Superinfections in critically ill COVID-19 patients - an association with Dexamethasone?

Bacground: Super-infections in COVID-19 patients with acute respiratory distress syndrome (ARDS) on mechanical ventilation were initially reported to be rare. Little is known of their incidence after dexamethasone was introduced as standard care. We aimed to determine the incidence and characteristics of superinfections in mechanically ventilated COVID-19 patients during the course of the COVID-19 pandemic, and explore the possible impact of the introduction of dexamethasone as standard therapy. Methods: In this national, multi-center, observational, retrospective study we included patients ≥ 18 years admitted from March 1 st 2020 to January 31 st 2021 with polymerase chain reaction (PCR)-confirmed SARS-CoV-2 infection treated with invasive mechanical ventilation. Data was collected from electronic health records. Patient characteristics, clinical findings, microbiology, length of stay and 90-day survival were examined with backwards stepwise multiple regression. Results: 155 patients (115 men, mean age 62 years, range 26-84 years) were included. 73 patients (47%) had a total of 101 superinfections where pneumonia dominated (70%). Superinfections were more commonly observed in patients receiving dexamethasone (67% vs 30%, p<0.0001), and in patients with pre-existing autoimmune disease (18% vs 5%, p<0.01). Invasive fungal infections were reported exclusively in dexamethasone-treated patients [9/72 (13%) vs 0/83 (0%), p<0.0001]. There was no difference in 90-day survival between patients with and patients without superinfections (64% versus 73%, p=0.238). In multiple regression analysis, superinfection was associated with dexamethasone use [OR 5.35 (2.62 – 11.35), p<0.001], pre-existing autoimmune disease [OR

Take home message: Dexamethasone therapy has increased survival in COVID-19 ARDS.We demonstrate in a national multi-center retrospective study that it also is strongly associated with superinfections in patients receiving invasive mechanical ventilation.

Introduction
Bacterial or fungal superinfections following viral infections are well known from seasonal influenza epidemics, mostly in hospitalized patients and in particular in those needing treatment in an intensive care unit (ICU) [1].At the onset of the SARS-CoV-2 pandemic, similar complications were anticipated.However, in early reports, the incidence of superinfections in hospitalized COVID-19 patients was remarkably low, with estimates not exceeding 16%, although in some studies, these infections had a negative impact on outcome [2][3][4][5][6][7].Later publications suggested that COVID-19 patients did not show the same propensity for superinfections as patients with Influenza virus, and concluded that use of empiric antimicrobial therapy was not warranted [8,9].Still, in critically ill COVID-19 patients with acute respiratory distress syndrome (ARDS) on mechanical ventilation, a markedly higher rate of superinfections was reported [4,[10][11][12][13].
During the summer of 2020, it was demonstrated in randomized trials that dexamethasone improved survival in hospitalized patients with severe COVID-19 [14,15].Dexamethasone was rapidly implemented as standard of care in most countries.After the introduction of this therapy in Norway, the impression was that more Covid-19 patients acquired super-infections.In particular, we observed numerous cases of Enterobacteriaceae and fungus in patients with suspected pneumonia receiving mechanical ventilation.At present, data on the impact of dexamethasone therapy on the incidence of superinfections in hospitalized severely ill COVID-19 patients is limited.A multi-center retrospective study was therefore established in the south-eastern region of Norway to examine if superinfections in hospitalized COVID-19 patients on mechanical ventilation have increased during the COVID-19 pandemic, and whether the incidence of such complications was impacted by the introduction of dexamethasone as standard care.

Study characteristics and data collection
This was a multi-center, observational, retrospective study with seven participating hospitals in the south-eastern region of Norway.The protocol was approved by the Regional Committee for Health and Research Ethics for Exclusion criteria were withdrawal of consent following distribution of a written summary of the study protocol to living eligible patients.In deceased patients, implied consent was assumed.
Data were manually collected from electronic medical records and charts and plotted into an electronic database (Ledidi.com,Oslo, Norway).Variables included gender, age, co-morbidities, blood values and clinical characteristics (on admission and during ICU stay), incidents of superinfections, putative causative organisms, use of dexamethasone and other immune-modulating drugs, targeted and/or empirical use of antimicrobial agents, length of stay in ICU and hospital, and 90-day survival (obtained from hospital records automatically updated from the Norwegian National Population Registry).

Definition of superinfections
A superinfection was defined as an incident when the following occurred: 1) clinical deterioration, 2) findings of pathogenic microorganisms (bacterial, fungal or viral) by cultures (bacteria and fungus), Galactomannan antigen assay (GM-EIA) (BioRad, Hemel, Hempstead, UK) or PCR (inhouse methods for fungus and virus) in clinical specimens and 3) the introduction of targeted antimicrobial therapy or the continuation of empiric antimicrobial therapy in agreement with the findings.Findings of coagulase-negative staphylococci in a single blood-culture and Candida spp. in airways were considered culture contaminators or commensal flora and were therefore not counted as superinfections.

Definition of pulmonary infections
The definition of ventilator associated pneumonia (VAP) includes radiographic infiltrates.However, the radiological presentation of COVID-19 ARDS precluded use of new pulmonary findings as an indication of VAP.Pulmonary infection was therefore in our dataset based on 1) respiratory deterioration and/or an increase in inflammatory biochemical markers and 2) microbiological findings and 3) the introduction of targeted antimicrobial therapy or the continuation of empiric antimicrobial therapy in agreement with the findings.

Statistical analysis
Reported values represent counts (%) and medians (25 th -75 th centiles) unless otherwise noted.Chi square test and Wilcoxon rank sum test were used to compare demographic and clinical characteristics among patients with and without superinfection, and to compare infections, microbiological findings and outcome with and without dexamethasone therapy.Factors associated with presence of infection were analyzed using backwards stepwise multiple logistic regression with p-value thresholds 0.20 to enter and 0.10 to leave the model (Fit Model platform, Personality Stepwise, SAS-JMP 13.2.0 software for Mac, SAS Institute, Cary, NC, USA).Potential dichotomous (yes/no) predictors of superinfection explored included gender, hypertension, diabetes, chronic heart disease, chronic lung disease, chronic kidney disease, autoimmune disease, malignancy, ongoing immunosuppression prior to COVID-19 infection, dexamethasone administered related to COVID-19, and other immune-modulating drugs administered.Potential continuous predictors of superinfection included age by decile, body mass index (BMI), C-reactive protein (CRP), and neutrophil and lymphocyte counts.Results are reported as odds ratio (OR) with 95% confidence levels (95% CI) for having an infection versus no infection.
The statistical significance level was set at p<0.05.

Patient characteristics
156 patients fulfilled inclusion criteria, of which one opted out.A total of 155 patients (115 men, mean age 62 years, range 26-84 years) were included from seven participating hospitals, with median 36 patients (range 5-50) per center (Table 1).Time from symptom start to hospital admission was 8 days (IQR 6-11).Time from hospital admission to administration of dexamethasone was 1 day (IQR 1-5).Time from hospital admission to ICU admission was 2 days (IQR 1-4).All patients had received mechanical ventilation; seven received ECMO therapy.Lowest value of PaO2/FiO2 (P/F) ratio during the ICU stay was 14 kPa (105 mmHg; IQR 9-20).Overall 90-day survival rate was 69%.Patients' risk factors for severe COVID-19, clinical characteristics, use of dexamethasone and other immunomodulating drugs, length of stay in the ICU and hospital, and outcome are shown in Table 1.

Occurrence of superinfections
A total of 101 superinfections were detected in 73/155 (47%) of patients.First occurences included 64 monomicrobial and 9 polymicrobial infections.Fifty-five patients (75%) had only one superinfection (50 monomicrobial and 5 polymicrobial).Eighteen patients (25%) developed a second infection (14 with monomicrobial and 4 with polymicrobial).Eight of these (11% of total) experienced a third infection (7 monomicrobial and 1 polymicrobial), and two patients had a fourth incident while in the ICU.Respiratory infections were found in 70% of the infections, followed by 8% urinary infections, 7% bloodstream infections and 3% combined respiratory and bloodstream infection.
The timing of superinfections in relation to hospital admission is presented in Figure 1.Table 2 summarizes all significant microbial findings and specimens of origin.In many patients, findings of the same microbe in more than one specimen occurred.Gram-negative rods were the dominante pathogens identified, while Staphylococcus aureus was the most common organism detected.Fungal infection was diagnosed in 9 patients, five fulfilling the the recently redefined criteria for possible invasive aspergillosis in research and four with Pneumocystis jirovecii pneumonia (16).
Comparing the patient groups with and without superinfections, we found that patients with superinfections more often had received dexamethasone or an equivalent dose of methylprednisolone (Table 1).Patients with superinfections also more often had pre-existing autoimmune diseases, and had higher maximum neutrophil count, lower minimum PaO2/FiO2 ratio and longer hospital stay, ICU-stay and time on ventilator support.However, 90-day survival rates for patients with superinfections were not significantly lower than for those without superinfections (64% versus 73%, p=0.238).Moreover, high CRP levels or low number of lymphocytes, other co-morbidities than autoimmune disorders, or use of immunosuppressive drugs at the time of hospital admission were not significantly associated with the occurrence of superinfections.Use of hydroxychloroquine was statistically associated with a lower number of superinfections, but importantly, in our patient cohort, hydroxychloroquine use stopped abruptly at the same time as dexamethasone was introduced as standard therapy.

Discussion
In this study of 155 consecutively treated COVID-19 patients requiring mechanical ventilation, we demonstrate that 47% experienced 1-3 superinfections according to our definition throughout their ICU stay.Significantly more superinfections occurred after the introduction of dexamethasone as standard treatment.The use dexamethasone in these critically ill COVID-19 patients was independently associated with occurrence of superinfections, with an odds ratio of 5.4.As expected, COVID-19 patients with superinfections had longer hospital stays and time on ventilatory support.However, in this small study we could not demonstrate a statistically significant effect of superinfections on 90-day survival.Our findings suggest that whereas dexamethasone has been shown to increase survival in critically ill COVID-19 patients, it also appears to increase the risk of clinically relevant superinfections.
While it is not surprising that use of corticosteroids increases the risk of infections in COVID-19 patients, available data on the incidence of superinfections in COVID-19 patients in the dexamethasone era is scarce.In initial studies on its use in COVID-19, superinfections were not included as secondary outcome [14][15].A later meta-analysis suggested a possible increase in secondary infections following corticosteroid therapy, but occurrence of secondary infection was not a pre-defined endpoint in the majority of the studies included, so the results should be interpreted with caution [17].In a single-center study on severely ill patients, Saade et al.
reported that dexamethasone was associated with increased risk of superinfection, but importantly, this population had a high prevalence of underlying immune defects (due to malignancies and organ transplantation (34%) [12].A recent study of ICU patients from three French ICUs reported a higher incidence of superinfections than that of the present study, but they found no association between the occurrence of VAP and the use of dexamethasone [18].One possible contributor to discrepancies between our studies is differences in the utilization of broad-spectrum antibiotics, previously demonstrated to be independently associated with superinfections in COVID-19 [11].Although 95% of our patients received empiric antibiotics, this to a large extent involved the use of relatively narrow-spectrum agents, due to moderate levels of antimicrobial resistance in the Scandinavian countries as demonstrated in EARS-Net [19].Furthermore, in the French study the nondexamethasone treated "first wave" patients were almost twice as likely to be on a ventilator at ICU admission and had 50% longer ventilator and ICU time than the dexamethasone treated "second wave" COVID-19 patients, potentially influencing the results.Of note, the definition of pneumonia in mechanically ventilated patients was quite similar to ours.
In the present study, we found an independent association between the use of dexamethasone and the occurrence of superinfections in mechanically ventilated COVID-19 patients.Importantly, no such association was found between the use of immunosuppression prior to hospital admission or for other pre-existing co-morbidities including malignancies.Of note, an association was found between the occurrence of superinfection and preexisting autoimmunity, although the number of patients with autoimmunity was relatively low.The association with dexamethasone was strongly present also after adjusting for occurrence of autoimmune disease.
Pneumonia was the most frequent form of infection observed, and nine patients (6%) had probable invasive fungal infection.Early in the pandemic, COVID-19-associated pulmonary aspergillosis (CAPA) was reported with varying incidence, with a median of 13.5%, ranging from 2.5% to 35% in a review [20].In three prospective trials, the incidence was somewhat higher (14% to 38%) [21][22][23].In ICU patients receiving mechanical ventilation, Marr et al. suggested an incidence of 20-30% [24].However, few reports on the impact of corticosteroids used in COVID-19 have so far been published, with the exception of its association with the development of mucormycosis in Indian patients with diabetes mellitus [25,26].Although of low incidence, in our study invasive fungal infections in COVID-19 patients receiving invasive ventilation was strongly associated with dexamethasone therapy.
A clear limitation of this study is its observational, retrospective design.However, it was a multicenter study and included most Norwegian hospitals where COVID-19 patients were treated before dexamethasone was implemented as standard therapy in patients with ARDS.Due to the limited value of radiological diagnostics in assessing VAP in COVID-19 ARDS, the diagnosis was based solely on perceived clinical deterioration and microbiological findings, possibly leading to overestimation of true superinfections.Finally, patient numbers were relatively low possibly influencing some sub-analyses such as the occurrence of fungal infection.

Conclusion
We report that critically ill, invasively ventilated COVID-19 patients receiving dexamethasone had an almost five times higher odds ratio for contracting superinfections while in the ICU.These superinfections were associated with prolonged mechanical ventilation and ICU and hospital stay.Other anti-inflammatory and immunosuppressive agents are now being tested in COVID-19 patients, often in combination with dexamethasone.Based on the present findings, a particular focus on superinfections is warranted.

Figure Captions
South-East Norway (REK: 219370).The study population included all patients at or above 18 years of age admitted from March 1 st 2020 until January 31 st 2021 who fulfilled the following inclusion criteria: (1) SARS-CoV-2 infection confirmed by polymerase chain reaction (PCR), (2) ARDS, and (3) treatment during hospitalization included invasive mechanical ventilation and/or extra-corporal membrane oxygenation (ECMO).

Fig. 1
Fig. 1 Time from admission until superinfection in 155 COVID-19 patients on invasive mechanical ventilation Total count was 118 positive specimens.Red dots indicate findings in patients receiving or having received dexamethasone or an equipotent dose of prednisolone as COVID-19 treatment (72 of 155 patients).

Figures
Figures

Figure 1 Time
Figure 1