The Burden of Viruses in Pneumonia Associated With Acute Respiratory Failure An Underappreciated Issue

Transcription

1 [ Original Research ] The Burden of Viruses in Pneumonia Associated With Acute Respiratory Failure An Underappreciated Issue Andrew F. Shorr, MD, MPH; Kristen Fisher, MD; Scott T. Micek, PharmD; and Marin H. Kollef, MD BACKGROUND: Pneumonia associated with mechanical ventilation (MV) results in substantial mortality and represents a leading reason for the use of antibiotics. The role of viruses in this setting is unclear. Identifying a viral cause in such instances could facilitate antibiotic stewardship. METHODS: We performed a secondary analysis of a prospective cohort with pneumonia requiring MV. We included both cases occurring in the community and hospital-onset cases and classified patients according to the cause of the pneumonia. The prevalence of viral pathogens represented the primary end point. We identified variables independently associated with isolation of a viral organism as the sole pathogen. RESULTS: The cohort included 364 patients, and a virus was the sole pathogen in 79 cases (21.7%). The most common viruses included rhinovirus/enterovirus (n ¼ 20), influenza A (n ¼ 12), and respiratory syncytial virus (n ¼ 11). The rate of in-hospital death was high (37.2%) and did not differ from that seen in other patients (36.5%). The duration of MV, hospital length of stay, and 30-day readmission rates also did not differ based on the cause of pneumonia. Two variables were independently associated with recovery of a virus: an Acute Physiology and Health Evaluation II score of < 26 (adjusted odds ratio [AOR], 0.51; 95% CI, ; P ¼.027) and stem cell transplantation (SCT) (AOR, 4.39; 95% CI, ; P ¼.001). A sensitivity analysis excluding patients who underwent SCT did not substantially alter our observations. CONCLUSIONS: Viruses represent a major cause of pneumonia in critically ill patients requiring MV. Identifying such subjects presents an opportunity for discontinuing antibiotics. Clinicians should consider systematically evaluating patients with pneumonia requiring MV for viral pathogens. CHEST 2017; -(-):--- KEY WORDS: acute respiratory failure; bacteria; outcomes; pneumonia; virus ABBREVIATIONS: AOR = adjusted OR; CAP = community-acquired pneumonia; HAP = hospital-acquired pneumonia; LOS = length of stay; MV = mechanical ventilation; SCT = stem cell transplantation; VAP = ventilator-associated pneumonia AFFILIATIONS: From the Department of Medicine (Dr Shorr), Pulmonary and Critical Care Medicine Section (Dr Shorr), Medstar Washington Hospital Center, Washington, DC; the CardioPulmonary Associates of St. Lukes Hospital (Dr Fisher), Chesterfield, MO; the St. Louis College of Pharmacy (Dr Micek); and the Department of Medicine (Dr Kollef), Division of Pulmonary and Critical Care Medicine, Barnes Jewish Christian Hospital, St. Louis, MO. FUNDING/SUPPORT: The authors have reported to CHEST that no funding was received for this study. CORRESPONDENCE TO: Andrew F. Shorr, MD, MPH, Medstar Washington Hospital Center, 110 Irving St NW, Washington, DC 20010; andrew.shorr@gmail.com Copyright Ó 2017 American College of Chest Physicians. Published by Elsevier Inc. All rights reserved. DOI: chestjournal.org 1

2 Pneumonia remains both a leading reason for mechanical ventilation (MV) and a potential complication of MV. Hence, the impact of pneumonia in the ICU remains substantial. 1,2 As such, multiple quality efforts focus on reducing the burden of pneumonia in the ICU, regardless of whether the pneumonia results in admission to the ICU or if it represents a consequence of MV. 2 Similarly, pneumonia in the setting of acute respiratory failure (ARF) continues to be a leading reason for antibiotic administration. 3,4 Appropriate and timely antibiotic administration is key to limiting morbidity and mortality in ARF and pneumonia. 5 This nexus between appropriate antibiotic therapy and outcomes has been demonstrated in diverse types of pneumonia, including severe community-acquired pneumonia (CAP), nosocomial pneumonia leading to ARF, and ventilatorassociated pneumonia (VAP). 5 Because of this relationship, intensivists often are quick to prescribe broad-spectrum antimicrobial agents in suspected cases of pneumonia. However, as a consequence, many patients are likely exposed unnecessarily to agents that are not specifically indicated. This overuse of antibiotics contributes to the escalating prevalence of resistant bacteria encountered in health care and further limits the options available to treat many emerging pathogens. One way to improve antibiotic prescribing and to increase rates of antibiotic de-escalation is to identify subjects who do not require further antibiotics because their pneumonic syndrome is caused by a viral pathogen. 6 Properly determining that the cause of a specific pneumonia is a viral pathogen can reassure the clinician that it is safe to discontinue further antibiotics (should no bacterial pathogen simultaneously be identified). In other words, better appreciating the range of pathogens encountered in pneumonia arising in or complicating ARF can foster improved antibiotic stewardship. Similarly, understanding the significance of viruses in ICU-related pneumonia could facilitate the development of both clinical decision aids and diagnostic tools that also might augment antibiotic stewardship. The significance of viral organisms in pneumonia historically was believed to be limited to immunosuppressed and transplant populations. Several contemporary analyses, however, illustrate the significance of viruses generally as important causes of pneumonia. Jain et al, 7 for instance, in an observational study of hospitalized patients with CAP noted that viruses were a more prevalent cause of the infection than were bacteria. Similarly, investigators have implicated viral organisms as a major cause of VAP. 8,9 Hong et al 9 isolated viruses in > 20% of VAP cases. To examine the epidemiology and outcomes associated with viral pneumonia seen in ARF, we conducted a retrospective analysis of pneumonia seen in patients undergoing MV. We hypothesized that viruses were common causes of pneumonia associated with ARF. We sought to describe the prevalence of viral organisms. We further aimed to assess the outcomes and risk factors associated with a viral cause in patients with pneumonia and respiratory failure requiring MV. Methods Study Overview This study was a secondary analysis of a prospective cohort of adult patients diagnosed with pneumonia in the setting of respiratory failure requiring MV during a 1-year period at a single center (January to December 2016). Prior aspects of this analysis, particularly information regarding bacterial causes and appropriate initial therapy, have been described elsewhere. 10 Briefly, the study was conducted between January 1, 2016 and December 31, We included only adults (age $ 18 years) admitted to the hospital for at least 48 hours and evaluated cases of pneumonia, regardless of whether it had a community or hospital onset. All subjects had to undergo MV for at least 24 hours to be included in the study cohort. In other words, we examined suspected cases of CAP, hospital-acquired pneumonia (HAP), and VAP. We excluded subjects transferred from other health-care facilities. As this study was retrospective, the hospital s institutional review board waived any need for informed consent (IRB No ). End Points and Definitions The isolation of a viral organism as the sole pathogen identified served as the primary end point for the study. We defined CAP, HAP, and VAP in accordance with the American Thoracic Society/Infectious Disease Society of America position statement on nosocomial pneumonia. 11 Initially, we identified patients for a potential diagnosis of pneumonia and concurrent ARF based on the physician ordering respiratory cultures. Following case finding based on the ordering of cultures, chest imaging was reviewed by one investigator (M. H. K.) (1) to ensure that there was an infiltrate consistent with a diagnosis of pneumonia and (2) to classify the pneumonia type based on the timing of the pneumonia relative to the onset of hospitalization and MV. We further required that all cases meet at least meet two of the following criteria: fever (> 38 C or < 35 C), leukocytosis or leukopenia, and purulent respiratory secretions to ensure the presence of pneumonia. We classified the results from respiratory cultures as demonstrating either a viral or a bacterial organism or as culture negative. Patients with both a bacterial and viral pathogen (ie, mixed infections) were classified as bacterial for purposes of our analysis given our focus on antibiotic stewardship. Respiratory specimens obtained through various approaches (eg, sputum, tracheal aspirate, and BAL) were included. We also reviewed results from blood cultures, pleural fluid cultures, and 2 Original Research [ - # - CHEST ]

3 urinary antigen tests to categorize subjects regarding the microbiological cause of their pneumonia. To identify viral agents, we relied on the results from a variety of viral specimens that were sent for qualitative nucleic acid tests for respiratory viruses (FilmArray Respiratory Panel; BioFire Diagnostics, Inc.). If only yeast was recovered, that was seen as a contaminant, and the test result was characterized as culture negative. All decisions regarding the ordering of respiratory cultures were undertaken by the patient s primary clinical team and were not guided by a formal protocol. Nonetheless, the ordering of viral testing was considered part of the routine standard evaluation for respiratory failure and pneumonia in the ICU. We recorded patient demographic characteristics along with comorbid illnesses such as coronary artery disease, congestive heart failure, COPD, and others. We calculated a Charlson comorbidity score for each patient to capture the global burden of chronic illness. 12 To assess immune status, we noted if the patient was receiving corticosteroids (> 10 mg of prednisone daily or equivalent), other immunosuppressive agents, or chemotherapy. We also recorded if the patient had undergone any form of transplantation (solid or stem cell transplantation [SCT]). For severity of illness, we calculated the Acute Physiology and Chronic Health (APACHE) II score along with determining whether the patient was in shock at the time of the pneumonia diagnosis. 13 Pneumonia type (eg, CAP, HAP, VAP) was determined, as noted earlier, based on definitions provided in recent guidelines. With respect to outcomes, we assessed in-hospital mortality along with the duration of MV, the ICU length of stay (LOS), and the overall hospital LOS (measured from the time of pneumonia diagnosis). Finally, we calculated the 30-day hospital readmission rate as a function of the cause of pneumonia. Statistical Analysis We compared categorical variables with the Fisher exact test and continuous variables with either the Student t test or the Mann- Whitney U test, as appropriate. Comparisons of continuous variables were analyzed with the Student t test if the data were parametrically distributed. If such data were nonparametric, we used the Mann- Whitney U test. All tests were two-tailed and P <.05 was considered to represent statistical significance. Time to event outcomes (eg, hospital LOS, duration of MV) were compared with the log-rank test. To determine factors independently associated with recovery of a viral cause, we used logistic regression. The regression was a stepwise backward approach, and we entered all variables significant at the 0.15 level in the univariate analysis into the model. Variables were assessed for colinearity. We assessed goodness of fit with the Hosmer- Lemeshow test. AORs and 95% CIs are presented when appropriate. After an initial review of data demonstrated a significant association between SCT and a viral cause, we performed a sensitivity analysis excluding patients who underwent SCT. All analyses were performed using SPSS, version 24.0 (IBM Corp.). Results The final cohort included 364 subjects (mean age, years; 44.2% men) with a high severity of illness (mean APACHE II score, ). Viruses alone were noted in 79 subjects (21.7%). The viruses isolated were diverse and included rhinovirus/enterovirus (n ¼ 20), influenza A (n ¼ 12), respiratory syncytial virus (n ¼ 11), metapneumovirus (n ¼ 8), parainfluenza (n ¼ 7), adenovirus (n ¼ 6), cytomegalovirus (n ¼ 5), and influenza B (n ¼ 1). There was no evident seasonality associated with the recovery of a virus as the cause of pneumonia. Viral testing was performed on respiratory samples obtained in > 95% of cases. In one case of viral pneumonia, concurrent blood culture results revealed a bacterial pathogen. In patients with other than solely viral infections, the majority had bacterial infections. Specifically, in 167 subjects (45.9% of the entire cohort), a bacterial organism was identified. In the remainder (n ¼ 118), neither a bacterial nor viral organism was recovered. In one patient, there was coinfection with both a bacterial pathogen and a virus. This subject had a communityonset infection. As Table 1 reveals, there were no differences between those with a viral cause alone and others with respect to demographic variables or comorbidities. Reflecting this, the median Charlson comorbidity score for the two cohorts was 3. The use of noncorticosteroid immunosuppressive drugs was more common in persons with viral pneumonia, but this difference only approached statistical significance. However, as Table 1 indicates, there was a strong link between SCT and a viral cause. Those who had received or were undergoing SCT were approximately 2 1 / 2 times (OR, 2.51; 95% CI, ) more likely to have a virus identified as the sole cause for their pneumonia in the setting of MV. In contrast, the cause of pneumonia (Fig 1) did not vary based on pneumonia type. For example, two-thirds of all viral cases represented community-onset infection, as did two-thirds of bacterial and culture-negative events. Regarding severity of illness, the APACHE II score was slightly lower in those with a viral infection. This difference, however, only represented a trend toward statistical significance. There was no difference in the prevalence of shock as a function of the type of culprit organism. Strikingly, 67.1% of viral pneumonias were complicated by shock. Outcomes (Table 2) were similar between patients with a viral pathogen and those with other causes of pneumonia. Crude mortality rates were high, with more than 37% of persons with respiratory failure and pneumonia from a virus dying while hospitalized. Viral pneumonia, furthermore, was associated with a median duration of MV of 6 days. chestjournal.org 3

4 TABLE 1 ] Patient Characteristics Variable Virus (n ¼ 79) Other (n ¼ 285) P Value Demographics Age, mean SD, y Male sex, % Race White, % Black, % Other, % Comorbidities CHF, % COPD, % DM, % ESRD, % Malignancy, % Charlson comorbidity score, median Immune status Corticosteroid administration, % Administration of other immunosuppressive medications, % HIV, % Solid organ transplantation, % Stem cell transplantation, % Metastatic malignancy, % HIV, % Severity of illness Shock, % APACHE II score, mean þ SD APACHE ¼ Acute Physiology and Chronic Health Evaluation; CHF ¼ congestive heart failure; DM ¼ diabetes mellitus; ESRD ¼ end stage renal disease. Logistic regression revealed two variables that were independently associated with the isolation of only a virus in pneumonia necessitating MV. An APACHE II score > 26 was linked to a lower risk of a viral organism (AOR, 0.51; 95% CI, ; P ¼.027). Conversely, SCT resulted in a significantly increased likelihood of a virus as the cause of pneumonia and respiratory failure (AOR, 4.39; 95% CI, ; P ¼.001). There was a trend toward viruses being a less frequent cause (AOR, 0.29; 95% CI, p; P ¼.100) in VAP. The overall model had good fit as demonstrated by the Hosmer-Lemeshow C-statistic (P ¼.791). In the sensitivity analysis excluding persons who had undergone or were receiving SCT, the cohort included 330 subjects, 20.3% of whom were classified as having a viral cause. The baseline characteristics of the remaining subjects did not generally vary as a function of whether the pneumonia was classified as viral or nonviral (data not shown). As with the overall analysis, two factors remained independently associated with the recovery of a virus alone. First, an APACHE II score > 26 continued to be connected with a reduced likelihood of a viral cause (AOR, 0.41; 95% CI, ; P ¼.010). Second, treatment with a noncorticosteroid immunosuppressive agent remained associated with a higher prevalence of viral organisms (AOR, 2.36; 95% CI, ; P ¼.010). More broadly, nearly 73% of the cohort was in no way immunocompromised (eg, met no criteria for being immunosuppressed). Among these patients with normal immune systems, viruses were isolated in 20.1%, indicating the generally high prevalence of such organisms. For the remaining cohort that was in any way immunosuppressed, a viral pathogen was responsible in 26.0% of instances (P ¼.255). Discussion This analysis of a cohort of patients with pneumonia while undergoing MV indicates that viruses are the sole cause of pneumonia in approximately 20% of cases. Moreover, viral pneumonia related to respiratory failure 4 Original Research [ - # - CHEST ]

5 print & web 4C=FPO % of Patients Viral Community Onset (n = 237) results in severe critical illness and often is associated with shock. Crude mortality rates and resource use in the setting of viral pneumonia and ARF are substantial and nearly indistinguishable from those related to bacterial infection. Since there are few effective agents to treat viral pulmonary infections, the scenario is essentially akin to patients with severe bacterial pneumonia who receive either delayed or inappropriate antibiotic therapy. Others have examined the epidemiology of pneumonia in more narrow cohorts of subjects. For example, Choi et al 8 documented the high prevalence of viruses in patients hospitalized with community-onset pneumonia. Specifically, they noted that more than one-third of those with health-care-associated pneumonia were infected with a virus. In a follow-up study, they observed that for 22.5% of persons with VAP, the infection was caused by a virus. As we found, they observed that the most common viruses recovered were respiratory syncytial virus and parainfluenza virus. Hong et al 9 further reported a high prevalence of viruses recovered in persons with HAP needing ICU admission. Garbino TABLE 2 ] Outcomes Other (Bacterial & Culture Negative) Hospital-acquired Ventilator-associated Pneumonia (n = 98) Pneumonia (n = 29) Figure 1 Viral etiology based on pneumonia type. Although no statistical difference in the overall distribution of pneumonia types between the population (P ¼.357), ventilator-associated pneumonia was less frequent amont viral cases (P <.05). Outcome Virus (n ¼ 79) Other (n ¼ 285) P Value Hospital mortality, % ICU LOS, median, d Hospital LOS, median, d Duration of MV, median Readmission, % LOS ¼ length of stay; MV ¼ mechanical ventilation. et al, 14 focusing mainly on immunosuppressed subjects, estimated that 17% of such patients undergoing BAL for pneumonia were infected with a viral organism. Finally, in an earlier report investigating the burden of viral pneumonia in nonventilated HAP, we documented that approximately 25% of patients had a virus as the sole culprit pathogen. 15 Our current analysis builds on and adds to these earlier evaluations not only by confirming the relatively high prevalence of viral pathogens in severe pneumonia but also by describing the mortality and morbidity burden associated with these infectious agents. Moreover, we did not restrict our analysis to a specific cohort of patients based on either pneumonia type (eg, CAP vs VAP) or immune status. Hence, we show more broadly that viruses are generally a key causative agent for pneumonia in critically ill patients with ARF, regardless of various pneumonia and patient characteristics. Those patients with viral pneumonia in the setting of ARF represent a potential target for antibiotic stewardship. Formally identifying that a virus is the only infectious agent recovered should facilitate the discontinuation of antibiotics. Since pneumonia represents the most common reason given by physicians for prescribing antibiotics in the ICU, rationally reducing their use when they will not confer benefit should constrain both costs and the ensuing selection pressure that drive antibiotic resistance. However, one cannot determine if a virus is present unless it is sought. In other words, our observations indicate that a search for a viral cause in ARF and pneumonia is prudent, especially if bacterial culture results are unrevealing. Thus, we recommend wider use of viral diagnostic technologies in patients with pneumonia who are receiving MV in the ICU. Unfortunately, clinical characteristics alone do not reliably allow the clinician to predict in whom a virus is unlikely to be present. As our multivariate analysis underscores, few factors are independently associated with the isolation of a virus. Even our attempt to enhance one s ability to stratify patients as to the likelihood of a viral cause through our sensitivity analysis reveals that baseline characteristics do not differ in substantial ways between those with and those without a viral organism as the sole cause for their pneumonia. In short, the select items we identified as independently connected with isolation of virus occur commonly in subjects with a bacterial infection. More importantly, the consistent finding, across various settings and types of pneumonia, that viruses cause chestjournal.org 5

6 nearly one-quarter of such infections emphasizes the need for newer and enhanced rapid diagnostic modalities for viral pathogens along with a necessity for clinical trials exploring novel treatment paradigms. Our study labors under several important limitations. First, this effort is technically retrospective in design and susceptible to various types of bias. However, all the data analyzed were prospectively collected. Second, viral testing was not required in all suspected cases, and our results thus likely represent a lower bound estimate of the true prevalence of viral causes. Likewise, not every patient underwent a standardized evaluation, and select diagnostic tests (eg, serum titers for certain bacterial pathogens) were not routinely ordered. Given that more extensive testing uniformly results in identifying greater numbers of culprit pathogens, we cannot exclude that bias from this phenomenon confounds our results. Our confirmation of the frequency of viruses in severe pneumonia, therefore, likely represents a lower-level estimate of the prevalence of such organisms. Third, no diagnostic test has perfect sensitivity and specificity. Therefore, there is the possibility of misclassification bias. Subjects with true bacterial infections were likely missed. For instance, we did not use procalcitonin to explore how many culturenegative patients might have actually had a high suspicion for a bacterial infection. Fourth, our project represents results from a single large academic hospital. Its generalizability to other settings must not be presumed. Fifth, some of the associations we noted in the multivariate analyses may reflect confounding by indication. In other words, immunosuppressed patients may not actually be at higher risk for viral pneumonia. Rather, because of the existing perception that they are, physicians were more aggressive with viral diagnostic techniques in those patients. The high rate of viral testing, conversely, suggests this may be a more narrow concern. Nonetheless, these various issues indicate that our conclusions should be viewed as hypothesis generating. Allaying many of these worries, however, is the general consistency of our basic findings with those of others who have examined this query from different vantage points. In conclusion, viral pathogens are important causes of pneumonia in patients with ARF who require MV. The viruses recovered are heterogeneous in type and reflect a degree of diversity similar to that encountered with bacterial organisms. Outcomes for patients receiving MV due to viral pneumonia are poor, and clinical covariates do not differentiate patients with distinct microbial causes. More aggressive efforts to search for viral pathogens represent a potential means for enhancing antibiotic stewardship. Acknowledgments Author contributions: A. F. S. is guarantor of the paper, taking responsibility for the integrity of the work as a whole, from inception to published article. A. F. S., K. F.,S.T.M.,andM.H.K.haveallmade substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data; drafting of the submitted article or critical revision for important intellectual content; and provision of final approval of the version to be published. They agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Financial/nonfinancial disclosures: The authors have reported to CHEST the following: M. H. K. is supported by the by the BJC Healthcare Foundation. A. F. S. has received research support from, served as a consultant to, or been a speaker for Accelerate, Achaogen, Alios, Allergan, Aridis, Astellas, AstraZeneca, Bayer, Cidara, Entasys, MedCo, Melinta, Merck, Nabriva, Paratek, and Tetraphase. None declared (K. F., S. T. M.). References 1. Nicolau DP, Dimopoulos G, Welte T, Luyt CE. Can we improve clinical outcomes in patients with pneumonia treated with antibiotics in the intensive care unit? Expert Rev Respir Med. 2016;10: Schreiber MP, Shorr AF. Challenges and opportunities in the treatment of ventilator-associated pneumonia. Expert Rev Anti Infect Ther. 2017;15: Doernberg SB, Chambers HF. Antimicrobial stewardship approaches in the intensive care unit. Infect Dis Clin North Am. 2017;31: Kollef MH, Bassetti M, Francois B, et al. The intensive care medicine research agenda on multidrug-resistant bacteria, antibiotics, and stewardship. Intensive Care Med. 2017;43(9): Zilberberg MD, Nathanson BH, Sulham K, Fan W, Shorr AF. Carbapenem resistance, inappropriate empiric treatment and outcomes among patients hospitalized with Enterobacteriaceae urinary tract infection, pneumonia and sepsis. BMC Infect Dis. 2017;17: Arnold HM, Micek ST, Skrupky LP, Kollef MH. Antibiotic stewardship in the intensive care unit. Semin Respir Crit Care Med. 2011;32: Jain S, Self WH, Wunderink RG, et al. Community-acquired pneumonia requiring hospitalization among U.S. adults. N Engl J Med. 2015;373: Choi SH, Hong SB, Ko GB, et al. Viral infection in patients with severe pneumonia requiring intensive care unit admission. Am J Respir Crit Care Med. 2012;186: Hong HL, Hong SB, Ko GB, et al. Viral infection is not uncommon in adult patients with severe hospital-acquired pneumonia. PLoS One. 2014;9:e Fisher K, Trupka T, Micek ST, Juang P, Kollef MH. A prospective one-year survey of combined pneumonia and respiratory failure. Surg Infect (Larchmt). 2017;18(7): Original Research [ - # - CHEST ]

7 11. Kalil AC, Metersky ML, Klompas M, et al. Executive summary: management of adults with hospital-acquired and ventilator-associated pneumonia: 2016 clinical practice guidelines by the Infectious Diseases Society of America and the American Thoracic Society. Clin Infect Dis. 2016;63: Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol. 1994;47: Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13: Garbino J, Soccal PM, Aubert JD, et al. Respiratory viruses in bronchoalveolar lavage: a hospital-based cohort study in adults. Thorax. 2009;64: Shorr AF, Zilberberg MD, Micek ST, Kollef MH. Viruses are prevalent in non-ventilated hospital-acquired pneumonia. Respir Med. 2017;122: chestjournal.org 7