Introduction
A clinical conundrum in viral CAP is diagnosis and management of potential bacterial co-infection. This question becomes crucial when determining the best therapy and weighing the necessity of antibiotics for bacterial infection against the risks of unnecessary antimicrobial exposure. There is substantial variability in reported rates of bacterial co-infection among patients with confirmed viral CAP, and co-infection rates may differ based on viral pathogens or severity of illness. If a patient truly has an infection solely due to a virus, then antibiotics are unnecessary and can only be associated with patient harm.1–3 However, there is concern over potentially missing cases of bacterial co-infection and withholding antibiotics inappropriately. The question of whether clinical presentation, rapid diagnostic tests and/or biomarkers may be utilized to rule in or out bacterial co-infection in the setting of confirmed viral CAP remains unanswered. This review aims to offer practical insights into the current understanding of this clinical dilemma in hospitalized adults.
Epidemiology
CAP is one of the most common causes for hospitalization. In the United States alone, it is estimated that CAP accounts for > 1.5 million hospitalizations and 100,000 inpatient deaths annually.4 Estimations of the burden of CAP secondary to viral pathogens vary, influenced by methodology, timing of diagnosis, patient population and seasonality. While one meta-analysis estimated that viral infection was observed in 24.5% of patients hospitalized with CAP, some studies approximate rates as high as 40%.5–7 Considering these estimations in combination with epidemiologic data reported by Ramirez et al, viral CAP results in approximately 368,000 to 600,000 hospitalizations in the United States annually. It should be noted this was prior to the SARS-CoV-2 (COVID-19) global pandemic.4 COVID-19 further exacerbated the already significant burden of illness caused by viral pneumonia.8
Due to the lack of standardized viral testing protocols across hospitals and the seasonal nature of viral illnesses, the overall burden of viral CAP with or without bacterial co-infection remains uncertain. Studies examining bacterial co-infection in hospitalized patients with viral CAP have yielded variable findings but have helped shed light on bacterial co-infection rates, including in specific viral illnesses like influenza, SARS-CoV-2, and respiratory syncytial virus (RSV).
Early utilization of polymerase chain reaction (PCR) testing data from 2009 at a single center analyzed respiratory secretions and demonstrated 14% of 242 patients with CAP had mixed viral-bacterial etiologies.9 Another study by Falsey et al in 2013 of 348 patients hospitalized with viral respiratory infections utilized a variety of microbiologic tests and/or a serum procalcitonin (PCT) >0.25 ng/mL, and found evidence of bacterial co-infection in 18.4% of cases. Interestingly, in this same study, overall antibiotic prescription rates did not differ significantly between patients with and without evidence of bacterial co-infection, with approximately 90% of both groups receiving antibiotics. However, shorter durations of antibiotic therapy were observed in the viral alone infection compared to mixed viral-bacterial group (4.2 ± 4.6 days vs 6.2 ± 9 days, p=0.04).6 A study in 2016 by Voiriot et al aimed to stratify CAP into bacterial, viral, and mixed (viral-bacterial co-infection), or no etiology using multimodal screening with traditional bacterial cultures, Streptococcus pneumoniae and Legionella pneumophilia urinary antigens, and PCR testing of respiratory secretions. This study found 26% of patients presented with viral pneumonia with bacterial co-infection.10 Despite variable observed rates, none of these studies found that bacterial co-infection was present in the majority of viral CAP cases.
Influenza
Post-influenza bacterial super-infection is well documented, with Staphylococcus aureus being the most frequently observed etiology.11 Retrospective studies have reported inconsistent rates of bacterial co-infection, ranging from 5.1% to 50%.12–18 It appears that in the setting of influenza, bacterial co-infection rates may vary based on time from admission and severity of illness. Additionally, methodology of diagnosing co-infection may differ across studies. For instance, Hughes et al, who reported a 5.1% infection rate, retrospectively evaluated microbiologic samples including respiratory and blood cultures, urinary antigens and PCR testing if available at any point within the hospital admission.14 In contrast, Pandey et al who reported a higher 50% co-infection rate, retrospectively evaluated only critically-ill patients for co-infection using blood or respiratory tract cultures.15 It should be noted that multiple studies have observed higher co-infection rates in critically-ill patients, which are associated with increased mortality.15–17
SARS-CoV-2
Though significant variability has been reported, the largest currently available meta-analysis estimates bacterial co-infection rates at 5.3% within 48 hours of hospital admission and 18.5% thereafter, likely representing hospital-acquired bacterial infection.19 A large retrospective cohort of 1,050 COVID-19 patients aimed to quantify bacterial co-infection within 48 hours of intubation confirmed via bacterial growth from respiratory or blood culture, or positive S. pneumoniae or L. pneumophilia urinary antigens. Evidence of bacterial co-infection was present in 9.7% of these intubated patients.18 Three retrospective studies in critically-ill patients reported growth from blood or respiratory cultures within 48 hours of hospital admission in 7.6 - 23% of cases.12,15,20 Shah et al retrospectively reviewed respiratory and blood cultures within 7 days of admission and observed an increased probability of being admitted to the ICU if bacterial co-infection was documented (60% vs 28.5%, p<0.0001).21 These studies coupled with meta-analysis data suggests that the probability of bacterial co-infection appears to be higher patients who are critically-ill or hospitalized for longer durations.
Respiratory syncytial virus
In the adult population, RSV bacterial co-infection rates are less frequently studied and are estimated to be between 12.1%-29%.13,22,23 Similar to influenza, co-infection has been associated with increased mortality relative to RSV alone.12,21 Studies have reported antibiotic prescribing rates over 75% in hospitalized populations.13,24 These data, similar to reports in COVID-19 infection, suggest a mismatch in antibiotic prescribing and actual bacterial co-infection rates.
Based on available evidence, it is challenging to make precise estimates of how frequently viral CAP with bacterial co-infection occurs. However, the observation of increased severity of illness and death in co-infected patients highlights the importance of recognizing this mixed disease state when it occurs and managing it appropriately. Nonetheless, it also appears that the reported antibiotic prescribing rates are much higher than co-infection rates, highlighting a potential target for antimicrobial stewardship and judicious prescribing (Table 1).
Diagnostic Considerations
Pneumonia is an infection of pulmonary parenchyma. The clinical diagnosis of pneumonia is based on a history of cough, sputum production, fever, dyspnea, and the identification of new infiltrates on chest radiography.27 When evaluating patients with confirmed viral CAP, it is crucial to use clinical assessments and diagnostic data when evaluating for potential bacterial co-infection.
Radiographic Imaging
Radiographic patterns of respiratory viral pneumonia can vary significantly. A study assessing pneumonias secondary to influenza, adenovirus, RSV, parainfluenza, and bacterial pneumonia found substantial overlap on computerized tomography imaging.28 Chest radiographs may show unilateral or bilateral consolidations, ground-glass or nodular opacities, or may lack specific findings altogether.29 Bacterial co-infection on radiography is more likely to show classic findings of bacterial pneumonia, such as focal consolidations, particularly lobar pneumonia, bronchopneumonia and bronchiolitis, cavitation or the development of an empyema or pleural effusions.29,30 Therefore imaging alone, in the absence of other clinical information, is likely insufficient to definitively differentiate between viral and bacterial pneumonias.
Biomarkers (Procalcitonin, C-Reactive Protein)
Biomarkers and inflammatory markers including PCT and C-reactive protein (CRP) have been used in CAP to assess treatment response, aid in evaluating risk of bacterial infection and adjust treatment plans. Generally, PCT and CRP are likely to be elevated in mixed viral-bacterial co-infection compared to viral CAP alone.31 PCT values <0.25 ng/mL is consistent the absence of bacterial co-infection, and a level >0.5 ng/mL is suggestive of possible bacterial co-infection in viral CAP.32,33
Increased CRP levels have also been observed in patients with confirmed viral-bacterial co-infection. In a study by Ahn et al., patients with H1N1 influenza were retrospectively evaluated for bacterial co-infection using sputum Gram stains, tracheobronchial aspirate cultures, blood cultures and Legionella spp. urinary antigens. Authors found a CRP >10 mg/dL was suggestive of bacterial co-infection, with a positive predictive value of 41% and a negative predictive value of 54%. In contrast, PCT procalcitonin had an improved negative predictive value of 84%.31 In critically-ill patients with COVID-19, increased CRP and PCT levels were observed in cases of bacterial co-infection, although neither marker was deemed to have excellent positive predictive value.20 Some experts have suggested a CRP cutoff of 15 mg/dL as being potentially predictive of bacterial infection.34,35
Additional biomarkers which are known to be elevated in pneumonia, including interleukin (IL)-5, IL-6, human interferon induced protein 10, and tumor necrosis factor alpha have been investigated for their utility in discriminating viral and bacterial etiologies. However, none of these biomarkers showed statistically significant differences when comparing patients with bacterial only, mixed etiology and viral only infection. The authors concluded that PCT was significantly higher in mixed etiology and bacterial infection relative to viral only infection.34
Bacteriology and Cultures
Bacterial culture growth is the most definitive microbiologic method for diagnosing bacterial infection. When evaluating for potential bacterial co-infection in viral CAP, various modalities that have been used in studies, including urinary antigens for S. pneumoniae or Legionella spp, as well as newer technologies such as rapid diagnostic platforms.36,37 However, a limitation with current platforms is their ability to detect a limited number of bacterial pathogens.38
One prospective study in 82 critically-ill patients with COVID-19 utilized a multiplex PCR panel alongside standard respiratory cultures. The authors reported the negative predictive value was 99.9% and the positive predictive value was 66.7% for the PCR panel, suggesting its potential use in ruling out bacterial co-infection. However, given the small sample size and a documented case of a false negative PCR panel, where growth was detected from a respiratory culture, this may result in a bacterial co-infection diagnoses being inappropriately dismissed.39 While these tools do have clinical utility and are likely to detect the most common causes of bacterial pneumonia, growth from culture of a respiratory sample is the most definitive method of confirming bacterial co-infection.
The presence of bacterial growth from culture is contingent upon the ability to collect an adequate, uncontaminated expectorated sputum sample. Alternative sputum sample collection methods, such as bronchial alveolar lavage, are unlikely to be widely used in the setting of CAP, especially less severe cases, due to the invasive nature of the procedure.30,40 It is generally recommended to obtain a respiratory culture prior to administering antibiotics.40 However, data on the utility of sputum cultures after antibiotic exposure is mixed.41–43 Miyashita et al concluded that patients with antibiotic exposure were less likely to have positive sputum cultures, whereas Abers et al found that in patients with bacteremic pneumococcal pneumonia, sputum cultures were still likely to be positive after initial antibiotic exposure. Under ideal circumstances, cultures should always be collected prior to antibiotic exposure but it appears they may still have clinical utility even after antibiotics have been administered.41,43
Clinical Status
Several studies have concluded that mixed viral-bacterial pneumonia is associated with increased severity of infection and worse clinical outcomes compared to viral infection alone. In a large retrospective study spanning 8 influenza seasons from 2010-2018, it was found that in patients with confirmed influenza there was a total co-infection rate of 19.6%. However, in patients who were admitted to the intensive care unit (ICU), the co-infection rate was 51.3% when utilizing respiratory and blood cultures, L. pneumophilia and C. pneumoniae serum IgM and IgG, and respiratory PCR testing.17 Another retrospective study by Bergman et al reported 23% of critically-ill patients with COVID-19 had bacterial co-infection based on growth from blood or respiratory cultures or L. pneumophilia or S. pneumoniae urine antigens.12 This rate is higher than the co-infection rate of 5.3% within 48 hours of admission reported in a recent meta-analysis of patients with COVID-19, but more similar to rates reported >48 hours after hospitalization at 18.4%.19 Additionally, a recent study found that in patients with RSV, bacterial co-infection was a risk factor for inpatient mortality (12.9% vs. 4.9%, p=0.01).22 Finally, a retrospective study in ICU patients with CAP found that patients with viral-bacterial co-infection were more likely to require mechanical intubation relative to patients with a single pathogen.10
It appears in multiple viral respiratory illnesses that bacterial co-infection is associated with increased severity of illness and poor clinical outcomes. This pattern suggests that the pre-test probability of bacterial co-infection is likely higher in critically-ill patients relative to more stable patients with viral CAP. It appears clinical status may help inform suspicion for viral-bacterial co-infection.
Guideline Recommendations and Management Strategies
The appropriate consideration and management of CAP due to possible viral-bacterial co-infection is addressed in both the 2019 Infectious Diseases Society of America (IDSA) CAP and 2023 European Respiratory Society (ERS), European Society of Intensive Care Medicine (ESICM), European Society of Clinical Microbiology and Infectious Diseases (ESCMID) and Latin American Thoracic Association (ALAT) severe CAP guidelines.40,44
The IDSA guidelines advocate for antibiotic use even in the setting of confirmed viral pneumonia at the point of diagnosis, recommending to “initially treat empirically for possible bacterial infection or co-infection” in the context of positive rapid viral testing. The guidelines also emphasize that PCT “alone cannot be used to justify withholding antibiotics from patients with CAP.” While the IDSA guidelines do not specifically address an individual viral disease state besides influenza, they reiterate a recommendation to start antibiotics. In patients with influenza, the guidelines state that there if is “no evidence of a bacterial pathogen (including low procalcitonin level) and early clinical stability, consideration could be given to earlier discontinuation of the antibiotic treatment at 48 to 72 hours.”40
The ERS/ESICM/ESCMID/ALAT severe CAP guidelines note that PCT “may help differentiate co-infections (bacterial) in patients with viral [severe CAP] … Outcomes related to antibiotic stewardship programs would likely benefit from a reduction in antibiotic use, which decreases likelihood of adverse effects. Despite such a decrease in antibiotic use, a negative impact on outcomes would most likely not occur.”44 Both the IDSA and ERS/ESICM/ESCMID/ALT guidelines note that further research is needed in utilizing both rapid viral diagnostic tests and PCT to aid in ruling out co-infections and reducing unnecessary antibiotic exposure.40,44
Widespread antibiotic use has been extensively documented in viral CAP, particularly in the setting of COVID-19 infection, with published prescribing rates as high as 80%, despite a much lower documented co-infection rate.19,26 The IDSA has published separate guidelines on the management of COVID-19 which state “the apparent discordance between bacterial and fungal co-infection in patients with COVID-19 at presentation and the use of antibacterial therapy has potential negative effects, namely in antimicrobial resistance.”45 The National Institutes of Health Coronavirus Disease 2019 (COVID-19) Treatment Guidelines recommend against the use of antibiotics in COVID-19 patients except in “proven or suspected bacterial co-infection.”46 Though both sets of guidelines point out antibiotic overprescribing is a major issue in this patient population, they do not provide specific criteria to guide clinicians on discontinuing or withholding antibiotic therapy.45,46 This highlights the overall challenge of optimal antibiotic prescribing in patients afflicted with viral CAP. The consequences of frequent antibiotic use and need to curb inappropriate exposure is more apparent than ever before. In 2022, the Centers for Disease Control and Prevention (CDC) published a report detailing how the COVID-19 pandemic had reversed previous progress on antimicrobial resistance, noting the alarming 15% rise in multi-drug resistant infections between 2019 and 2020.47 This trend suggests that current antibiotic prescribing in patients with viral CAP is likely contributing to antimicrobial resistance.
Relatively small, observational or quasi-experimental studies have attempted to investigate antibiotic discontinuation in patients with confirmed viral CAP. Moore et al and Heesom et al conducted single-center trials and observed no difference in mortality in COVID-19 patients when antibiotics were discontinued based on a recommendation from antimicrobial stewardship programs using PCT guidance.48,49 Moradi et al conducted a multi-site quasi-experimental study describing a pre and post period when the electronic health record began displaying a recommendation to consider discontinuing antibiotics to prescribers if adult patients had a PCT result < 0.25 ng/mL and a positive viral respiratory PCR within 48 hours of each another. This study found a reduction in mean antibiotic days (5.8 vs 8, p <0.001) without a difference in antibiotic restarts after discontinuation.50 Taken together, these reports demonstrate reduced antibiotic exposure without adverse clinical outcomes when utilizing rapid viral diagnostic testing results and PCT to stop antibiotics.
It is imperative for clinicians to limit antibiotic exposure when possible by carefully assessing for bacterial co-infections utilizing the tools that are available to them. As previously stated, risk of bacterial co-infection appears to be higher in critically-ill patients. However, in the absence of growth from bacterial cultures, other positive bacterial screening tests or elevated PCT, clinicians should carefully consider the risk versus benefit of continuing antibiotics and evaluate if discontinuation is in the patient’s best interest. If clinical status worsens thereafter, antibiotics may always be resumed if deemed appropriate.
Future Directions
Global experts in the management of CAP have called for further research in viral CAP, including the use of PCT and diagnostics to limit antibiotic exposure.40,44 Researchers should prioritize the development of practical and low-resource methodologies for evaluating patients with viral CAP for bacterial co-infections, aiming to demonstrate similar or superior clinical outcomes when antibiotic exposure is limited. Though retrospective data exists in limiting antibiotics in patients with viral CAP, these studies are limited by being single-center and uncontrolled.48–50 There is a need for a standardized clinical and microbiologic approach to assessing risk of bacterial co-infection which should be developed and rigorously tested in a prospective, randomized controlled trial. Ideally, such an approach would be validated in multiple viruses, geographic regions and patient populations, including those who are critically-ill or immunocompromised. If successful in demonstrating improved or similar clinical outcomes compared to standard of care, such an intervention would revolutionize the management of viral CAP. This would enable clinical practice guidelines to provide evidence-based and high-quality guidance for clinicians.
Conclusions
It remains to be definitively determined if and when it is safe and optimal to discontinue or withhold antibiotics in patients with viral CAP. To date, no randomized controlled trials addressing antibiotic prescribing in this clinical scenario have been conducted. While expert opinion has proposed an approach to managing patients with viral CAP, prospective clinical trials are needed given the frequency of admissions and antimicrobial prescribing. Clinicians should tailor their medical decision-making based on individual patient factors including but not limited to the probability of viral-bacterial co-infection, prognostic factors, microbiologic and other lab results and risks of adverse events associated with antibiotic use. Future studies should aim to develop practical, standardized methodology for evaluating patients with viral CAP for risk of bacterial co-infection and provide guides for clinical decision-making including optimal prescribing of antibiotics.
Funding
the authors received no specific funding for this work
Conflicts of interest
None to disclose
Acknowledgements
none