Impact of dexamethasone on the incidence of ventilator-associated pneumonia and blood stream infections in COVID-19 patients requiring invasive mechanical ventilation: a multicenter retrospective study

Background Dexamethasone decreases mortality in patients with severe coronavirus disease 2019 (COVID-19) and has become the standard of care during the second wave of pandemic. Dexamethasone is an immunosuppressive treatment potentially increasing the risk of secondary hospital acquired infections in critically ill patients. We conducted an observational retrospective study in three French intensive care units (ICUs) comparing the first and second waves of pandemic to investigate the role of dexamethasone in the occurrence of ventilator-associated pneumonia (VAP) and blood stream infections (BSI). Patients admitted from March to November 2020 with a documented COVID-19 and requiring mechanical ventilation (MV) for ≥ 48 h were included. The main study outcomes were the incidence of VAP and BSI according to the use of dexamethasone. Secondary outcomes were the ventilator-free days (VFD) at day-28 and day-60, ICU and hospital length of stay and mortality. Results Among the 151 patients included, 84 received dexamethasone, all but one during the second wave. VAP occurred in 63% of patients treated with dexamethasone (DEXA+) and 57% in those not receiving dexamethasone (DEXA−) (p = 0.43). The cumulative incidence of VAP, considering death, duration of MV and late immunosuppression as competing factors was not different between groups (p = 0.59). A multivariate analysis did not identify dexamethasone as an independent risk factor for VAP occurrence. The occurrence of BSI was not different between groups (29 vs. 30%; p = 0.86). DEXA+ patients had more VFD at day-28 (9 (0–21) vs. 0 (0–11) days; p = 0.009) and a reduced ICU length of stay (20 (11–44) vs. 32 (17–46) days; p = 0.01). Mortality did not differ between groups. Conclusions In this cohort of COVID-19 patients requiring invasive MV, dexamethasone was not associated with an increased incidence of VAP or BSI. Dexamethasone might not explain the high rates of VAP and BSI observed in critically ill COVID-19 patients. Supplementary Information The online version contains supplementary material available at 10.1186/s13613-021-00876-8.

Background Among patients admitted to the intensive care unit (ICU) with a severe form of coronavirus disease 2019 (COVID- 19), up to 80% [1] require invasive mechanical ventilation (MV). The mortality of these patients has been reported to be as high as 37% [1]. A large randomized controlled trial [2,3] demonstrated that the use of dexamethasone resulted in a lower 28-day mortality in patients who were receiving invasive MV. Consequently, during the second wave of pandemic, dexamethasone has become the standard of care for COVID-19-related pneumonia in the ICU [4,5]. The frequency of severe forms of acute respiratory distress syndrome (ARDS) [1,6] exposes COVID-19 patients to a high risk of nosocomial infections [7]. Unexpected incidence of ventilator-associated pneumonia (VAP) has been recently reported in large series, as high as 50% for COVID-19 patients as compared to 30% for influenza [8]. Very high late-onset VAP rate has been described in patients under extracorporeal membrane oxygenation (ECMO) for severe COVID-19-related ARDS, again as compared with influenza patients [9]. Besides, a high frequency (67%) of bloodstream infections (BSI) [10] has been reported. Considering the wide use of dexamethasone during the second wave of COVID-19 pandemic, its role in the development of secondary bacterial infections comes under question. We therefore conducted a retrospective multicenter observational study, with a before-after design, to compare the incidence of VAP and BSI between patients of the first and second waves of COVID-19, before and after the publication of the RECOVERY [2] trial and the generalization of the use of dexamethasone. We also compared the duration of MV, length of stay and mortality according to the use of dexamethasone.

Study design and population
We conducted an observational retrospective study, with a before-after design, in three ICUs from two University hospitals in Southern France. Patients were included if they had been admitted to the ICU for a SARS-CoV-2 documented acute respiratory failure (from a pharyngeal or pulmonary sample RT-PCR) and required MV for at least 48 h. First wave of pandemic covered from March 10th to May 29th 2020 and second wave from August 14th to November 7th 2020 (end of the study period). Patients for whom withholding of treatments was decided during the first 48 h after ICU admission, aged under 18, deprived of liberty or without social protection were not included.

Baseline assessment and data collection
Data were collected from the electronic patient's file. Demographic characteristics, comorbidities, severity at ICU admission, date of COVID-19 first symptoms and RT-PCR positivity, date of ICU admission, date of intubation and invasive MV, need for ECMO, antiviral treatment, initial bacterial co-infection and antibiotics received at ICU admission, nosocomial infections (VAP and BSI) with microbiological documentation and recurrences, duration of invasive MV, ICU and hospital stay, status at day 28, day 90, ICU and hospital mortality were obtained. Occupational rates, invasive MV and ECMO, hydroalcoholic solution and antibiotic (piperacillintazobactam) consumption during the study period were recorded in the three participating ICUs. Patients were classified according to dexamethasone treatment, administered as in the RECOVERY trial (i.e. intravenous infusion of 6 mg/day during 10 days [2,3]). Patients treated with dexamethasone were called "DEXA+". In these patients, the delay from the initiation of dexamethasone to MV was recorded.
Patients who did not receive dexamethasone were called "DEXA−".
In all patients, the use of rescue immunomodulatory therapies (RIT) was also recorded: the use of methylprednisolone for persistent ARDS as previously described [11], IL-1 inhibitors (anakinra), ruxolitinib or tocilizumab.

VAP and BSI definitions
In patients receiving MV for at least 48 h, VAP was diagnosed when the following criteria were met [12][13][14]: • New or progressive persistent infiltration on chest radiograph. • At least two of the following: • New onset of fever. • Purulent endotracheal aspirate.
Keywords: COVID-19, Dexamethasone, Ventilator-associated pneumonia, Bloodstream infection, Mechanical ventilation (In patients with ARDS, in which demonstration of radiologic deterioration is difficult, at least two of the preceding criteria sufficed).
• Positive culture from broncho-alveolar lavage (BAL) or a positive quantitative culture from endotracheal aspirate (ETA) specimen.
Patients with tracheostomy were considered at risk of VAP only during the period of MV.
ICU-acquired BSI was defined as at least one positive blood culture for bacteria or fungi, drawn at > 48 h after ICU admission was positive. For coagulase-negative staphylococci and other common skin contaminants, at least two consecutive blood cultures positive for the same pathogen at different times and sites were necessary [15].

Relapse and recurrence of VAP
Recurrence of VAP was defined as a new onset of clinical symptoms following a partial or complete regression of the clinical signs after adequate antibiotic treatment with at least one positive bacterial culture at a significant concentration. Relapse was defined as a recurrence involving at least one of the initial causative bacteria; otherwise, it was considered a superinfection [7,16].

Study outcomes
The main study outcome was the incidence of VAP according to the use of dexamethasone. Crude and cumulative incidence, considering death, duration of MV (extubation) and rescue immunosuppressive therapy as competing factors were compared in DEXA+ and DEXA− groups.
Secondary outcomes were the incidence of BSI, recurrence of VAP or BSI, the time from MV to first VAP, the duration of MV, the ventilator-free days (VFD) at day 28 and day 60, ICU and hospital length of stay, ICU, hospital and day 60 mortalities according to the use of dexamethasone.
The same outcomes were analysed in the subgroups of patients who had received RIT associated or not with dexamethasone.

Statistical analysis
Statistical analysis was performed using SPSS Version 20 (IBM SPSS Inc., Chicago, IL, USA) and cmprsk package from R software, version 3.2.3.
Continuous variables are expressed as means ± SD or as median with range (min, max), and categorical variables are reported as count and percentages. Comparisons of means values between two groups were performed using Student's t-test or Mann-Whitney U. Comparisons of percentages were performed using Chi-square test or (Fisher's exact test, as appropriate).
The cumulative incidence function with competing events was used to estimate VAP first episode [17]. Comparisons were done using the Fine and Gray model [18]. The competing risks for VAP were death, extubation and the use of rescue immunosuppressive therapy. All the tests were two-sided. Univariate and multivariate analyses were performed. The statistical significance was defined as p < 0.05. We secondary constructed four groups according the use of dexamethasone and rescue immunosuppressive therapy. Then, we performed for quantitative variables, analysis of the variance (ANOVA) with Bonferroni post hoc tests when significance was (p < 0.05) and for categorical variables, multiple comparisons with Kruskal-Wallis test with Tukey post hoc tests when significance was (p < 0.05).

Patients' characteristics at ICU admission
During the study period, 151 patients were included in the three participating ICUs (Fig. 1). Among them, 84 were treated with dexamethasone, all but one during the second wave of pandemic. Dexamethasone was administered before the period at risk of VAP or BSI. Patients' characteristics at ICU admission are summarized in Table 1. There was no difference except for age, DEXA+ patients being slightly older (66 ± 11 vs. 62 ± 13 years old; p = 0.05), and antibiotics at ICU admission that were less frequently prescribed for DEXA+ patients (52 (62%) vs. 58 (87%); p = 0.001). Patients from the second wave were intubated later after ICU admission as compared to patients from the first wave (1 (0-4) day vs. 0 (0-1) day; p = 0.006). Bacterial co-infections were not different between groups. Dexamethasone was administered 1 (0-4) days before the beginning of MV. (Table 2) VAP occurred, respectively, in 63 vs. 57% (p = 0.43) for DEXA+ and DEXA− patients. The incidence of VAP was 26‰ days under MV overall, 31‰ for DEXA+ patients and 21‰ for DEXA− patients (p = 0.16).The first VAP occurred earlier after the onset of invasive MV in the DEXA+ patients (5 (3-10) vs. 9 (4-15) days; p = 0.02).

Microbiology of VAP and BSI
Gram-negative bacteria (66%) and especially Enterobacteriaceae and non-fermenting Gram-negative bacilli were the most commonly retrieved pathogens during the first episode of VAP. Gram-positive pathogens were mainly methicillin-susceptible Staphylococcus aureus (MSSA) and Enterococcus spp. There was a trend towards more Enterobacteriaceae VAP in the DEXA+ group (72 vs. 52%; p = 0.06). Non-fermenting Gramnegative bacilli were more frequent in DEXA− patients (48 vs. 23%; p = 0.02).
BSI involving Gram-positive pathogens and especially Enterococcus spp. and coagulase-negative Staphylococci were the most frequent. However, Gram-negative bacteraemia was more frequent among DEXA+ patients (11 (39%) vs. 3 (13%); p = 0.04). The source of BSI was a bacteraemic pneumonia for 12 (27%) patients, a catheter-related BSI for 7 (16%) patients and was of unknown origin (primary BSI) for 25 (57%) patients. Microbiological details of BSI in the two groups are provided in Table 5.

Use of rescue immunosuppressive therapy
RIT was used in 46 (30%) patients. There was no difference between the two groups regarding the use of at least one rescue therapy. The details of treatments are presented in Additional file 2: Table S2. The outcomes according to the use of RIT in both groups (DEXA+ and DEXA− patients) are provided in Additional file 3: Table S3. The use of RIT was associated with a higher rate of VAP and BSI regardless the use of dexamethasone.

Discussion
In this retrospective cohort comparing the first and second waves of COVID-19 pandemic in three ICUs, dexamethasone was not associated with an increased incidence of VAP and BSI among patients under MV. Previous series had found discordant data on the frequency of nosocomial infections in COVID-19 patients [19,20]. In our series, the incidence of VAP was higher than described in non-COVID ARDS [21] and comparable to the incidence reported in a recently published large COVID-19 cohort [6]. Noteworthy, in this latter study, very few patients received dexamethasone. The only study that reported the incidence of nosocomial infections in COVID-19 patients treated with dexamethasone found a 12% incidence of VAP and 8% BSI [19], however with a 28-day follow-up only. We performed a 60-day follow-up which might explain the higher incidence described herein. In our cohort, the incidence of BSI was higher than reported in non-COVID patients series [22,23]. Our results are consistent with those from Buetti et al. [24] who reported 14.9% of BSI in the COVID-19 group and only 3.4% in the non-COVID-19 patients. In this study as in ours, corticosteroids did not increase the risk of developing BSI. The high rate of BSI we report might be due to the long duration of MV and ICU stay, partly explained by the 19% patients under ECMO. Surge in the number of critically ill patients challenging the full respect of usual infection control practices as well as the systematic use of gloves [25] have also been raised to explain the high rate of bacteremia. The disruption of the gut barrier caused by SARS-CoV-2 might also increase intestinal permeability and bacterial translocation, favouring the development of BSI, as it has been shown that SARS-CoV-2 productively infects human gut enterocytes [26]. In our cohort, Gram-negative pathogens and especially Enterobacteriaceae and Pseudomonas  11:87 aeruginosa were predominant, as it has been shown in other series, but MSSA and Enterococcus spp. VAP were more frequently retrieved than in other series [6,19].
Apart from dexamethasone treatment, several aspects of COVID-19 might lead to increased risks of nosocomial infections: the complexity of host-response to SARS-CoV-2 infection, including moderate to severe systemic inflammation and/or marked systemic immune suppression [27] as well as the pulmonary vasculopathy with endothelial dysfunction and endothelialitis [28,29]. The high rate of relapses observed, especially in patients receiving adequate antibiotics with monitored pharmacokinetics [7], questions the local diffusion of antibiotics into "COVID-19" lungs.
Although not affecting the incidence of VAP or BSI, treatment with dexamethasone resulted in some differences between patients: VAP occurred earlier and involved less frequently non-fermenting Gram-negative bacteria but rather Enterobacteriaceae. This was unexpected as dexamethasone has been shown to increase susceptibility to Pseudomonas aeruginosa pneumonia in animal models through suppressing iNOS gene expression and peroxynitrite production [30]. The lower use of antibiotics (especially 3rd generation cephalosporins) received at the time of ICU admission during the second wave in dexamethasone patients has probably contributed to modify the patients microbiota. Indeed, guidelines published updated between the 2 first waves discouraged the systematic use of antibiotics given the relatively low rates of bacterial co-infections [4].
Patients receiving dexamethasone had more VFD at D28 as it was found in the CODEX-study [19], a shorter duration of MV and ICU length of stay. These results were in accordance with those from Villar et al. in non-COVID-19 ARDS patients [31]. However, in our cohort, mortality was not different between groups contrary to what was found in the RECOVERY-trial [3]. In a recently published study, Contou et al. [32] did not find any difference concerning the prognosis during the first and second waves. There was no difference in duration of MV, conversely to our results. However, in this paper, authors reported a high ICU mortality, especially for patients requiring MV (57% in the first wave and 75% during the second wave). In our cohort, we found a 38% hospital mortality. This mortality discrepancy might be related to different case mix and hardens comparison with our results.
Finally, we found an association between the use of rescue immunosuppressive treatment and a higher rate of VAP and BSI. Although of importance, this result Fig. 3 Constructing timelines showing the occupational rate, invasive MV and ECMO rates, hydroalcoholic solution consumption, antibiotic (piperacillin-tazobactam) consumption and dexamethasone across the study period. DDD daily defined dose, ECMO extracorporeal membrane oxygenation, IMV invasive mechanical ventilation deserves to be confirmed in a larger cohort and considering potential confounding factors.
Our study has several limitations that should be underlined. Firstly, it is a retrospective design. Secondly, given the relatively small size of the cohort, the study might be underpowered.
Third, the large use of antibiotics at ICU admission during the first wave of COVID-19 pandemic, very limited during the second wave, might have affected the microbiology of secondary bacterial complications. Dexamethasone patients were the less exposed to early antibiotics which might have affected the delay of VAP occurrence but not the global incidence. Fourth, the pulmonary samples used to diagnose VAP as well as the antibiotics prescribed differed according to physicians' practices. However, we did not observe any difference in VAP or BSI incidence or recurrences between centres. Lastly, some patients received rescue immunomodulatory therapies later during their ICU stays [33], which might have favoured nosocomial infections. These patients were equally split in both groups and this confounding factor was considered as a competing event and did not modify the cumulative incidence of VAP.

Conclusions
In this cohort of severe forms of COVID-19 patients requiring invasive MV, dexamethasone was not associated with an increase in the incidence of VAP or BSI. Dexamethasone treatment was associated with more ventilator-free days at day-28. Dexamethasone treatment might not explain the high rate of VAP and BSI observed in critically ill COVID-19 patients.