Delirium Severely Worsens Outcome in Patients with COVID-19—A Retrospective Cohort Study from Temporary Critical Care Hospitals

Delirium is a sign of deterioration of homeostasis and worse prognosis. The aim of this study was to investigate the frequency, risk factors and prognosis of delirium in patients with COVID-19 in a temporary acute setting hospital. A retrospective cohort analysis of data collected between October 2020 and February 2021 from two temporary acute care hospitals was performed. All consecutive hospitalized patients ≥18 years old with COVID-19 were included. An assessment of consciousness was carried out at least two times a day, including neurological examination. Delirium was identified through retrospective chart review according to DSM-5 criteria if present at least once during hospitalization. Analysis included 201 patients, 39 diagnosed with delirium (19.4%). Delirious patients were older (p < 0.001), frailer (p < 0.001) and the majority were male (p = 0.002). Respiratory parameters were worse in this group with higher oxygen flow (p = 0.013), lower PaO2 (p = 0.043) and higher FiO2 (p = 0.006). The mortality rate was significantly higher in patients with delirium (46.15% vs 3.70%, p < 0.001) with OR 17.212 (p < 0.001) corrected for age and gender. Delirious patients experienced significantly more complications: cardiovascular (OR 7.72, p < 0.001), pulmonary (OR 8.79, p < 0.001) or septic (OR 3.99, p = 0.029). The odds of mortality in patients with COVID-19 presenting with delirium at any point of hospitalization were seventeen times higher.


Introduction
Novel respiratory syndrome, known as COVID-19, was first reported at the end of December 2019 and soon became a global health threat [1]. With millions of people affected and suffering worldwide, the global burden for COVID-19 survivors is unprecedented.
The COVID-19 pandemic introduced new criteria for immediate planning in healthcare to accommodate a rapidly growing number of admissions to critical care wards during the pandemic's peak. Temporary hospitals were created to overcome the sudden surge of acutely ill and deteriorating patients into the intensive care units (ICUs) and toprovide professional medical help despite bed and resources shortages. The number of temporary acute care units introduced into the systems of all countries meant that the ability to care for many patients in a short time was higher, but on the other hand, also meant that the use of systematic patient assessments, using dedicated scales, might have been compromised [2].
Central nervous system (CNS) involvement during COVID-19 was identified early during the initial phase of the pandemic, therefore awareness of different aspects of the effects of SARS-CoV-2 virus on the central nervous system should be acknowledged by clinicians and scientists. A large observational study carried out by Mao et al. showed a high prevalence of central nervous system disorders, such as dizziness, headache, impaired consciousness, acute cerebrovascular disease, ataxia or presence of seizures [3]. Infection with SARS-CoV-2 causes acute brain dysfunction in the form of delirium in a significant proportion of patients in the acute stage of COVID-19 [4]. Data from the intensive care units reported that delirium prevalence ranged from 45% to 84%, depending on the delirium identification tools and definitions used in the studies [5][6][7]. Delirium is an acute brain disorder, potentially reversible, that commonly occurs in critically ill patients with a pathomechanism related to neuroinflammation and oxidative stress. Delirium identification is based on clinical observation and is characterized by rapid onset, significant symptom fluctuation during the day, disturbance of the wake and sleep cycle and changes in thinking, memory and behavior [8].
Undoubtedly, delirium monitoring should always be carried out in a formalized manner, using dedicated, well-established guidelines [9] and validated diagnostic scales, such as the Confusion Assessment Method for Intensive Care Unit (CAM-ICU) or the Intensive Care Delirium Screening Checklist (ICDSC) [10]. However, the adherence to clinical guidelines in temporary acute care hospitals, including delirium identification and management may have been and still may be low due to staff shortages and heavy workload [2]. In many places acutely ill patients were unable to receive a formal admission to the ICU and were provided with a bed at high-dependency acute units, which employed a wide range of non-ICU healthcare professionals, not trained in using validated scales to assess acute consciousness impairment.
In such a situation, formal monitoring of acute brain dysfunction within the temporary ICU is usually impossible to be implemented with the personnel turnover being too high and/or the willingness to implement an additional monitoring scale too low. The current COVID-19 pandemic has made it necessary to adapt the guidelines to real-life scenarios of a sudden increase in the number of patients over a short time [11,12]. The authors of this analysis focused on reporting real-life data regarding monitoring of acute changes of consciousness in a non-research, ad-hoc organized medical facility that had no training options for formal delirium screening. We hypothesized that it is crucial to identify delirium, as an early and often the only sign of deterioration of homeostasis (hypoxia, infection, ion disturbance, senses impairment or drug withdrawal) and treat accordingly. Therefore, the objective of this study was to investigate the frequency, risk factors and prognosis of delirium in patients with COVID-19 admitted to acute care temporary hospitals at the Pomeranian Medical University in Szczecin, Poland.

Materials and Methods
The authors conducted an observational retrospective cohort data analysis from 3 acute care units of 2 temporary hospitals in Szczecin, Poland. The data comes from an ad-hoc high-dependency acute care unit ("oxygen unit") with passive oxygen, high-flow nasal oxygen therapy (HFNOT) and non-invasive ventilation (NIV). If any of the patients deteriorated, they were admitted to a temporary ICU ("ventilation unit"). The data was collected between October 2020 and February 2021, after receiving permission from the university hospital management.

Ethical Considerations
The study received a waiver from the Bioethical Committee of the Pomeranian Medical University due to its retrospective, observational nature (decision no. KB-0012/15/02/2021/Z dated 3 February 2021).

Study Population
Adult patients (≥18 years old) with positive antigen tests (approved in Poland with diagnostic sensitivity ≥90% and specificity ≥97%) or reverse transcription polymerase chain reaction (RT-PCR), hospitalized in the "oxygen unit" of temporary hospitals were included into the study. Authors excluded severely ill patients with predominant severe comorbidities without symptoms of coronavirus infection (especially pneumonia).

Data Collection
Data were retrieved from an electronic hospital database and included: study group characteristics (demographic data, comorbidities, addictions, medications on admission), COVID-19 symptoms on admission, laboratory testing on admission, respiratory and ventilation parameters, in-patient hospital treatment and complications. The assessment of consciousness was carried out at least 2 times a day, in the morning (between 9:00 and 11:00 a.m.) and in the evening (between 19:00 and 21:00 p.m.), including neurological examination and autopsychic and allopsychic orientation. Delirium was identified through a retrospective chart review method according to DSM-5 criteria if present at least once during hospitalization [7]. Data regarding outcome (death, hospital length of stay, ICU length of stay, complications, discharge information) were retrieved from the hospital computer database.

Statistical Analysis
All analyses were performed using licensed software Statistica 13 (StatSoft, Inc., Tulsa, OK, USA). The continuous variables are presented as mean, standard deviation (SD) and median. The categorical variables are presented as numbers and a percentage. The Mann-Whitney U test was used to compare continuous variables. The chi-square test or chi-square test with Yates's correction was used to compare qualitative data between the two groups of patients. The receiver operating characteristic (ROC) curve analysis was performed to determine the best cut-off value for predicting NLR values for delirium prediction. Moreover, the analysis of the relationship between the complications, follow up and delirium was performed using logistic regression with model analysis adjusted by data (age and gender). Kaplan-Meier analysis calculated the probability of survival. Statistical significance was set at p-value ≤ 0.05.

Results
The analysis included a group of 201 patients, 162 did not develop signs of delirium during hospitalization, whilst 39/201 were diagnosed with at least one episode of delirium (19.4%). Demographic data and comorbidities on admission are presented in Table 1. Patients with delirium were more often males (p = 0.002), older (p < 0.001) and individuals who scored higher in the clinical frailty scale (CFS) (p < 0.001). The majority of patients with delirium had been previously diagnosed with arterial hypertension (p = 0.023), chronic heart failure (p = 0.019), chronic kidney disease (p = 0.048) and diabetes treated with insulin (p = 0.002). Legend: COPD-chronic obstructive pulmonary disease, CPI-chronic peripheral ischemia, Me-median, BMI-body mass index, CFSclinical frailty scale, CKD-chronic kidney disease, EF-ejection fraction, GFR-glomerular filtration rate, ICA-internal carotid artery, n-number of patients, p-statistical significance, SD-standard deviation. Table 2 provides information on pre-admission medications used by the patients included in the study which were continued during hospitalization if no contraindications occurred. Individuals diagnosed with at least one delirium episode during hospitalization were treated with diuretics (p = 0.037) and insulin (p = 0.002) more often in comparison with non-delirious patients. Individuals who were diagnosed with delirium during their hospital stay did not differ statistically from patients without consciousness impairment in terms of SARS-CoV-2-related symptoms upon admission to the hospital. The data are presented in Table 3. Table 3. Coronavirus-related symptoms on admission to the hospital.

Symptoms on Admission
No Delirium (n = 162)
Ventilation parameters on admission between the two groups were compared and are visible in Table 5. The use of non-rebreather masks among patients with delirium was higher (p = 0.010) and the oxygen flow rates were higher in this group (p = 0.013).
In patients with delirium, there were lower levels of partial pressure of oxygen (PaO 2 ) in arterial blood gas samples when compared to non-delirious individuals (p = 0.043). Fraction of inspired oxygen (FiO 2 ) rates were also higher in delirious patients (p = 0.006). A statistically significant higher percentage of delirium was observed in patients with assisted ventilation (HFNOT + NIV) than without this type of therapy (p < 0.001). Legend: ARDS-acute respiratory distress syndrome, BE-base excess, FiO 2 -fraction of inspired oxygen, HFNOT-high-flow nasal oxygen therapy, Me-median, n-number of patients, p-statistical significance, pCO 2 -partial pressure of carbon dioxide, pO 2 -partial pressure of oxygen, SD-standard deviation, SpO 2peripheral oxygen saturation. Table 6 presents the analysis of COVID-19-specific treatment administered during hospital stay. A greater percentage of patients with delirium required a therapeutic dose of low-molecular-weight heparin (LMWH) (p = 0.004). When assessing antibiotic therapy, a greater percentage of patients with delirium received treatment with azithromycin (p = 0.019) or another antibiotic other than ceftriaxone (p = 0.035). The analysis of steroid therapy showed that hydrocortisone was used in a greater number of patients with impaired consciousness (p < 0.001), and there was a higher maximum dose of dexamethasone (p = 0.037) among these individuals as well.
A detailed multivariable analysis of complications was performed. During treatment, a significantly greater number of delirious patients experienced cardiological (OR 7.720, p < 0.001), pulmonary (OR 8.788, p < 0.001) and sepsis (OR 3.991, p = 0.029). The mortality rate was higher in patients with delirium (46.15% vs. 3.70%, p < 0.001) with OR 17.212, p < 0.001). The data are presented in Tables 7 and 8. A survival probability assessment in delirious and non-delirious COVID-19 patients was performed using the Kaplan-Meier curve (Figure 1). It presents a statistically significant difference in 30-day survival between both groups in favor of patients without delirium (p < 0.001). A survival probability assessment in delirious and non-delirious COVID-19 patients was performed using the Kaplan-Meier curve (Figure 1). It presents a statistically significant difference in 30-day survival between both groups in favor of patients without delirium (p < 0.001).

Discussions
Delirium accounts for sudden, onset fluctuating impairment of consciousness commonly observed in critically ill patients that cannot be explained by preexisting neurological disorders [13,14]. The results of this study report that the odds of mortality in patients with COVID-19 presenting with delirium is seventeen times higher (as   Legend: AKI-acute kidney injury, CKD-chronic kidney disease, n-number of patients, OR-odds ratio, p-statistical significance, pCO 2 -partial pressure of carbon dioxide, pO 2 -partial pressure of oxygen. Note: Delirium * adjusted by age and gender.

Discussions
Delirium accounts for sudden, onset fluctuating impairment of consciousness commonly observed in critically ill patients that cannot be explained by preexisting neurological disorders [13,14]. The results of this study report that the odds of mortality in patients with COVID-19 presenting with delirium is seventeen times higher (as adjusted to age and gender) as compared to patients without this consciousness disturbance during hospital stay and reach an incidence rate of 46.15%. Delirium has been reported to contribute to worsened outcomes in severely ill patients, from prolonged hospitalization, to increased risk of long-term cognitive impairment, neuropsychiatric disorders and even death [15,16]. Much attention has been placed recently on the CNS complications of COVID-19, especially in patients presenting with severe infection [17] and in the elderly population [18]. Various factors could lead to delirium development in COVID-19. Direct pathological effects of the SARS-CoV-2 virus on the brain cells, release of CNS inflammatory mediators, peripheral organ systems insufficiency, sedative strategies, prolonged mechanical ventilation time and social isolation have been listed as possible contributing factors [12]. ARDS-related hypoxemia and oxidative stress are further possible underlying causes [19]. Other authors have also reported systemic infections, metabolic and endocrine alterations, such as electrolyte disbalance, anemia, hyperglycemia and hypoalbuminemia as etiologies related to delirium occurrence in COVID-19 [20].
Increased mortality in the delirium group had been previously confirmed in a study by Rebora et al. in an Italian cohort of COVID-19 patients hospitalized in four acute medical units [21]. Authors report that 14.1% of patients presented with delirium on admission displayed almost a two-fold chance of in-hospital mortality compared to those in the non-delirium group [21]. Our results show that 19% of patients were identified to have delirium, but mortality was 17-times higher. In a study by Kennedy et al., 28% of elderly patients with COVID-19 admitted to ED displayed signs of delirium at presentation [22]. In a retrospective study of COVID-19 patients performed by Mao et al., the incidence rate for impaired consciousness reached 14.8% in severe infections defined according to the American Thoracic Society guidelines for community-acquired pneumonia [2]. These are quite contrary to the results of a meta-analysis performed by Nazari et al., who reported impaired consciousness in 1.9% of COVID-19 patients [23]. In the study by Mao et al., patients more prone to develop consciousness disturbances were older and more burdened with chronic diseases, especially arterial hypertension, compared to the non-severe infection group [2]. These outcomes stay in accordance with our results as we indicate significantly increased mean age and increased prevalence of arterial hypertension in the group of patients with impaired consciousness. In our research, heart failure, chronic kidney disease and diabetes mellitus treated with insulin injections on admission also contributed to increased risk for altered states of consciousness during hospitalization. Interestingly, none of the comorbidities predominated in the delirium group in comparison to the non-delirium group in a cohort analysis of French ICU patients with COVID-19 [4].
We further report that the male sex and higher CFS scores predispose to delirium in a COVID-19 cohort. The male sex has been previously associated with disease severity measured on admission to the ICU and mortality in a global-scale meta-analysis of more than 3,000,000 cases [24]; however, to the best of our knowledge, no study has previously linked male sex to increased prevalence of consciousness disturbances. Frailty, defined as ≥ 5 points CFS was found to significantly predict delirium in a hospital cohort of elderly patients diagnosed with SARS-CoV-2 infection [25]. Similar conclusions have been drawn by Spanish researchers [26].
The intensity of oxygen therapy in the study group had a statistically significant link with the occurrence of delirium, although initial peripheral oxygen saturation did not differ between the two groups. As expected, higher oxygen flows, as well as the need to use nonrebreathing masks or assisted ventilation (both HFNOT and NIV), and hence high FiO 2 , were associated with greater risk of delirium (Table 5). These assumptions were confirmed by Pun et al. [6]. Nazari et al. and Krewulak et al. drew similar conclusions [23,27]. The implementation of mechanical ventilation was inextricably linked with the occurrence of ARDS, which in turn predisposes to hippocampus damage and cognitive impairment. It is simultaneously influenced by the level of hypoxia [28,29]. According to Krewulak et al., delirium itself is associated with worse outcomes and mechanical ventilation is a modifiable risk factor for impaired consciousness. All of the above has also been confirmed by our research. Therefore, any interventions aimed at avoiding the implementation of mechanical ventilation, or shortening its duration, can improve the prognosis. The results of our research show that the survival rate was largely influenced by pulmonary complications, including fibrosis. The risk of death increased almost 17-fold in the case of fibrotic lesions in the lung tissue in delirious patients.
According to Rello et al., the way oxygen therapy is administered in patients with COVID-19 is of great importance. Moderate or severe hypoxemia despite HFNOT usually requires intubation [30], worsening the prognosis. However, this thesis is not confirmed by the Frat study [31]. According to their data, in patients with non-hypercapnic acute hypoxemic respiratory failure, treatment with high-flow oxygen, standard oxygen or noninvasive ventilation did not result in significantly different intubation rates. There was a significant difference in favor of high-flow oxygen in 90-day mortality.
In our study, we also tried to discover any specific laboratory findings on admission that could act as predictors of developing consciousness impairment during COVID -19, as no symptoms at presentation were found to be significantly linked to an increased risk of mental deterioration. We found that higher inflammatory parameters (such as leukocyte and neutrophil count, CRP, IL-6, PCT or LDH) were observed in patients whose mental state had deteriorated during their hospital stay. McNeil et al. concluded that IL-6 serum level is independently associated with delirium prevalence and length but found no correlation with C-reactive protein levels [32]. Knopp et al. correspondingly to our findings, revealed that increased CRP and NLR (neutrophil-to-leukocyte ratio) were linked to worse outcomes [33]. Multiple studies have shown that hyperinflammation and cytokine storm may trigger neurological symptoms in COVID-19 patients but found no single specific biomarker that can be used as a predictor of deterioration.
In our study, elevated creatinine and urea serum levels were linked to higher rates of cognition impairment. Interestingly, we found no significant difference in developing consciousness decline when it comes to ion composition on admission, but, as a human's body can maintain homeostasis in a wide range, we speculate that biochemical signs of kidney injury may act as a herald here. The kidney can be a direct target for SARS-CoV-2 leading to tubular damage, therefore patients with AKI should be considered as those with a higher risk of complications, including neuropsychiatric manifestations and death [34][35][36]. In a study by Toklu et al., over 25% of COVID-19 patients with neurological symptoms, including altered mental status and impaired consciousness, had hypernatremia. In addition, 20% of this cohort had persistent hypernatremia for over 48 h of their hospital stay [37]. Maguire et al. found that impairment of cognition, elevation in CRP, urea and NLR, but decrease in lymphocyte count, were linked to increased 30-day mortality [38]. Another finding in our study seems to be somewhat intuitive: higher levels of serum D-dimer is linked with higher prevalence of delirium. Coagulation is known to be significantly altered in the course of SARS-CoV-2 infection, thus giving us a wide spectrum of clinical presentations, i.e., pulmonary embolism, stroke, arterial and venous thrombosis, excessive bleeding, etc. Diaz-Perez reported that acute alteration in mental status in COVID-19 patients can be the only presentation of multiple small foci of cerebral strokes [39]. Moreover, pulmonary embolism leading to hypoxia can be an underlying cause of consciousness impairment [40][41][42]. In our results as well, elevated levels of cardiac T-troponin were linked to higher occurrence of delirium during hospital stay. Imazio et al. hypothesized that during SARS-CoV-2 infection, myocardial injury has rather a non-ischemic character and occurs secondary to hypoxia, sepsis, embolism or myocarditis [43]. A meta-analysis by Sandoval et al. showed that in COVID-19 patients with elevated troponin serum levels, the mortality rate could reach up to 75% [44]. Having mentioned all of the above, we suggest that elevated serum troponin on admission could act as an early warning for delirium development and death.
The neutrophil-to-lymphocyte ratio (NLR) is a novel, non-specific index of disease severity, previously studied in a vast range of inflammatory morbidities or malignancies, and in delirium [45]. It has been previously reported that a higher NLR may be associated with delirium in severely ill patients with acute ischemic stroke (NLR > 4.86) [46], delirium in hospitalized elderly patients (NLR > 3.626) [47] and in postoperative delirium after hip fracture (NLR ≥ 3.5) [48]. There is no consensus on what a normal range of NLR is, as it seems to vary slightly depending on age and race. Luo et al. analyzed almost 6000 healthy Chinese adults and proposed that the NLR should fall between 0.88 and 4.0 [49]. In a study on 2212 healthy Iranians by Moosazadeh et al., the NLR normal value was established between 1.0 and 2.4 [50]. Christiansen et al., in a large register-based study, aimed to assess NLR distribution in the Danish population by comparing patients in general practice and ICU. They concluded that in the GP group with low CRP ("healthy"), the NLR median was 1.85, and that the ICU group had a significantly higher median NLR [51]. In our research, we studied the Polish population, in which normal reference ranges of NLR had not been evaluated yet. Nevertheless, we found a statistically significant difference between NLR values between non-delirious and delirious COVID-19 patients, the delirious patients presenting with a higher index, with the cut-off point at ≥6.51. To our knowledge this is the first study to report this phenomenon.
We have considered short-term complications associated with COVID-19. They came from various systems, including cardiovascular, respiratory and urinary, with cardiological complications-mostly atrial fibrillation and heart failure-increasing the risk of consciousness disorders the most (by 7.7 times). Among pulmonary disorders, the strongest aggravating factor turned out to be fibrosis (risk increased 8 times), and to a lesser extent respiratory failure itself (6.3 times). The clinical manifestation of pneumonia was also worsening prognosis, as it was in the cases in a German report [52]. According to our results, sepsis increases the risk of developing mental impairment 4-fold. Despite the common knowledge about their link with worse prognosis in patients without COVID-19 [53,54], their interaction with SARS-CoV-2 infection and delirium is still not fully discovered and requires further research. To our knowledge, there is no available statistical data regarding this type of correlation. Many authors, including Völk [52] and Garcez [55], found one of the more frequent comorbidities, which were cerebrovascular diseases (ischemic stroke, TIA, etc.), which was not statistically significant in our research, despite the occurrence of such complications.
Limitations: The authors realize that this study is not without limitations. First, the retrospective character of the study should be acknowledged, rendering our ability to interpret the relationship between delirium and mortality as associative rather than cause and effect. The collected database reaches from 3 up to 7 months back. During that period some changes have been applied to the general guidelines of COVID-19 treatment (i.e., corticosteroids or antibiotics dosing), therefore its potential impact on the final prognosis in particular individuals might remain considerable. Second, the follow-up is terminated by the patient's discharge from the hospital, so no further data on the survival or patient condition was collected subsequently. Third, the core of the staff of the temporary hospital was formed by clinicians of several specializations, principally representing subspecialties of internal medicine (nephrologists, cardiologists, pulmonologists), but sporadically also surgeons, neurologists and obstetricians. Whilst laboratory tests remain irrefutable, the approach to assessing delirium may vary amidst different specialists. The study was carried out in two medical centers, in which different emphasis was put on each of the aspects of diagnostics as well as treatment. Thus, a minor bias in data acquisition might have occurred. Fourth, according to critical care guidelines, dedicated scales should be used to identify delirium in acutely ill patients. Introducing the use of validated delirium screening scales was the greatest difficulty in a temporary hospital due to a rapid personnel changeover and lack of time to perform an additional task during the wave of admissions. Nevertheless, in everyday clinical practice with a heavy workload, even the simplest measures identifying impairment of consciousness will identify the patients at risk of having worse outcomes. Monitoring and prevention of delirium (hypoxia, infection, ion disturbance, senses impairment or drug withdrawal) is one available mechanism by which to attempt to reduce mortality.

Conclusions
The results of this study report that the odds of mortality in patients with COVID-19 presenting with delirium during their hospital stay is over seventeen times higher as compared to patients without acute brain dysfunction. In critical illness, such as severe SARS-CoV-2 infection, delirium is an early and often the only sign of deterioration of homeostasis, and thus should be monitored and prevented to avoid increased mortality.