Discussion

In this study, we assessed the effect of the commonly used glucocorticoid aerosol therapy on the frequency of A. baumannii–positive cultures from lower respiratory tract samples in 3 ICUs in China. After controlling for other variables, our findings showed that glucocorticoid aerosol increased the risk for A. baumannii isolation from critically ill patients on invasive mechanical ventilation.

A. baumannii is ubiquitous in nature and is becoming more frequent in hospitals. In our study, 32% of patients acquired A. baumannii during the 30-day follow-up period. Over the past 2 decades, several studies have attempted to characterize and identify risk factors for A. baumannii colonization or infection. Invasive operations, such as endotracheal mechanical ventilation, inserted invasive devices, ICU stays, recent surgery, use of broad-spectrum antimicrobial drugs, ineffective antimicrobial therapy, and septic shock at diagnosis, are risk factors for MDR A. baumannii colonization or infection and for death^[5,31–33].

We determined that prolonged use of broad-spectrum antimicrobial drugs, invasive mechanical ventilation, and vasopressor treatment were independent risk factors for A. baumannii isolation from ventilated patients, which is consistent with previous studies^[5,31–33]. A previous study reported that cardiovascular organ failure was an independent risk factor associated with A. baumannii bloodstream infection^[34]. Of note, model 1 of our study showed that cardiovascular disease also was an independent risk factor for A. baumannii isolation. Another population-based study reported that patients with chronic heart failure had a markedly increased risk for hospitalization with pneumonia^[35], indicating a possible correlation between cardiovascular diseases and pneumonia. However, the specific underlying mechanisms by which cardiovascular disease promotes A. baumannii isolation remain unknown.

In critically ill patients undergoing mechanical ventilation, aerosol inhalation is a common intervention for treating various pulmonary diseases. An international survey demonstrated that 99% of 611 ICUs from 70 countries reported using aerosol therapy during mechanical ventilation, including noninvasive ventilation, and the most frequently delivered drugs were bronchodilators and steroids^[36]. A web-based survey involving 447 hospitals in mainland China recorded a high proportion of aerosol therapy in both invasive (90.8%) and noninvasive (91.3%) mechanical ventilation; bronchodilators (64.8%) and topical corticosteroids (43.4%) were the most commonly used drugs^[9]. Aerosol inhalation is aimed at reversing bronchoconstriction, decreasing the work of breathing, relieving dyspnea, modifying the inflammatory response^[19,37], ameliorating lung injury^[38], and reducing the rate of exacerbation in both asthma and COPD. However, in patients with COPD and asthma, inhaled corticosteroids are associated with an increased risk for upper respiratory tract infections^[39,40], pneumonia, and lower respiratory tract infections^[15,41].

Ventilated patients are already vulnerable to pneumonia. Therefore, evaluating whether commonly used aerosol therapy increases the risk for nosocomial pneumonia is crucial, especially when inhaled with glucocorticoids. Because A. baumanni–related pneumonia is associated with severe illness and death, we chose A. baumannii isolation as an outcome and explored its relationship with glucocorticoid aerosol therapy. Our study showed that glucocorticoid aerosol therapy was an independent risk factor for A. baumannii isolation from patients on ventilators. Compared with no aerosol inhalation, glucocorticoid aerosol inhalation increased the risk for A. baumannii by ≈1.5 times. Although further analysis revealed that glucocorticoid aerosol was not directly associated with 30-day mortality, it still might contribute to poor clinical prognosis due to its effect on A. baumannii isolation. As we described, A. baumannii, especially MDR A. baumannii pneumonia, is well recognized as a risk factor for death. In this study, we also found A. baumannii isolation was an independent risk factor for 30-day mortality in patients receiving invasive mechanical ventilation. Because glucocorticoid aerosol heightened the likelihood of acquiring A. baumannii, it might exert a secondary effect, death among A. baumannii–infected patients. Thus, further investigation in a much larger patient population could describe a downstream mortality effect of glucocorticoid aerosol therapy.

When we included glucocorticoid aerosol indications in multivariate analysis, we found COPD and asthma and possible ARDS were independent risk factors for A. baumannii isolation. Because these structural or underlying lung diseases and severe acute lung injury necessitate longer duration of mechanical ventilation, our results were compatible with previously described risk factors for A. baumannii infection. In contrast to glucocorticoid aerosol, we did not detect an association between aerosol inhalation without glucocorticoid and A. baumannii. Because both therapies generate aerosols, our previous concern that aerosols were a source of A. baumannii acquisition might not be reasonable.

Reasons why glucocorticoids increase the risk for A. baumannii isolation remain elusive. Previously considered sterile, healthy lungs harbor complex and dynamic microbiota communities^[42]. Pulmonary diseases, such as COPD^[43], asthma^[44], lung cancer^[45], and ARDS^[46], cause considerable alteration of lung microbiota. Pneumonia pathogenesis involves an abrupt and emergent disruption in the complex homeostasis of the lung microbial ecosystem^[47]. A recent study reported that inhaled corticosteroids altered the lung microbiota in both COPD patients and mouse models and impaired bacterial control in models with Streptococcus pneumonia infection^[14]. Antimicrobial peptides, also known as host defense peptides, are short and generally positively charged peptides that participate in the regulation of the host’s antibacterial actions and immune defense^[48]. Another study observed that cathelicidin, an antimicrobial peptide, was impaired by inhaled corticosteroids among COPD patients by increasing protease cathepsin D, thereby promoting the proliferation of Streptococcus^[14]. In a bacterial 16S rRNA gene sequencing and host transcriptomic analysis, another study reported that as COPD severity increased, the airway microbiome becomes associated with decreased abundance of Prevotella bacteria in concert with downregulation of genes promoting epithelial defense associated with inhaled corticosteroid use^[49]. Evidence also suggests that in asthma, inhaled corticosteroids can alter the relative abundance of genera in airway microbiome^[50]. Therefore, inhaled corticosteroids play a primary role in lung microbiota disruption and host-defense suppression, which could explain why glucocorticoid aerosol contributed to the increased risk for A. baumannii isolation from patients on ventilators.

Clinicians should individualize patient care and manage treatments on the basis of subgroup analysis results. Our results showed that, in most subpopulations, regardless of the presence or absence of the prespecified characteristics, patients were at higher risk for A. baumannii isolation when receiving glucocorticoid aerosol. Based on our findings, we recommend that intensive care teams more carefully consider the risk of widespread usage of glucocorticoid aerosol in patients with invasive mechanical ventilation. Because glucocorticoid aerosol therapy had a much greater favorable effect on A. baumannii isolation in the subgroup of patients on vasopressors for ≥3 days, clinicians should be particularly cautious about giving glucocorticoid aerosol to patients on long vasopressor treatments. Because patients with diabetes, hematologic malignancy, and shorter ICU stays were at relatively lower risk for glucocorticoid aerosol–associated A. baumannii acquisition, glucocorticoid aerosol could be considered when appropriate for indications in these patients. Altogether, our study suggests ICU teams need to identify the specific patient subgroups that will truly benefit from glucocorticoid aerosol therapy rather than more generalized administration. Limiting glucocorticoid aerosol use might be considered as part of existing antimicrobial stewardship bundles. In addition, defining the duration of glucocorticoid aerosol therapy might help maximize benefit while reducing associated risk. Interventional studies exploring the effects of different glucocorticoid aerosol therapy durations on the occurrence of various types of nosocomial pneumonia are needed.

The first limitation of our study is that, because it was an observational study, confounders that might influence the effect of glucocorticoid aerosol on A. baumannii isolation remain, even after adjusting by subgroup analysis, propensity score matching analysis, and multivariable Cox regression. A similar situation exists for the analysis of risk factors for death. Second, a potential time bias remains, in which patients might have been exposed to risk factors before study enrollment. However, we have excluded patients with prolonged exposure to several well-known risk factors to minimize the possible effect of time bias. Third, exclusion of certain cases might pose a potential selection bias, including survivorship bias. Future randomized controlled interventional studies are expected to confirm our findings and minimize the effects of confounders and the above biases. Fourth, the study included only 3 ICUs in China and did not focus on pathogens other than A. baumannii. Research involving more centers and more cases could explore whether glucocorticoid aerosol is a common risk factor in the overall nosocomial pneumonia risk.

In conclusion, we found glucocorticoid aerosol therapy was an independent risk factor for A. baumannii isolation from patients receiving invasive mechanical ventilation. Because of high mortality rates associated with A. baumannii–related nosocomial pneumonia, clinicians should carefully consider both the beneficial and harmful effects of glucocorticoid aerosol before administering this therapy.