Occurrence, Risk Factors, and Outcomes of Pulmonary Barotrauma in Critically Ill COVID-19 Patients: A Retrospective Cohort Study

Objective Pulmonary barotrauma has been frequently observed in patients with COVID-19 who present with acute hypoxemic respiratory failure. This study evaluated the prevalence, risk factors, and outcomes of barotrauma in patients with COVID-19 requiring ICU admission. Methods This retrospective cohort study included patients with confirmed COVID-19 who were admitted to an adult ICU between March and December 2020. We compared patients who had barotrauma with those who did not. A multivariable logistic regression analysis was performed to determine the predictors of barotrauma and hospital mortality. Results Of 481 patients in the study cohort, 49 (10.2%, 95% confidence interval: 7.6–13.2%) developed barotrauma on a median of 4 days after ICU admission. Barotrauma manifested as pneumothorax (N = 21), pneumomediastinum (N = 25), and subcutaneous emphysema (N = 25) with frequent overlap. Chronic comorbidities and inflammatory markers were similar in both patient groups. Barotrauma occurred in 4/132 patients (3.0%) who received noninvasive ventilation without intubation, and in 43/280 patients (15.4%) who received invasive mechanical ventilation. Invasive mechanical ventilation was the only risk factor for barotrauma (odds ratio: 14.558, 95% confidence interval: 1.833–115.601). Patients with barotrauma had higher hospital mortality (69.4% versus 37.0%; p < 0.0001) and longer duration of mechanical ventilation and ICU stay. Barotrauma was an independent predictor of hospital mortality (odds ratio: 2.784, 95% confidence interval: 1.310–5.918). Conclusion s. Barotrauma was common in critical COVID-19, with invasive mechanical ventilation being the most prominent risk factor. Barotrauma was associated with poorer clinical outcomes and was an independent predictor of hospital mortality.

Studies on the epidemiology of barotrauma in critical COVID-19 patients remain uncommon. Its risk factors are not well characterized, especially since many of the published studies were performed in patients receiving IMV, whereas barotrauma has been observed in other patients with COVID-19 [5]. Tis study evaluated the prevalence of barotrauma, risk factors, and outcomes of pulmonary barotrauma in COVID-19 patients admitted to the ICU because of acute hypoxemic respiratory failure.

Setting and Patients.
Tis retrospective cohort study was conducted in King Abdulaziz Medical City in Riyadh, Saudi Arabia, a tertiary-care center with more than 1000-bed capacity. Its Intensive Care Department had seven diferent ICUs, four of which were designated as COVID-19 units during the pandemic [10]. Te ICUs operated as closed units with 24-hour per day, 7-days per week on-site coverage by board-certifed intensivists [10]. In this study, we included patients who were older than 14 years (the cutof age for admission to an adult ward/ICU in Saudi Arabia), confrmed to have COVID-19, and admitted to the adult ICU between March 1 and December 31, 2020, because of acute hypoxemic respiratory failure that was treated by any form of oxygen therapy. A confrmed COVID-19 case was defned as one with a clinical presentation consistent with COVID-19 and detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA in a respiratory specimen by a real-time reverse transcription polymerase chain reaction. Critical COVID-19 was defned as having acute respiratory failure, septic shock and/or multiple organ dysfunction [11]. Patients who were transferred from other institutions with a known history of pulmonary barotrauma and those who did not undergo a chest X-ray or computed tomography were excluded. Te study was approved by the Institutional Review Board of the Ministry of National Guard Health Afairs, Riyadh, Saudi Arabia.

Data Collection.
All patients' clinical data and information were retrieved from the electronic health record system of the hospital (BESTCare). Data included sociodemographic factors such as age, gender, body mass index (BMI), chronic comorbidities (diabetes, hypertension, chronic obstructive pulmonary disease, asthma, previous pneumothorax, interstitial lung diseases), and smoking. Other collected data included chest X-ray and computed tomography fndings (location of infltrates, presence of subcutaneous emphysema and location, presence of pneumomediastinum, presence of pneumothorax and location) for the frst 10 days in ICU, laboratory results (including white blood cell counts, neutrophil and lymphocyte counts, infammatory markers, ferritin levels, and D-dimer), treatments such as vasopressors, central venous catheters, renal replacement therapy, corticosteroids and dose, use of different oxygen therapies (conventional oxygen therapy lowfow oxygen devices such as nasal prongs, mask with or without oxygen reservoir, Venturi mask systems), high-fow nasal cannula, NIV and IMV, timing of barotrauma in relationship to IMV, ventilator settings in the frst 24 hours (max tidal volume, max positive end expiratory pressure (PEEP), peak pressure, and plateau pressure) for patients who received IMV, and insertion of chest tube (number and location).
Te primary outcome of the study was hospital mortality. Te secondary outcomes were ICU mortality, duration of mechanical ventilation, need for a tracheostomy, and length of stay in the ICU and hospital.

Statistical Analysis.
Te prevalence of pulmonary barotrauma (with 95% CI) was calculated in the study cohorts as well as in subgroups who received diferent forms of oxygen therapy and mechanical ventilation. Te study patients were categorized into two groups: patients who developed pulmonary barotrauma and those who did not. Te descriptive statistics were presented as frequency and percentage for categorical variables and as mean with standard deviation or median with the frst and third quartiles (Q1, Q3) for continuous variables, depending on the normality of their distribution. Multivariable logistic regression analysis was performed to evaluate the risk factors for pulmonary barotrauma and for hospital mortality. In the models, the independent variables were those with p value <0.25 between groups [12]. To evaluate the risk factors for mortality associated with barotrauma, we compared the characteristics and management of patients with barotrauma who survived and did not survive. Te results of the regression analyses were presented as an OR with a 95% CI. We used Statistical Package for the Social Sciences (version 21) software for statistical analysis. A test was considered signifcant if the p value was <0.05.
Tere were no signifcant diferences in most baseline characteristics, including presence of comorbidities, BMI, and laboratory fndings, between patients with or without barotrauma (Table 1). Tere were no signifcant diferences in the patterns of infltration on CXR between the two groups. IMV was more common in patients who developed barotrauma compared to those who did not; 87.8% of patients with barotrauma were intubated compared with 54.9% of those without barotrauma (p < 0.0001). Te mechanical ventilator settings on the frst day of intubation are described in Table 1. Tese settings of the mechanical ventilator were not diferent between the two groups.
Respiratory bacterial cultures were taken in 222 patients. Teir results are shown in Figure 2. More patients with barotrauma had respiratory cultures than those without barotrauma. Normal respiratory fora and yeast were isolated more often in patients with barotrauma.

Outcomes.
Te median duration of mechanical ventilation for all patients was 11 days (interquartile range: 5, 18 days); 16 days (interquartile range: 9, 30 days) for patients with barotrauma; and 10 days (interquartile range: 5, 18 days; p < 0.001) for those without barotrauma. Tracheostomy was also performed more frequently in patients with barotrauma (p � 0.016). Te lengths of stay in the ICU and hospital were signifcantly longer in patients with barotrauma compared to those without barotrauma (Table 3).
Te overall mortality in the ICU was 30.8% and in the hospital, 40.3%. Te ICU and hospital mortality rates of patients who developed barotrauma were signifcantly higher than those who did not (59.2% versus 27.5% for ICU mortality, p < 0.001, and 69.4% versus 37.0% for hospital mortality, p < 0.001) ( Table 3).
On multivariable logistic regression analysis (Table 4), in which age, comorbid conditions such as obesity, hypertension, diabetes, asthma, chronic obstructive pulmonary disease, interstitial lung disease, heart failure, chronic kidney disease, stroke, cancer hyperlipidemia, arthritis, IMV,   Te characteristics and management of patients with barotrauma who survived and did not survive are shown in Table 5. Nonsurvivors had a higher BMI and lymphocyte count, had less asthma as a comorbidity, and received less high-fow nasal oxygen and more IMV with a higher fraction of inspired oxygen on the frst day of IMV.

Discussion
Our study found that pulmonary barotrauma occurred in 49 out of 481 patients (10.2%) with critical COVID-19 on a median of 4 days after ICU admission; pulmonary barotrauma was more common in patients who required artifcial ventilation, especially IMV; most patients with pulmonary barotrauma received conservative treatment without chest tube insertion, especially in patients who did not have pneumothorax; pulmonary barotrauma was associated with a worse outcome and was an independent risk factor for hospital mortality.
In our study, 49 out of the 481 patients developed barotrauma in the form of pneumothorax, pneumomediastinum, and subcutaneous emphysema. Tis is slightly lower than the rate (15.6%) observed in a recent metaanalysis among critically ill patients with COVID-19 [5]. Te prevalence of pulmonary barotrauma was the lowest in patients who did not require artifcial ventilation (2.9%, 95% CI: 0.4-10.1%), higher in patients who were treated with NIV (without intubation) (3.0%, 95% CI: 0.8-7.6%), and highest in patients who required IMV (15.4%, 95% CI: 11.3-20.1%). Tis was observed in other studies [9]. COVID-19 itself may increase the risk of barotrauma. A systematic review of COVID-19 patients receiving mechanical ventilation (13 observational studies, 1814 patients) found that 14.7% had at least one barotrauma event compared with 6.3% of patients with non-COVID acute respiratory distress syndrome [7]. Data from randomized controlled trials on patients with acute respiratory distress syndrome (2468 patients) showed an incidence rate of barotrauma of 6-8% [13,14].

Critical Care Research and Practice
Te pathophysiology of barotrauma in COVID-19 is not very clear. A study investigated the radiographic patterns of barotrauma in patients with COVID-19 and observed that 41/43 patients (95%) demonstrated concurrent pneumomediastinum and subcutaneous emphysema or pneumomediastinum alone as the initial abnormal air collection [15]. Te investigators concluded that this was consistent with pulmonary interstitial emphysema, where increased intrathoracic pressure causes overinfation of alveoli without adequate expansion of the associated vessel resulting in alveolar rupture and dissection of air into the bronchovascular sheath and then dissection into the mediastinum, pleural space, subcutaneous tissues, and retroperitoneum [15,16].
In the current study, we did not observe any signifcant association of comorbidities with pulmonary barotrauma. In contrast, hypertension and diabetes have been associated with barotrauma [17]. In our study, hemoglobin and lymphocyte count were slightly higher in barotrauma patients on univariate analysis. On multivariate analysis, the studied infammatory markers were not associated with barotrauma. Other studies observed that certain but not all infammatory markers were signifcantly elevated in patients with barotrauma [17,18], with lymphocyte count being the only infammatory marker associated with barotrauma on multivariate logistic regression [17]. We observed no association between NIV (without subsequent IMV) and the development of barotrauma. Similar fndings were seen in another study [17]. IMV was signifcantly associated with pulmonary barotrauma in the multivariable regression analysis in our study. Tis was also observed in another study [18]. Surprisingly, Hamouri et al. noted that IMV was associated with less barotrauma [17]. We also observed no relationship between early ventilator settings, including PEEP, and barotrauma, likely because most of the study patients received lung protective strategies. While Protti et al. showed that a higher PEEP was a risk for barotrauma [19], a survey of 38 Italian hospitals found that pulmonary barotrauma that occurred with ventilatory settings that may be considered nonprotective was relatively uncommon [20]. For example, when the plateau airway pressure was >35 cm H 2 O, 2/113 (2%) patients had barotrauma, and when the tidal volume was >8 ml/kg of ideal body weight and the plateau airway pressure was >30 cm H 2 O, 12/134 (9%) patients had barotrauma [20]. Hence, SARS-CoV-2 itself may play an important role in the pathophysiology of barotrauma. It may be speculated that SARS-CoV-2 infection leads to frail alveoli by direct and/or indirect (infammatory) alveolar injury, thus reducing epithelial-interstitial integrity [21,22]. A rise in transpulmonary pressure, which can be precipitated by severe cough, respiratory distress, patient self-inficted lung injury [23,24], and/or positive pressure ventilation and would not afect normal alveoli, can be beyond the stressstrain threshold for the epithelial-interstitial integrity and lead to the rupture of the frail alveoli and thus interstitial emphysema [22]. Our fnding of a slightly higher rate of barotrauma in patients who received IMV after failure of high-fow nasal oxygen and/or NIV (17.8% versus 9.6% for patients who only received IMV, p � 0.09) suggests that failure of noninvasive ventilatory support may increase the risk of barotrauma, possibly through patient self-inficted lung injury, which may not have been fully mitigated by the noninvasive ventilatory support. Further studies are needed to understand the precise pathophysiology of barotrauma in COVD-19 and the role of the diferent infammatory mediators.
We also found that patients with barotrauma had more respiratory bacterial cultures, likely because most received IMV, which would facilitate taking a deep tracheal aspirate for culture. Normal respiratory fora grew more often in patients with barotrauma. Whether bacterial coinfection or superinfection increases the risk of barotrauma in patients with COVID-19 is not clear and requires further analysis.
In the current study, chest tube insertion was performed in only 21/49 patients (42.9%) with barotrauma. Tis might be in part due to the occurrence of isolated pneumomediastinum or subcutaneous emphysema without pneumothorax. A study reported that 39/51 patients (76.5%) with pulmonary barotrauma were treated with a chest tube [17].
Barotrauma was associated with worse outcomes in our study, including a longer duration of IMV, a longer stay in the ICU, and higher mortality. Tese fndings were also observed in other studies [18,25]. Additionally, the rates of ICU and hospital mortality were signifcantly     [5,18,25]. Whether the increased mortality was solely due to barotrauma or that barotrauma indicated a more severe respiratory illness is not clear.
Te results of this study should be interpreted based on its strengths and limitations. Te strengths include the relatively large sample size and the inclusion of patients receiving diferent forms of oxygen therapy. Te limitations include the retrospective design at a single center, which will limit its generalizability. Te prevalence of barotrauma was below what was assumed for sample size calculation, which reduced the study's power to detect signifcant risk factors for barotrauma. Moreover, the presence of barotrauma was only assessed during the frst 10 days of the ICU stay, and so more events may have occurred later; therefore, the prevalence may have been underestimated. However, we believe that most barotrauma occurs early during critical COVID-19. Other limitations include the unavailability of the levels of important infammatory markers, such as interleukins, and the lack of assessment for acute respiratory distress syndrome and of periodic follow-ups. We should note that the association between barotrauma and IMV does not imply causality. It might be related to an unmeasured confounder and may be related to the severity of the respiratory disease leading to barotrauma, requirement of mechanical ventilation, and increased mortality at the same time.

Conclusions
Barotrauma was common in patients with critical COVID-19, especially in those receiving IMV. Barotrauma was associated with worse outcomes, including mortality. Whether barotrauma itself worsens outcomes or whether its occurrence is a marker of more severe illness, it remains unclear. Caution during intubation, by avoiding aggressive Ambu bagging, and the implementation of lung protective strategies during NIV and IMV are warranted when caring for patients with COVID-19 and severe hypoxemic respiratory failure.

Data Availability
Te data used to support the fndings in this study are available from the corresponding author on request. Te data are not publicly available due to institutional policies of maintaining the confdentiality of patient data.

Ethical Approval
Tis study was approved by the Institutional Review Board of the Ministry of National Guard Health Afairs, Riyadh, Saudi Arabia (IRBC/0981/21, 20 May 2021).