Intensivists admit to being less than 50% confident about more than 30% of patients they diagnose with ventilator-associated pneumonia. Sensitivity and positive predictive value of quantitative bronchoalveolar lavage cultures are only about 50% and 70% relative to histology. There are no official standards for concepts such as worsening oxygenation, increased secretions, or progressive infiltrates.
Intensivists admit to being less than 50% confident about more than 30% of patients they diagnose with ventilator-associated pneumonia. Sensitivity and positive predictive value of quantitative bronchoalveolar lavage cultures are only about 50% and 70% relative to histology. There are no official standards for concepts such as worsening oxygenation, increased secretions, or progressive infiltrates.
Intensivists admit to being less than 50% confident about more than 30% of patients they diagnose with ventilator-associated pneumonia. Sensitivity and positive predictive value of quantitative bronchoalveolar lavage cultures are only about 50% and 70% relative to histology. There are no official standards for concepts such as worsening oxygenation, increased secretions, or progressive infiltrates.
722 www.ccmjournal.org March 2014 Volume 42 Number 3
B edside clinicians know better than anyone that diagnos- ing ventilator-associated pneumonia (VAP) can be dev- ilishly difcult. Intensivists admitted to being less than 50% condent about more than 30% of patients they diagnose with VAP (1). Their uncertainty is well placed. Autopsy series afrm that clinical diagnostic criteria miss a third or more of true cases and wrongly suggest VAP in almost one half of all cases (2). Requiring microbiological conrmation does little to improve condence. The sensitivity and positive predictive value of quantitative bronchoalveolar lavage cultures are only about 50% and 70%, respectively, relative to histology (37). As with clinical diagnoses, so too with surveillance designa- tions. Infection preventionists are just as likely to get it wrong. In fact, infection preventionists may be more prone to error because they often rely on indirect evidence, such as radio- graphic reports rather than images, to make their determina- tions. The problem is further compounded by subjectivity. There are no ofcial standards for concepts such as worsen- ing oxygenation, increased secretions, or progressive inltrates. Surveyors are instead left to their own discretion. Not surpris- ingly, reasonable practitioners often disagree (8, 9). In this issue of Critical Care Medicine, the article by Ste- vens et al (10) reminds us just how unreliable and variable VAP classications can be. Stevens et al (10) developed six clinical vignettes describing possible cases of VAP. They distributed the vignettes to a nationally representative sample of infection preventionists and hospital epidemiologists and asked them to rate whether each patient has VAP or not. They found almost no agreement between survey respondents. Answers followed a near normal distribution suggesting near random variation in VAP assignments. Although it is possible that agreement may have been better if respondents had had access to the full medical charts or bedside evaluations rather than just vignettes alone, the ndings by Stevens et al (10)are consistent with other published surveys using different methodologies (8, 9). These sobering results remind us yet again that VAP is a fundamentally unreliable outcome measure. This is true regardless of whether the application is benchmarking hos- pitals, determining remuneration, studying the impact of different prevention strategies, or evaluating the efcacy of new treatments. VAP is simply too subjective, inaccurate, and prone to bias to yield dependable answers in any of these spheres. Furthermore, the use of this VAP surveillance deni- tion had led to the misinterpretation and misuse of pneumo- nia bundles of care (11). These results also remind us of why the Centers for Dis- ease Control and Prevention (CDC) retired their longstanding clinical surveillance denitions for VAP and created venti- lator-associated event (VAE) denitions instead (12). VAE denitions were designed to enhance objectivity and facilitate automated surveillance by eliminating subjective criteria (such as worsening oxygenation or increased secretions) and substituting quantitative alternatives instead (e.g., an increase in the daily minimum positive end-expiratory pressure [PEEP] of 3 cm H 2 O sustained for 2 d) (13). More broadly, the VAE framework was designed to overcome the limited accuracy of VAP denitions by shifting from diagnosis-specic surveil- lance to syndrome-based surveillance. Ventilator-associated conditions (VAC), the core concept within the VAE framework, can be thought of as a surveillance marker for the syndrome of nosocomial respiratory deteriorations. To get the most out of VAE surveillance, however, we have to appreciate its unique strengths and respect its limitations. First and foremost, VAC is not VAP. Pneumonia only accounts for a small fraction of VACs just as pneumonia only accounts for a small fraction of attributable mortality in ICUs (14). Acute respiratory distress syndrome, severe sepsis, pulmonary Copyright 2013 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000436119.32758.69 *See also p. 497. Key Words: surveillance; variability; ventilator-associated pneumonia Dr. Klompas lectured for Premier Healthcare Alliance and received support for travel and accommodation expenses from Society of Health- care Epidemiologists of America, Infectious Disease Society of America, and Infectious Disease Association of California. His institution received grant support from CDC. Dr. Kalil disclosed that he does not have any potential conficts of interest. Michael Klompas, MD, MPH Department of Population Medicine Harvard Medical School and Harvard Pilgrim Health Care Institute; and Department of Medicine Brigham and Womens Hospital Boston, MA Andre C. Kalil, MD, MPH Infectious Diseases Division Department of Internal Medicine University of Nebraska Medical Center Omaha, NE The Last Breath of the Ventilator-Associated Pneumonia Surveillance Denition* Editorials Critical Care Medicine www.ccmjournal.org 723 edema, atelectasis, and other complications account for many more deaths. VAC detects many of these complications in addition to pneumonia (15, 16). To prevent VACs and improve outcomes for all ventilated patients, we need new ventilator bundles that include strategies that better target all these con- ditions (17). Second, VAC is a surveillance concept not a clini- cal concept. VACs strengths lie in its objectivity, its suitability for electronic analysis, and its strong association with adverse outcomes (18). It is designed to give a population-level pic- ture of complication rates rather than patient-level diagnostic information to inform immediate management. The price we pay for VACs simplicity and objectivity is that VAC will likely miss some milder pneumonias and overcall some nonprevent- able deteriorations. In return, though, VAC should provide a more consistent yardstick to measure relative complication rates between institutions and across time. The article by Stevens et al (10) is also a welcome prompt to think more deeply about VAEs objectivity. VAE denitions have the ring of objectivity insofar as they are built solely upon quantitative criteria and computers can successfully apply them (18). VAEs objectivity cannot be taken for granted, however. There are potential sources of variability second- ary to patient care and surveillance practices that we need to understand better and guard against if VAE rates are to be valid benchmarks. On the patient care side, clinicians and centers may differ in how they choose to manage ventilator settings for similar scenarios. This is an important area for investigation and quantication. On the surveillance side, there is a very real possibility that surveyors charged with manually nding and processing daily minimum PEEPs and FIO 2 will nd different cases than electronic systems that analyze ventilators minute- by-minute settings. It is also likely that human assessors manu- ally screening ventilator settings and antibiotic exposures will occasionally make mistakes. Hence, we welcome Stevens et al (10) suggestion that research to demonstrate the interfacil- ity and interobserver reliability of the new denition (VAE) is needed before adopting it for hospital comparison purposes. To enhance VAEs objectivity, we need to encourage as many hospitals as possible to automate VAE surveillance and we need to dene common standards for the interpretation of electronic ventilator data. For example, should a single minute spent at a very low PEEP qualify as an eligible daily minimum PEEP? It will both help hospitals and increase the credibility of VAE surveillance if CDC could disseminate VAE detection computer algorithms or web services to assure hospitals that their cases are valid and comparable to those of other centers. Centraliz- ing the development and dissemination of analytic algorithms will also allow CDC to detect attempts to game the new de- nitions by manipulating ventilator settings in small clinically meaningless ways that serve only to undermine VAE detection. After so many years using CDCs former VAP surveillance denitions, we can now safely conclude that any potential agreement achieved between observers, or between hospitals, was no better than chance alone. The study by Stevens et al (10) clearly shows that the old VAP denition reached its very last breath. Although we believe that VAE denitions are a major step forward, these new denitions now require scientic repli- cation and external validity to assess their suitability for bench- marking, accreditation, and remuneration. REFERENCES 1. Koulenti D, Lisboa T, Brun-Buisson C, et al; EU-VAP/CAP Study Group: Spectrum of practice in the diagnosis of nosocomial pneumo- nia in patients requiring mechanical ventilation in European intensive care units. Crit Care Med 2009; 37:23602368 2. Tejerina E, Esteban A, Fernndez-Segoviano P, et al: Accuracy of clini- cal defnitions of ventilator-associated pneumonia: Comparison with autopsy fndings. J Crit Care 2010; 25:6268 3. Kirtland SH, Corley DE, Winterbauer RH, et al: The diagnosis of ven- tilator-associated pneumonia: A comparison of histologic, microbio- logic, and clinical criteria. Chest 1997; 112:445457 4. Fbregas N, Ewig S, Torres A, et al: Clinical diagnosis of ventilator associated pneumonia revisited: Comparative validation using imme- diate post-mortem lung biopsies. Thorax 1999; 54:867873 5. Papazian L, Thomas P, Garbe L, et al: Bronchoscopic or blind sam- pling techniques for the diagnosis of ventilator-associated pneumo- nia. Am J Respir Crit Care Med 1995; 152:19821991 6. Marquette CH, Copin MC, Wallet F, et al: Diagnostic tests for pneu- monia in ventilated patients: Prospective evaluation of diagnostic accuracy using histology as a diagnostic gold standard. Am J Respir Crit Care Med 1995; 151:18781888 7. Torres A, el-Ebiary M, Padr L, et al: Validation of different techniques for the diagnosis of ventilator-associated pneumonia. Comparison with immediate postmortem pulmonary biopsy. Am J Respir Crit Care Med 1994; 149:324331 8. Klein Klouwenberg PMC, Ong DSY, Bos LDJ, et al: Interobserver agreement of Centers for Disease Control and Prevention criteria for classifying infections in critically ill patients. Crit Care Med 2014; 41:23732378 9. Klompas M: Interobserver variability in ventilator-associated pneumo- nia surveillance. Am J Infect Control 2010; 38:237239 10. Stevens JP, Kachniarz B, Wright SB, et al: When Policy Gets It Right: Variability in U.S. Hospitals Diagnosis of Ventilator-Associated Pneumonia. Crit Care Med 2014; 42:497503 11. Kalil AC, Wiener-Kronish JP: Is the evidence for benefts from venti- lator-associated pneumonia bundles reliable enough for implementa- tion in a general hospital? Crit Care Med 2012; 40:348350 12. Magill SS, Klompas M, Balk R, et al: Developing a new, national approach to surveillance for ventilator-associated events. Crit Care Med 2013; 41:24672475 13. Centers for Disease Control and Prevention: Ventilator-Associated Event Protocol. 2013. Available at: http://www.cdc.gov/nhsn/acute- care-hospital/vae/index.html. Accessed February 25, 2013 14. Melsen WG, Rovers MM, Groenwold RH, et al: Attributable mortal- ity of ventilator-associated pneumonia: A meta-analysis of individual patient data from randomised prevention studies. Lancet Infect Dis 2013; 13:665671 15. Klompas M, Khan Y, Kleinman K, et al; CDC Prevention Epicenters Program: Multicenter evaluation of a novel surveillance paradigm for complications of mechanical ventilation. PLoS One 2011; 6:e18062 16. Hayashi Y, Morisawa K, Klompas M, et al: Toward improved surveil- lance: The impact of ventilator-associated complications on length of stay and antibiotic use in patients in intensive care units. Clin Infect Dis 2013; 56:471477 17. Klompas M: Ventilator-associated events surveillance: A patient safety opportunity. Curr Opin Crit Care 2013; 19:424431 18. Klompas M, Magill S, Robicsek A, et al; CDC Prevention Epicenters Program: Objective surveillance defnitions for ventilator-associated pneumonia. Crit Care Med 2012; 40:31543161