The Prevalence and the Impacts of Pulmonary Bacterial Coinfections and Secondary Bacterial Pneumonia in Patients with Severe Inuenza Pneumonitis

Background: This study investigated the prevalence and clinical outcomes of pulmonary bacterial coinfections and secondary bacterial pneumonia in patients with severe inuenza pneumonitis. The causative pathogens and their clinical impacts were analyzed. Methods: We retrospectively analyzed the data of adult patients with severe inuenza pneumonitis admitted to our medical ICUs from January 2014 to May 2018. Bacterial confection (in rst 48 h) and secondary bacterial pneumonia (from 48 h to 1 week) were conrmed by chest radiographs and positive ndings in the respiratory specimen obtained from lower airway. The risk factors of pulmonary coinfection were evaluated. The outcomes of patients with or without pulmonary coninfection or secondary bacterial pneumonia were also analyzed. Results: We identied 117 critically ill patients with laboratory-conrmed inuenza pneumonitis admitted to the medical ICUs. Klebsiella pneumoniae (31.4%) and Staphylococcus aureus (22.8%) were the most commonly identied bacteria in patients with bacterial coinfection. A high proportion of methicillin-resistant Staphylococcus aureus was noted. Liver cirrhosis and diabetes mellitus were the independent risk factors for bacterial coinfection. Acinetobacter baumannii (28%) and S. aureus (25%) were the most often identied bacteria in patients with secondary bacterial pneumonia. Patients with secondary bacterial pneumonia had longer period of mechanical ventilation, longer ICU stay and hospital stay, and higher mortality. Conclusions: Bacterial coinfection or secondary infection in patients with severe inuenza pneumonitis were associated with higher rates of morbidity and mortality in ICU patients. Earlier diagnosis and appropriate therapy, especially in patients with liver cirrhosis and diabetes mellitus, should be cautiously considered.


Introduction
In uenza, including the most common causes of human in uenza A subtypes H1N1, H3N2, and in uenza B, is an acute viral respiratory infection that causes widespread annual epidemics infecting up to 20% of the population and resulting in signi cant morbidity and mortality. [1,2] Most of the affected people have mild illness comprising a sudden onset of fever, sore throat, cough, runny rose, headache, myalgia, and malaise. [2] However, in uenza can also cause severe illness or even death, especially in high-risk individuals such as elderly people, those with certain comorbid chronic diseases, and immunocompromised patients. [3,4] The World Health Organization estimates that global in uenza epidemics result in 3-5 million cases of severe illness and 290,000-650,000 deaths annually. [5,6] Remarkably, from April 12, 2009, to April 10, 2010, there were 60.8 million cases and 12,469 deaths in the United States due to swine-origin in uenza A (pandemic 2009 A/H1N1). [7] In uenza epidemics cause morbidity and mortality from direct viral effects, coinfection of bacterial pneumonia, and secondary bacterial complications. Evidence indicating in uenza predisposing to bacterial coinfections has been observed in seasonal in uenza epidemics, past pandemics, pathology studies, and animal models. [8][9][10][11][12] In uenza can cause cell damage and death within the host's airway, which leads to an upregulation in the production of toxins. This phenomenon generates an in ammatory response, causing the release of cytokines and chemokines, more commonly referred to as a cytokine storm. This storm is a variable that promotes damage to pulmonary tissue, leaving the pulmonary tissue vulnerable to infection. Moreover, bacterial pathogens in the lower respiratory tract may cause in uenza infection. [13] In previous published studies (between 2009 and 2012), the most important bacterial coinfections during an in uenza epidemic were Streptococcus pneumoniae and Staphylococcus aureus. [14,15] However, empiric antibiotics were not widely available to cover those bacterial pathogens of both coinfection and secondary bacterial pneumonia, and it is believed that it led to higher mortality rates in critical patients. Inappropriate initial antibiotics for pneumonia are usually associated to extended intensive care unit stay and are linked with an increased risk of mortality. [16,17] Therefore, in the present study, we collected and analyzed the prevalence data of bacterial coinfection (during the rst 48 h of hospitalization) and secondary bacterial pneumonia infection (after 48 h to 1 week) of patients with laboratory-con rmed in uenza pneumonitis admitted to our medical intensive care units (ICUs). Our aims are to determine the distribution of bacterial coinfections and secondary bacterial infections in patients with severe in uenza in the ICUs and to identify the most common bacterial species to provide more appropriate therapy. The impacts of bacterial coinfections and secondary infections in severe in uenza patients were also analyzed.

Patients
We retrospectively evaluated the data of patients (aged > 20 years) with laboratory-con rmed in uenza infection admitted to the medical ICUs between January 2014 and May 2018 in the Chiayi Chang-Gung Memorial Hospital, a 1369-bed primary care and referral regional teaching hospital that included two adult medical ICUs, in southern Taiwan. The inclusion criteria were (1) patients with con rmed in uenza pneumonitis accompanied by complications of multiorgan failure (altered function in two or more organ systems during an acute illness), unstable blood pressure (systolic blood pressure < 90 mmHg or mean arterial pressure < 60 mmHg), or acute respiratory failure requiring mechanical ventilation and ICU admission and (2) all these patients should have received antiviral treatment with either oseltamivir or peramivir. These patients were categorized into (1) cases with bacterial coinfection among severe cases of in uenza requiring medical ICU admission, (2) severe cases of in uenza requiring medical ICU admission without bacterial coinfection, (3) cases of secondary bacterial infection among severe cases of in uenza requiring medical ICU admission, and (4) severe cases of in uenza requiring medical ICU admission without secondary bacterial infection. We also collected the detailed clinical and physiological data using Acute Physiology and Chronic Health Evaluation (APACHE) II score. Patients were excluded from the study if they were aged < 20 years or had mild u-like symptoms and treated as outpatients or in general wards.

Diagnosis of in uenza infection
In uenza virus A or B infection was con rmed by a positive nding in the respiratory specimen (nasopharyngeal swab and/or pharyngeal swab) using the rapid in uenza diagnostic test (Formosa One Sure Flu A/B Rapid Test Kit) or reverse transcriptase-polymerase chain reaction (QiAamp Viral RNA Mini Kit; TAIGEN Bioscience Corporation, Taiwan).

Diagnosis of pulmonary bacterial coinfection
Bacterial coinfection was con rmed by a positive nding on chest radiograph (in ltration and/or consolidation change) along with a positive culture in the respiratory specimen (obtained from sputum, endotracheal aspiration, bronchial washing, or bronchoscopic bronchoalveolar lavage samples) in the rst 48 h of hospitalization.

Diagnosis of secondary bacterial pneumonia
Secondary bacterial infection was con rmed using clinical symptoms (for example, fever, cough, increased sputum production, and tachypnea) and a positive nding on chest radiograph (in ltration and/or consolidation change) along with a positive culture in the respiratory specimen (obtained from sputum, endotracheal aspiration, bronchial washing, or bronchoscopic bronchoalveolar lavage samples) between 48 h to 1 week after hospital admission.

Data collection and de nitions
A standardized form was designed for clinical data collection. The data were primarily retrieved from the hospital's electronic medical records and were supplemented by a secondary manual search. The following data were collected: age, gender, body mass index, associated medical conditions, APACHE II score during rst 24 h admitted to the ICU, results of laboratory tests, results of bacterial specimens from sputum, endotracheal aspiration, bronchial washing, or bronchoalveolar lavage, in-hospital complications, days of invasive mechanical ventilation, days of ICU stay, days of hospital stay, and overall mortality.

Statistical analysis
The clinical characteristics, underlying medical diseases, and laboratory ndings as well as complications of survivors and non-survivors were compared using the chi-square test for categorical variables and the independent sample t-test for numerical variables. Differences were considered to be statistically signi cant at p < 0.05. Signi cant variables in the univariate analyses were entered into a multivariate logistic regression model to identify independent predictors. Data were entered and analyzed using the Statistical Package for the Social Sciences statistical software (version 26.0; SPSS Inc., Chicago, IL, USA). days, respectively. The most commonly associated underlying disease was hypertension, followed by diabetes mellitus, chronic obstructive pulmonary disease (COPD) and liver cirrhosis. Table 1 summarizes the clinical characteristics and outcomes of the included patients.

Impacts of pulmonary bacterial coinfection and secondary bacterial infection
Compared with patients without coinfection, those diagnosed with severe in uenza coinfected along with bacterial pneumonia had a signi cantly higher severity (higher APACHE II scores) and longer ICU stay.
Although not signi cantly different, patients with bacterial coinfection were of older age and had low body mass index (BMI). There was no signi cant difference in the duration of mechanical ventilation, hospital stay and mortality between patients with or without bacterial coinfections (Table 3). In addition, although there were no signi cant differences in age and BMI, severe in uenza patients with secondary bacterial pneumonia had signi cantly higher APACHE II score, longer ICU stay and hospital stay. Numerically longer mechanical ventilation use and higher mortality (21.4 vs. 9.6 days and 31.8% vs. 18.9%, respectively) were also found in patients with secondary bacterial infection, although the difference did not reach to a statistical signi cance (Table 4).

Discussion
The results of our study showed that pulmonary bacterial coinfection occurred in 27% (32 /117) of patients with severe in uenza pneumonitis. Those with bacterial coinfection had higher severity and longer ICU stay. Diabetes mellitus and liver cirrhosis were the independent risk factors for bacterial coinfection in patients with severe in uenza pneumonitis. We also demonstrated that 19% (22/117) of patients with severe in uenza pneumonitis had secondary bacterial pneumonia during the course of hospitalization. Those with secondary bacterial pneumonia had higher severity and longer ICU and hospital stay.
Unlike previous studies that reported that S. pneumoniae was the most frequent cause of bacterial coinfection,[18-20] the most common coinfection pathogens identi ed in our study were K. pneumoniae (31.4%), followed by S. aureus (22.8%), S. pneumoniae (11.4%), and P. aeruginosa (11.4%). This result was similar to a previous report from Taiwan. [21] The higher percentage of K. pneumoniae coinfection could be related to the higher rates of chronic diseases such as diabetes mellitus, COPD or liver cirrhosis in our patients. A previous study reported signi cant increases in the incidence of MRSA infections in the past decade, particularly community-associated MRSA (CA-MRSA). [22] In the pandemic of coronavirus disease 2019 (COVID-19), MRSA was also the most common and important bacteria in coinfection. [23] In our study, 22.8% of the identi ed isolates were S. aureus among patients with coinfection, and 75% of these isolates were MRSA. This nding of coinfection pathogens supports the Infectious Disease Society of America recommendations for broad-spectrum antibiotic coverage for in uenza-related pneumonia, particularly to cover CA-MRSA in patients with in uenza-related pneumonia. [24] In case of secondary bacterial pneumonia, A. baumannii (28%) was the most often identi ed bacterium, followed by S. aureus (25%) and K. pneumoniae (17%). S. pneumoniae was not detected among secondary infection. Meanwhile, 66.7% of the S. aureus isolated were MRSA. Other drug-resistant specimens were also detected in patients with secondary infection (MDRAB 13%, MHT-positive K. pneumoniae 8%, and CRPA strain 8%). Those drug-resistant organisms should be covered by empiric antibiotic therapy if nosocomial pneumonia was suspected in these patients. Previous studies have described that cellular immune response was decreased, whereas antibody production was intact, in patients with advanced liver cirrhosis, and IFN-γ response was relatively poor [29]. In addition, in uenza virus itself can cause hepatitis, and in uenza-induced toxic metabolites and proin ammatory cytokines such as TNF-α, IL-1, and IL-6 might contribute to hepatic damage [30,31].
On the other hand, neutrophil chemotaxis and adherence to intracellular bactericidal activity, opsonization, vascular endothelium, phagocytosis, and cell-mediated immunity are all deteriorated in diabetic patients with hyperglycemia. [32,33] Therefore, patients with liver cirrhosis and diabetes mellitus can have an immunosuppression status due to viral and bacterial infections, resulting in adverse complications and increasing the overall mortality. Therefore, in uenza vaccine should be recommended to cirrhotic and diabetic patients.
These 117 critically ill patients with laboratory-con rmed in uenza pneumonitis and were admitted to the medical ICU in the present study, had the 28-day mortality and in-hospital mortality rates being 15.4% and 21.3%, respectively. This result was similar to that of a previous study involving 444 adult hospitalized patients with in uenza in the United States. [34] A mortality rate of 20.6% had also been reported in a prospective multicenter observational cohort study of 2,059 patients admitted to ICUs for in uenza infection. [35] Previous studies have mentioned that community-acquired bacterial coinfection can predict severity and mortality in patients admitted with in uenza-associated pneumonia.
[36] Our study demonstrated that compared with patients without coinfection, those diagnosed with bacterial coinfection with had a signi cantly higher severity (higher APACHE II score) and longer ICU stay, but there was no signi cant difference in the duration of mechanical ventilation, length of hospital stay or mortality rate. The rate of mortality was similar to a previous report, [37] which may be related to the increasing recognition of in uenza, so that the intensivist could provide more optimal care in the ICU.
A higher mortality was reported in patients with nosocomial infections among those hospitalized with severe in uenza A.
[38] Our research found the ICU stay and hospital stay were signi cantly longer in this patient group. Longer period of mechanical ventilation and higher mortality (21.4 vs. 9.6 days, P = 0.055 and 31.8% vs. 18.9%, P = 0.183, respectively) were also found in patients with secondary bacterial infection, although the difference had no statistical signi cance. This study provides data to clinicians to recognize that secondary bacterial infection is a contributor for increased morbidity and mortality in patients with severe in uenza pneumonitis. The statistical insigni cance could be related to the small case numbers in this study.
This study has some potential limitations. First, given the retrospective nature of the study, we did not collect information on previous in uenza or pneumococcal vaccination. The impact of previous in uenza vaccination could not be evaluated in this study. Second, due to the small case number, the statistical power might not be su cient to assess the risk factors for mortality within this subpopulation. Third, the study population consisted of adult patients; therefore, the results cannot be generalized to pediatric patients. The strengths of this study include a detailed description of medical disease information at presentation and the entire bacterial specimen data of critically ill patients with in uenza. We have also emphasized the key factors associated with bacterial coinfection and secondary bacterial infection in critically ill in uenza patients and provided data to physicians to select the appropriate empiric antibiotics to minimize patient morbidity and mortality.

Conclusions
Bacterial coinfection and secondary infection were important complications in patients with severe in uenza in the ICU, with K. pneumoniae being the most common coinfection pathogen and A. baumannii being the most commonly identi ed pathogen in secondary infection. A high prevalence of MRSA in patients with coinfection and secondary infection was also noted. Liver cirrhosis and diabetes mellitus were independent risk factors for bacterial coinfection. Moreover, bacterial secondary infection appeared to be associated with higher rates of morbidity and mortality in ICU patients. Prevention and treatment of bacterial coinfection and secondary infection should be an integral component of a pandemic plan, particularly for patients with severe in uenza infection requiring ICU care.
Abbreviations ICU: intensive care unit.
BMI: Body mass index.