition, the presence of delirium and its duration and severity are risk factors for long-term cognitive sequelae in patients who survive critical illnesses (Salluh & Latronico, 2019). Delirium, reported in up to 84% of patients with COVID-19, can be associated with several mechanisms, including direct infection and parenchymal injury, toxic-metabolic encephalopathy, epilepsy, or immune-mediated injury. However, cases of meningoencephalitis caused by SARS-CoV-2 are rare and more investigation is needed to clarify direct virus injury to the central nervous system (CNS) (Espíndola et al., 2020, Helms et al., 2020; Khan et al., 2020; Mao et al., 2020; Pun et al., 2021; Zubair et al., 2020).

We formulated the hypothesis that delirium in patients with COVID-19 is frequent, severe, and may last over time. In addition, there is a lack of knowledge of the possible long-term effects of SARS-CoV2 on the CNS.

Thus, well-constructed, comparative studies are needed to understand the real extent and prognosis of delirium in patients with COVID-19.

Objectives

We aim to evaluate the prevalence of delirium, its characteristics, namely duration and severity, in a cohort of critically ill patients with COVID-19 compared with a cohort of critically ill patients admitted because of other respiratory infections.

Methods

Patient Selection

We performed a cohort study including two cohorts of patients with COVID-19 and patients without COVID-19 to compare similar patients who underwent similar practices in the same unit. We retrospectively analyzed all consecutive patients with respiratory symptoms and laboratory-confirmed SARS-CoV-2 infection admitted to the infectious disease intensive care unit (ID-ICU) of our tertiary center between March 2020 and December 2020 (patients with COVID-19). We also reviewed all consecutive patients with clinically confirmed respiratory infection admitted to the ID-ICU between January 2018 and December 2019 (patients without COVID-19).

Patients were excluded if moribund (died in the first 24 hours after admission), those still under deep sedation when life-supporting measures were withdrawn, and patients with no Acute Physiology and Chronic Health Evaluation (APACHE) score calculated. The study protocol was approved by the Local Ethics Committee.

Parameter Acquisition

We assessed electronic medical records and retrieved demographic and clinical variables such as sex, age, comorbidities including vascular risk factors, previous neuropsychiatric diseases that could impact the development of delirium, including cognitive impairment, impairment of sensory inputs, namely visual, auditory, olfactory, gustatory or tactile, previous cerebrovascular disease or demyelinating disease, presence of epilepsy, and previously diagnosed psychiatric illness and chronic medication including previous use of benzodiazepines and antipsychotics. The APACHE II score, Simplified Acute Physiology Score (SAPS) II and III Score, and organic dysfunction at admission according to Marshall and Meakins criteria were retrieved, as it could influence the delirium characteristics. We also reviewed the length of stay in the ID-ICU, need and duration of endotracheal intubation (ETI), the usage of the prone position, and development of bacterial superinfection, as the length of stay and these factors associated with increased length of stay could impact the development or duration of delirium.

Antipsychotics and benzodiazepines were registered and maximum daily dosages of antipsychotics used were converted to dose equivalents of olanzapine according to the International Consensus Study of Antipsychotic Dosing (Gardner, Murphy, O'Donnell, Centorrino, & Baldessarini, 2010).

Outcome Measures

Outcome measures were defined as the presence, duration, and severity of delirium. Development of delirium was assessed by revising neurologic status descriptions in electronic medical records. Although there are currently several bedside assessment scales validated for the assessment of delirium, namely confusion assessment method for the intensive care unit (CAM-ICU), its implementation in our unit is still underway. Because of the lack of CAM-ICU documentation on several patients from both cohorts, the presence of delirium was assessed only by recorded accurate daily descriptions of patients’ mental status. Duration of delirium was defined by every day that the neurologic status recorded on patients’ charts was compatible with hypoactive or hyperactive delirium, namely impaired consciousness, disorientation, confusion, or agitation. Maximum daily dosages of antipsychotics used were converted to dose equivalents of olanzapine according to the International Consensus Study of Antipsychotic Dosing (Gardner et al., 2010) and used as a delirium severity marker. For accurate statistical analysis, dose equivalents of olanzapine were also dichotomized into a nominal variable with a cut-off value of 10 mg/day.

Statistical Analysis

We compared the baseline categorical and continuous independent variables between cohorts and outcomes (developed delirium or not) with chi-square test and Mann-Whitney test, according to variables characteristics and distribution. Possible confounders with a p < 0.1 in univariable analysis and factors with plausibility to influence results were included in the multivariable model. We performed backward elimination to identify independent predictors of the outcome. The effect of COVID-19 infection on defined outcomes was accessed using a binary logistics regression model, generating adjusted odds ratios (aOR) and 95% confidence intervals (CI). All statistical analyses were performed using IBM SPSS Statistics® Version 26.0 (Armonk, New York. IBM Corp). Significance was set at p < 0.05.

Results

From a total of 177 potential patients (128 patients with COVID-19 and 49 patients without COVID-19), we excluded 38 patients because of the presence of at least one exclusion criteria (Figure 1 ).

Figure 1

Decision tree for included patients

*Moribund (died in the first 24 hours after admission), still under deep sedation when life-supporting measures were withdrawn or with no Acute Physiology and Chronic Health Evaluation (APACHE) score calculated.

We included 139 patients (99 in the COVID-19 cohort and 40 in the non-COVID-19 cohort) in the final analysis. During the COVID-19 pandemic, our 6-bed intensive care unit was upgraded into a 10-bed intensive care unit to fully respond to the needs, which explains the disparity in the total number of patients included in both cohorts.

The respiratory infectious agents identified in the 139 patients are listed in Table 1 .

Table 1

Infectious respiratory agent

Infectious Agent	Frequency N=139 (100)
SARS-CoV-2	99 (71.2)
Pneumocystis jirovecii	1 (0.7)
Legionella	2 (1.4)
Streptococcus pneumoniae	8 (5.8)
Influenza A	7 (5.0)
Influenza B	1 (0.7)
Staphylococcus aureus	1 (0.7)
Cytomegalovirus	2 (1.4)
Leishmania	2 (1.4)
Not Identified	16 (11.5)

SARS-CoV-2: Severe Acute Respiratory Syndrome Coronavirus 2

Both cohorts had a few different characteristics (Table 2 ). Patients with COVID-19 were older, had less severe dysfunctions according to the defined severity admission scores (SAPS II, SAPS III, and APACHE), higher prevalence of vascular risk factors (hypertension, diabetes, dyslipidemia) although lower rates of smoking/ex-smoking habits and fewer organ dysfunctions at admission (renal, cardiovascular, hematologic, and neurologic).

Table 2

Baseline characteristics of all patients and comparison between cohorts

Variables	COVID-19 Cohort (n=99)	Non-COVID-19 Cohort (n=40)	P-value
Age, in years – Mean (SD)	63 (18)	57 (22)	0.020*
Male – n (%)	60 (60.6)	21 (52.5)	0.380
Comorbidities – n (%)	90 (90.9)	34 (85.0)	0.309
Hypertension – n (%)	57 (57.6)	15 (37.5)	0.032*
Diabetes Mellitus – n (%)	36 (36.4)	7 (17.5)	0.029*
Dyslipidemia – n (%)	50 (51.5)	12 (30.0)	0.021*
Chronic Cardiac Disease – n (%)	19 (19.2)	4 (10.0)	0.187
Smoker/Ex-Smoker – n (%)	14 (14.1)	17 (42.5)	0.001⁎⁎
Alcohol Abuse – n (%)	6 (6.1)	6 (15.0)	0.089
Asthma – n (%)	2 (2.0)	1 (2.5)	0.860
COPD – n (%)	6 (6.1)	6 (15.0)	0.089
OSAS – n (%)	7 (7.1)	0 (0)	0.084
Liver Chronic Disease – n (%)	7 (7.1)	6 (15.0)	0.146
Renal Chronic Disease – n (%)	7 (7.2)	6 (15.0)	0.146
Chronic Hematogenous Disease – n (%)	3 (3.0)	1 (2.5)	0.866
Previous Neurologic Disease – n (%)	12 (12.1)	6 (15.0)	0.647
Active Neoplasia – n (%)	8 (8.1)	4 (10.0)	0.715
HIV Infection – n (%)	1 (1.0)	6 (15.0)	0.001⁎⁎
Previous Transplantation – n (%)	4 (4.0)	2 (5.0)	0.801
Auto-immune Disease – n (%)	6 (6.1)	3 (7.5)	0.766
Previous antipsychotic use – n (%)	2 (2.0)	2 (5.0)	0.341
Previous Benzodiazepines use – n (%)	26 (26.3)	9 (22.5)	0.644
APACHE Score – Mean (SD)	16 (5)	22 (8)	<0.001⁎⁎
SAPSII Score – Mean (SD)	31 (14)	46 (17)	<0.001⁎⁎
SAPSIII Score – Mean (SD)	51 (14)	66 (14)	<0.001⁎⁎
Respiratory Dysfunction – n (%)	90 (90.1)	38 (95.0)	0.419
Renal Dysfunction – n (%)	15 (15.2)	15 (37.5)	0.004⁎⁎
Cardiovascular Dysfunction – n (%)	23 (23.2)	21 (52.5)	0.001⁎⁎
Hematologic Dysfunction – n (%)	13 (13.1)	13 (32.5)	0.008⁎⁎
Neurologic Dysfunction – n (%)	10 (10.1)	15 (37.5)	<0.001⁎⁎
Liver Dysfunction – n (%)	15 (15.2)	8 (20.0)	0.486
Length of stay IDICU, in days – median (IQR)	8 (16)	12 (23)	0.019⁎⁎
Need of ETI – n (%)	42 (42.4)	26 (65.0)	0.016⁎⁎
Length of ETI use, in days – median (IQR)	13 (12)	12 (16)	0.749
Prone position used – n (%)	42 (42.4)	9 (22.5)	0.014⁎⁎
Bacterial superinfection – n (%)	34 (34.3)	10 (25.0)	0.284
Development of Delirium – n (%)	42 (42.4)	23 (58)	0.107

APACHE - Acute Physiology and Chronic Health Evaluation (APACHE) II score; COPD = chronic obstructive pulmonary disease; ETI = endotracheal intubation; HIV = human immunodeficiency virus; ID-ICU = infectious diseases intensive care unit; OSAS = obstructive sleep apnea syndrome; SAPS II = simplified acute physiology score.

p-value<0.05

p-value<0.005

Comorbidities were present in 90.9% of patients with COVID-19, of which 12% had previous neuropsychiatric disease, including five mild cognitive impairment, two previous stroke, one epilepsy, one central demyelinating disease, one diabetic neuropathy, one toxic necrotizing myopathy, and one maniac depressive disorder. No patient had severe visual impairment. Only 2% and 26.3% of patients with COVID-19 previously took antipsychotics and benzodiazepines, respectively.

Both cohorts presented similar frequency in the development of delirium: 42% in patients with COVID-19 versus 58% in patients without COVID-19. Magnetic resonance imaging (MRI) or electroencephalography (EEG) was performed in 9 (5 patients with COVID-19) and 10 patients (5 patients with COVID-19), respectively. 1 EEG was considered normal (patient without COVID-19), and the other 9 showed diffuse slowing compatible with unspecified encephalopathy. Only 1 MRI of the 5 patients with COVID-19 showed a concomitant cerebral venous thrombosis, whereas the others lacked any concomitant cerebral insult. The duration of delirium was similar (3 to 4 days) (Table 3 ). The presence of COVID-19, previous neurologic disease, antipsychotic use, or benzodiazepine use were not associated with the development of delirium. However, there was a higher prevalence of male patients, autoimmune disease, higher severity scores at admission, and cardiovascular, hematologic, and neurologic organic dysfunctions in the group that developed delirium. Patients that developed delirium also had longer ID-ICU stays, a higher need for endotracheal intubation (ETI), and higher rates of bacterial superinfection. Patients with COVID-19 required a higher dose of antipsychotics to control delirium, with a median of 10.2 mg/day versus 0.68 mg/day (Table 4 ).

Table 3

Baseline characteristics of patients with and without delirium

	Development of Delirium
Variables	Yes (N=65)	No (N=74)	P-value
Age, in years – Mean (SD)	63 (16)	61 (15)	0.154
Male – n (%)	44 (67.7)	37 (50.0)	0.035*
Comorbidities – n (%)	59 (90.8)	65 (87.8)	0.578
Hypertension – n (%)	39 (60.0)	33 (44.6)	0.070
Diabetes Mellitus – n (%)	22 (33.8)	21 (28.4)	0.487
Dyslipidemia – n (%)	32 (49.2)	31 (41.9)	0.386
Chronic Cardiac Disease – n (%)	14 (21.5)	9 (12.2)	0.138
Smoker/Ex-Smoker – n (%)	18 (27.7)	13 (17.6)	0.152
Alcohol Abuse – n (%)	8 (12.3)	4 (5.4)	0.148
Asthma – n (%)	1 (1.5)	2 (2.7)	0.637
COPD – n (%)	7 (10.8)	5 (6.8)	0.401
OSAS – n (%)	5 (7.7)	2 (2.7)	0.180
Liver Chronic Disease – n (%)	8 (12.3)	5 (6.8)	0.262
Renal Chronic Disease – n (%)	7 (10.8)	6 (8.1)	0.591
Chronic Hematogenous Disease – n (%)	2 (3.1)	2 (2.7)	0.895
Previous Neurologic Disease – n (%)	11 (16.9)	7 (9.5)	0.191
Active Neoplasia – n (%)	4 (6.2)	8 (10.8)	0.329
HIV Infection – n (%)	2 (3.1)	5 (6.8)	0.322
Previous Transplantation – n (%)	3 (4.6)	3 (4.1)	0.871
Auto-immune Disease – n (%)	1 (1.5)	8 (10.8)	0.028*
Previous antipsychotic use – n (%)	3 (4.6)	1 (1.4)	0.251
Previous Benzodiazepines use – n (%)	19 (29.2)	16 (21.6)	0.302
COVID-19 infection – n (%)	42 (64.6)	57 (77.0)	0.107
APACHE Score – Mean (SD)	20 (6)	16 (6)	0.001⁎⁎
SAPSII Score – Mean (SD)	40 (16)	32 (16)	0.003⁎⁎
SAPSIII Score – Mean (SD)	60 (15)	51 (15)	0.001⁎⁎
Respiratory Dysfunction – n (%)	61 (93.8)	67 (90.5)	0.471
Renal Dysfunction – n (%)	18 (27.7)	12 (16.2)	0.101
Cardiovascular Dysfunction – n (%)	33 (50.8)	11 (14.9)	<0.001⁎⁎
Hematologic Dysfunction – n (%)	17 (26.2)	9 (12.2)	0.035*
Neurologic Dysfunction – n (%)	20 (30.8)	5 (6.8)	<0.001⁎⁎
Liver Dysfunction – n (%)	12 (18.5)	11 (14.9)	0.569
Length of stay ID-ICU – median (IQR)	19 (19)	6 (6)	<0.001⁎⁎
Need of ETI – n (%)	53 (81.5)	15 (20.3)	<0.001⁎⁎
Length of ETI use – median (IQR)	16 (13)	8 (14)	0.197
Prone position used – n (%)	26 (40.0)	25 (33.8)	0.426
Bacterial superinfection – n (%)	31 (47.7)	13 (17.6)	<0.001⁎⁎

APACHE = Acute Physiology and Chronic Health Evaluation (APACHE) II score; COPD = chronic obstructive pulmonary disease; ETI = endotracheal intubation; HIV = human immunodeficiency virus; ID-ICU = infectious diseases intensive care unit; OSAS = obstructive sleep apnea syndrome; SAPS II= simplified acute physiology score.

p-value<0.05

p-value<0.005

Table 4

Comparison between patients with COVID-19 and patients without COVID-19 with delirium

	Patients with delirium (n=65)
Variables	COVID-19 Cohort (n=42)	Non-COVID-19 Cohort (n=23)	P-value
Age, in years – Mean (SD)	67 (14)	57 (16)	0.018*
Male – n (%)	30 (71.4)	14 (60.9)	0.384
Comorbidities – n (%)	40 (95.2)	19 (82.6)	0.093
Hypertension – n (%)	29 (69.0)	10 (43.5)	0.044*
Diabetes Mellitus – n (%)	17 (40.5)	5 (21.7)	0.127
Dyslipidemia – n (%)	25 (59.5)	7 (30.4)	0.025*
Chronic Cardiac Disease – n (%)	11 (26.2)	3 (13.0)	0.218
Smoker/Ex-Smoker – n (%)	8 (19.0)	10 (43.5)	0.035*
Alcohol Abuse – n (%)	3 (7.1)	5 (21.7)	0.087
Asthma – n (%)	1 (2.4)	0 (0)	0.456
COPD – n (%)	3 (7.1)	4 (17.4)	0.202
OSAS – n (%)	5 (11.9)	0 (0)	0.085
Liver Chronic Disease – n (%)	5 (11.9)	3 (13.0)	0.894
Renal Chronic Disease – n (%)	2 (4.8)	5 (21.7)	0.035*
Chronic Hematogenous Disease – n (%)	2 (4.8)	0 (0)	0.288
Previous Neurologic Disease – n (%)	5 (11.9)	6 (26.1)	0.145
Active Neoplasia – n (%)	3 (7.1)	1 (4.3)	0.654
HIV Infection – n (%)	0 (0)	2 (8.7)	0.052
Previous Transplantation – n (%)	1 (2.4)	2 (8.7)	0.246
Auto-immune Disease – n (%)	1 (2.4)	1 (4.3)	0.178
Previous antipsychotic use – n (%)	1 (2.4)	2 (8.7)	0.246
Previous Benzodiazepines use – n (%)	13 (31.0)	6 (26.1)	0.680
APACHE Score – Mean (SD)	18 (5)	24 (6)	<0.001⁎⁎
SAPSII Score – Mean (SD)	34 (14)	50 (14)	<0.001⁎⁎
SAPSIII Score – Mean (SD)	54 (15)	70 (11)	<0.001⁎⁎
Respiratory Dysfunction – n (%)	39 (92.9)	22 (95.7)	0.654
Renal Dysfunction – n (%)	9 (21.4)	9 (39.1)	0.127
Cardiovascular Dysfunction – n (%)	17 (40.5)	16 (69.6)	0.025*
Hematologic Dysfunction – n (%)	9 (21.4)	8 (34.8)	0.241
Neurologic Dysfunction – n (%)	9 (21.4)	11 (47.8)	0.027*
Liver Dysfunction – n (%)	8 (19.0)	4 (17.4)	0.869
Length of stay ICUID – median (IQR)	18 (19)	20 (24)	0.269
Need of ETI – n (%)	32 (76.2)	21 (91.3)	0.133
Length of ETI use – median (IQR)	13 (11)	12 (16)	0.815
Prone position used – n (%)	20 (47.6)	6 (26.1)	0.052
Bacterial superinfection – n (%)	23 (54.8)	8 (34.8)	0.123
Delirium Duration – median (IQR)	3 (5)	4 (2)	0.702
Antipsychotic equivalent dose, in mg/day – median (IQR)	10.3 (17)	0.68 (10)	0.034*

APACHE = Acute Physiology and Chronic Health Evaluation (APACHE) II score; COPD = chronic obstructive pulmonary disease; ETI = endotracheal intubation; HIV = human immunodeficiency virus; IDICU = infectious diseases intensive care unit; OSAS = obstructive sleep apnea syndrome; SAPS II = simplified acute physiology score.

p-value<0.05

p-value<0.005

In univariate binary logistics regression analysis, male sex, higher APACHE score, cardiovascular and neurologic dysfunction at admission, length of stay in ID-ICU, need for ETI, and bacterial superinfection were the only risk factors for the development of delirium. When adjusted for confounding variables, only male sex (aOR 2.64, 95% CI 1.02-6.82, p=0.046) and need of ETI (aOR 10.3, 95% CI 3.69-29.0, p<0.001) presented as independent risk factors to the development of delirium (Table 5 ).

Table 5

Prediction of development of delirium by logistic regression analysis

	Development of Delirium		Univariate Analysis	Multivariate Analysis¶
Variables	Yes (N=65)	No (N=74)	OR (CI)	aOR (CI)
Age, in years – Mean (SD)	63 (16)	61 (15)	1.02 (0.99-1.04)	1.02 (0.99-1.06)
Male – n (%)	44 (67.7)	37 (50.0)	2.10 (1.05-4.18)*	2.64 (1.02-6.82)*
APACHE Score – Mean (SD)	20 (6)	16 (6)	1.10 (1.03-1.16)**	0.99 (0.92-1.06)
COVID-19 infection – n (%)	42 (64.6)	57 (77.0)	0.55 (0.26-1.15)	1.04 (0.35-3.14)
Cardiovascular Dysfunction – n (%)	33 (50.8)	11 (14.9)	5.91 (2.64-13.2)**	1.11 (0.35-3.49)
Neurologic Dysfunction – n (%)	20 (30.8)	5 (6.8)	6.13 (2.15-17.5)**	3.65 (0.89-14.9)
Length of stay ID-ICU – median (IQR)	19 (19)	6 (6)	1.10 (1.06-1.15)**	1.04 (1.00-1.09)
Need of ETI – n (%)	53 (81.5)	15 (20.3)	17.4 (7.46-40.4)**	10.3 (3.69-29.0)**
Bacterial superinfection – n (%)	31 (47.7)	13 (17.6)	4.28 (1.98-9.26)**	1.22 (0.39-3.82)

aOR = adjusted odds ratio; APACHE = Acute Physiology and Chronic Health Evaluation (APACHE) II score; CI = confidence interval; ETI = endotracheal intubation; ID-ICU = Infectious diseases intensive care unit; OR= odds ratio.

¶Adjusted for age, sex, length of ID-ICU stay, need of ETI, neurologic dysfunction.

*p-Value <0.05; **p-Value <0.005.

In univariate binary logistic regression analysis, the presence of COVID-19, was the only risk factor for the use of dosages of antipsychotics above 10 mg/day used (p<0.05), however when adjusted for confounding variables we identified COVID-19 (aOR = 19.7, 95% CI 2.69-144.7, p<0.005) as well as length of stay in ID-ICU (aOR = 1.05, 95% CI 1.00-1.09, p<0.05) and male sex (aOR = 4.71, 95% CI 1.17-19.0, p<0.05) as independent risk factors for delirium severity (Table 6 ).

Table 6

Prediction of Antipsychotic dose equivalent in patients who developed delirium, by logistic regression analysis

	Antipsychotic dose equivalent > 10mg/day in patients who developed delirium		Univariate Analysis	Multivariate Analysis¶
Variables	Yes (N=25)	No (N=40)	OR (CI)	aOR (CI)
Age, in years – Mean (SD)	64 (16)	63 (16)	1.00 (0.97-1.04)	1.02 (0.98-1.06)
Male – n (%)	20 (80)	25 (60)	2.67 (0.83-8.56)	4.71 (1.17-19.0)*
APACHE Score – Mean (SD)	19 (5)	21 (7)	0.94 (0.86-1.02)	0.90 (0.81-1.00)*
COVID-19 infection – n (%)	21 (84)	21 (52.5)	4.75 (1.38-16.4)*	19.7 (2.69-144.7)**
Cardiovascular Dysfunction – n (%)	13 (52)	20 (50)	1.08 (0.40-2.94)	0.75 (0.23-2.45)
Neurologic Dysfunction – n (%)	10 (40)	10 (25)	2.00 (0.68-5.85)	1.64 (0.52-5.16)
Length of stay ID-ICU – median (IQR)	22 (17)	15 (21)	1.02 (0.99-1.06)	1.05 (1.00-1.09)*
Need of ETI – n (%)	21 (84)	32 (80)	1.31 (0.35-4.92)	0.71 (0.16-3.19)
Bacterial superinfection – n (%)	15 (60)	16 (40)	2.25 (0.81-6.24)	1.89 (0.53-6.80)

aOR = adjusted odds ratio; APACHE = Acute Physiology and Chronic Health Evaluation (APACHE) II score; CI = confidence interval; ETI = endotracheal intubation; IDICU = infectious diseases intensive care unit; OR = odds ratio.

¶Adjusted for age, sex, length of ID-ICU stay, need of ETI, neurologic dysfunction.

*p-Value <0.05; **p-Value <0.005.

Discussion

We did not find that the presence of COVID-19 increases the risk of developing delirium or its duration. Nevertheless, the development of delirium in critically ill patients with COVID-19 appears to be associated with much more severe manifestations with the need for higher dosages of antipsychotics to control the symptoms.

Acquired knowledge of SARS-CoV-2 infection has shown that neurologic symptoms could range from mild nonspecific or specific symptoms such as the loss of various sensory perceptions or more severe involvement. The mechanism of encephalopathy in COVID patients remains to be determined (Jarrahi et al., 2020).

Delirium is extremely common in the intensive care unit (ICU), especially among mechanically ventilated patients. Different mechanisms have been proposed to explain delirium, such as medication sedatives, namely benzodiazepines, opiates, sepsis, respiratory disease, older age, previous alcohol abuse, previous psychiatry medication, or underlying central nervous system disease (Cavallazzi, Saad, & Marik, 2012).

In a cohort of 58 ICU patients with COVID-19, Helms et al. reported that 84% developed neurologic symptoms, of which 69% developed delirium, and later, a cohort of 150 ICU patients reported delirium and/or altered neurologic exam in 84% of patients, of which 18% showed signs of delirium at ICU admission (Helms et al., 2020, Helms et al., 2020).

The largest cohort study performed, with more than 2000 patients with COVID-19, reported the presence of acute brain dysfunction (coma or delirium) to be more common and more prolonged than in other studies of acute respiratory failure without COVID-19; the prevalence of delirium was reported to be around 54% with a mean duration of 3 days when addressed through validated tools (Pun et al., 2021).

Our study, although retrospective, shows a slightly lower prevalence of delirium (42%) with a duration of 3 days, consistent with other multicenter COVID-19 studies (Helms et al., 2020b; Pun et al., 2021), improving the validity of our results. Although, our study disclosed that delirium in patients with COVID-19 is not more common or prolonged when compared with other respiratory failure patients admitted to the same unit with similar management protocols. The lower prevalence of delirium in patients with COVID-19 could be because delirium screening was not performed systemically in our unit. In addition, because of the prolonged data collection period of approximately nine months, management practices for COVID-19 and its associated complications were improved according to the newest evidence, which could have induced some variability in the analysis.

Although there was a similar prevalence of delirium in patients with COVID-19 than in patients without COVID-19, the latter presented with significantly higher severity scores and organ dysfunction, known risk factors for the development of complications associated with prolonged ID-ICU stay and delirium (Marra, Ely, Pandharipande, & Patel, 2017; Pun et al., 2019).

Patients with more severe disease and needing a higher level of care appear to have a higher incidence of neurologic complications (Mao et al., 2020), including delirium, encephalopathy, and signs of first neuron injury (Zubair et al., 2020). Nevertheless, it is still unknown if any of these complications are specific to COVID-19 and not just complications related to concomitant morbidities.

A recent study demonstrated that mechanical ventilation, use of restraints, benzodiazepine, opioid and vasopressor infusions, and antipsychotics were associated with a higher risk of delirium on the day after in patients with COVID-19 (Pun et al., 2021).

In addition to the risk factors for delirium in the ICU and neuropathogenesis of SARS-CoV-2, the pandemic also created circumstances that further increase the risk of delirium, such as isolation of patients, absence of family visitors and the inability to freely ambulate (Kotfis et al., 2020).

We also confirmed that male sex and need of ETI are independent predictors of the development of delirium when adjusted for age, sex, length of ID-ICU stay, need of ETI, and neurologic dysfunction at admission. In accordance with a recent paper demonstrating female sex to be associated with more delirium-free and coma-free days, and invasive mechanical ventilation to be associated with a higher risk of delirium the next day (Pun et al., 2021). Furthermore, COVID-19 mostly affects the male sex, possibly because of the higher number of ACE2-expressing cells in their lungs (Helms et al., 2020b).

COVID-19 infection, lengthier ID-ICU stay, and male sex at admission were independent risk factors for severe delirium in the ICU when adjusted for age, sex, length of ID-ICU stay, need of ETI, and neurologic dysfunction at admission, consistent with recent reports showing severe delirium in patients with COVID-19 (Khan et al., 2020)

Because of the retrospective design of our study, categorization of delirium type was not performed, and higher doses of antipsychotics used might suggest a higher incidence of hyperactive delirium. Published reports demonstrate different rates of delirium types in patients with COVID-19, either the similar incidence of hypoactive and hyperactive or higher incidence of hypoactive delirium (Khan et al., 2020; Pun et al., 2021).

This study has several limitations. This is a single-center study that decreases the external validity of our results. Because of the retrospective study design, assessment of delirium might be less sensitive. In addition, the two cohorts had few baseline differences, and these were taken into consideration as confounding factors in the logistic regression model. Because the non-COVID-19 cohort was from 2018-2019, the presence of COVID-19 meant not only the presence of infection but also all the measures implemented to “flatten the curve” including limited family visitors, worse nonverbal communication between healthcare professionals and the patient because of the use of individual protective equipment. The use of antipsychotic drugs as a surrogate marker of delirium severity may not show the full picture of the delirium. Although, the protocols for delirium management in the ID-UCI used were the same since 2018, demonstrating a similar approach in both cohorts. Because of the prolonged period of data collection (around 9 months), several practices in the routine care of patients with COVID-19 were changed, including postponing mechanical ventilation and routine use of video calls to family members that were implemented only after a few months of COVID-19 care.

To our knowledge, this is the first study to compare delirium in two cohorts of patients with COVID-19 and respiratory patients without COVID-19 treated in the same unit over different but close time periods.

By demonstrating the presence of more severe delirium associated with COVID-19, it becomes clear that there is a need for better interventions to minimize the development of delirium in these patients. Furthermore, the presence of severe hyperactive delirium in the context of a pandemic with over-crowded healthcare services could theoretically lead to the inter-hospital spread of the virus because of agitated, uncooperative patients before intubation. (Kotfis et al., 2020)

Despite the existence of the ABCDEF safety bundle (A-Assess, Prevent, and Manage Pain; B-Both Spontaneous Awakening Trials and Spontaneous Breathing Trials; C-Choice of analgesia and sedation; D-Delirium: Assess, Prevent, and Manage; E-Early mobility and Exercise; F-Family engagement and empowerment) to reduce the incidence of delirium and improve care of critically ill patients, further analysis of existing guidelines is somewhat harder to accomplish in the middle of a pandemic scenario, requiring adaptations and modifications. Family involvement was perhaps the most neglected component during the pandemic, harder to accomplish at first, but as the pandemic progressed, newer techniques including video calls, daily clinical updates from the medical team, and dedicated time to listen to family concerns and thoughts were a few solutions to keep the family involved in the caring of patients with COVID-19 (Kotfis et al., 2020).

With the prevention and correct management of delirium in patients with COVID-19, we are ensuring better recovery and reducing the risk of long-term neurologic complications of surviving patients.

Conflict of Interest Disclosure

Dr. Rafael Dias, Dr. João Paulo Caldas, Dr. André Silva-Pinto, Dr. Andreia Costa, Prof. Dr. António Sarmento and Prof. Dr. Lurdes Santos have nothing to disclose.