European Journal of Clinical Microbiology & Infectious Diseases (2020) 39:281–286 https://doi.org/10.1007/s10096-019-03720-x ORIGINAL ARTICLE Ventilator-associated pneumonia diagnosis: a prioritization exercise based on multi-criteria decision analysis João Ferreira-Coimbra 1 & Carmen Ardanuy 2 & Emili Diaz 3 & Marc Leone 4 & Gennaro De Pascale 5,6 & Pedro Póvoa 7,8,9 & Cristina Prat-Aymerich 10 & Ricardo Serrano-Garcia 11,12 & Jordi Solé-Violan 13,14 & Rafael Zaragoza 15 & Jordi Rello 16 Received: 16 August 2019 / Accepted: 20 September 2019 / Published online: 25 October 2019 # Springer-Verlag GmbH Germany, part of Springer Nature 2019 Abstract The aim was to provide global experts ranking on priorities in diagnostic tools for VAP in clinical practice. A multiple criteria decision analysis (MCDA) was performed to identify diagnosis tools for VAP diagnosis. Priority factors were identified after literature review. An international, multidisciplinary expert panel reviewed variables and ranked diagnostic tools. Experts from ten European hospitals participated. Regarding bedside clinical practices, seven required chest X-ray use in all patients, whereas six reported the use of blood cultures and endotracheal aspirate in all patients. Invasive techniques were routinely performed in seven sites. CRP, PCT, and Gram stains were performed in all patients by 5, 2, and 8, respectively. Impact on patient outcomes, safety, and impact on the decision to start antibiotic therapy were ranked as the top three relevant concerns (7.7/10, 7/10, and 6.9/ 10, respectively). Chest X-ray was ranked as the most important imaging technique to diagnose VAP (score 251.7). Apart from blood cultures, endotracheal aspirate culture was identified as the main collection method for the microbiological testing (scores of 274.8 and 246.8, respectively). Mini-BAL was the preferred invasive technique with a score of 208. Top three biomarkers were CRP (score 184.3), PCT (181.3), and WBC (166.4). Gram stain (192.5) was prioritized among laboratory diagnostic techniques. Using MCDA, it is recommended to perform a combination of diagnostic techniques including images (chest X-ray), culture of clinical specimens (blood cultures and endotracheal aspirate), and biomarkers (CRP or PCT) for VAP diagnosis at the bedside. Gram stain was ranked as the preferred laboratory technique. Keywords VAP . Diagnosis . MCDA . Multicriteria decision analysis . Ventilator-associated pneumonia . Microbiology diagnosis . Biomarkers . Imaging Introduction diagnosis, because gold-standard tests are still lacking. It is also missing clear evidence to prioritize diagnostic tests and Ventilator-associated pneumonia (VAP) is a chest infection how they impact clinical outcomes. As evidence is weak, ex- occurring 48 h after the initiation of mechanical ventilation. pert opinion should be taken into account. Multi-criteria deci- It is responsible for increasing hospital length of stay and costs sion analysis (MCDA) is widely used in decision-making in [1, 2]. Many efforts are still being performed to reduce their various areas in order to rank priorities and is also used in burden, especially using prevention bundles [3, 4], which medicine with the same purpose putting together evidence have been shown to be effective in reducing its rate. VAP is and expert opinion [6, 7]. a continuous process presenting with nonspecific clinical fea- Our aim is to create a set of priorities of tools for VAP tures [5], with some limitations regarding its definition and diagnosis in adult patients using a MCDA. Furthermore, we also want to provide data regarding experts’ preferences and Electronic supplementary material The online version of this article the use of diagnostic tools in their practice. (https://doi.org/10.1007/s10096-019-03720-x) contains supplementary material, which is available to authorized users. * João Ferreira-Coimbra Methods joaoferreiracoimbra@gmail.com An international panel of experts with a background in inten- Extended author information available on the last page of the article sive care medicine and clinical microbiology was directly 282 Eur J Clin Microbiol Infect Dis (2020) 39:281–286 invited by email to participate in this study by filling an elec- Two (2/10) reported the absence of the use of biomarkers in tronic survey (Supplementary material 1). Participants were 10–20% of patients, while 3/10 reported no use of CRP and chosen by study coordinator (JR), according to their expertise PCT. Gram stain and MALDI-TOF (matrix assisted laser de- in clinical practice, participation in clinical studies, and pub- sorption ionization-time of flight) were reported as the most lications. Experts also contributed to article revision and pro- used among microbiological techniques, performed in all pa- vision of the final version. tients by 8/10 and 6/10 physicians, respectively. A comprehensive list of VAP diagnostic tests and priority After an MCDA, impact in patient outcomes (7.7/10), risk criteria included in multi-criteria decision analysis (MCDA, to the patient (7.0/10), and impact on decision to start antibi- methodology described elsewhere [8]) were chosen by the otic therapy (6.9/10) were ranked as the major concerns for study coordinator after literature review [9–11]. Seven priority VAP diagnosis, followed by time to clinical relevant result criteria were defined by authors (JFC and JR) and independent- (6.0/10), test availability (5.7/10), need for professional expe- ly weighted by the panel, according to their relative importance rience (4.0/10), and cost (2.7/10). to each other. Diagnostic tests were evaluated individually Chest X-ray was ranked as the most important imaging against priority criteria, by each expert. These results were then technique (score 251.7). “None imaging technique use” strat- multiplied by the weight of each priority criterion to obtain egy obtained the least score (143.6). Blood cultures and endo- individual scores for each diagnostic test. Then, sum score tracheal aspirate culture were scored as the most important was obtained after addition of all the partial scores previously microbiological cultures (scores 274.8 and 246.8). Among achieved. Each variable was transformed into a 0–10 scale in invasive techniques, mini-bronchoalveolar lavage (mini- order to facilitate their comparison. All responses are categori- BAL) was ranked first (208.0), followed by bronchoalveolar cal variables since they are reported using descriptive statistics lavage (BAL; 204.9), the telescopic catheter (176.7), and (proportions and percentages). Median and interquartile range transbronchial biopsy (78.7). Among these techniques, mini- (IQR) were used to analyze nonparametric variables. BAL was intended as the safest (score 46.7) and BAL the Panel members were also inquired about their perceived riskier (after transbronchial biopsy, that is rarely used). relevance of each diagnostic tool and also about their use in Meanwhile, the same distribution appeared in need of profes- clinical practice. For the purpose of this study, VAP is defined sionals’ experience, where mini-BAL was ranked as the one as a chest infection that occurs after 48 h of endotracheal tube that needs less experience (Table 1). placement [9, 10]. This study focuses on adult patients. Due to CRP (score 184.3), PCT (score 171.3), and WBC (score the lack of patients, data in this study, ethical approval, and 166.4) were ordered as the most important biomarkers in informed consent were waived. VAP diagnosis. Gram stain had the most favorable score in microbiological techniques (score 192.5), followed by MALDI-TOF (score 180.0), PCR tests (score 177.5), and Results Gene Expert ® for ESBL (score 162.9). Regarding time to clinical relevant result, Gram stain, PCR tests, and MALDI- A panel of ten European experts with background in intensive TOF were top scored. Table 1 shows the overall results of care medicine and clinical microbiology accepted to partici- the MCDA analysis. According to panel members ranking, pate in our study. chest CT is the imaging technique that impacts the strongest Seven in 10 experts confirmed the use of chest X-ray in all on patient outcomes. Blood cultures, MALDI-TOF, and patients with VAP, while 5/10 and 4/10 admitted the use of PCR are the more important in test microbiology cultures chest CT and lung ultrasound in more than 50% of patients, and techniques, and PCT is intended as the biomarker with respectively. The whole group used at least one imaging tech- most impacting on patient outcomes. Gram stain, MALDI- nique for VAP diagnosis. Blood cultures and endotracheal TOF, and PCR tests were scored first in “Time to clinical aspirate were declared to be the most used methods to obtain relevant result”. Bronchoscopy techniques (included microbiological specimens for culture. All group members transbronchial biopsy) were considered as the unsafest for reported their use in more than 60% and 50% of cases, respec- VAP diagnosis (Table 1). tively. Any invasive technique is used in more than 50% of In this study survey, the panel was also asked to show VAP patients by 7 experts, while bronchoalveolar lavage is the their own relevance for each test (Table 2 shows, in a 10 most commonly used technique (5/10 reported its use in more value scale, expert preference in VAP diagnosis). Chest than 50% of patients). Transbronchial biopsy was reported as X-ray was the preferred imaging technique (score 8.5/ the least used bronchoscopy technique. All declined the use of 10), bronchoalveolar lavage culture the preferred away IL6 and IL10, as well as pro-adrenomedulin in VAP diagnosis to assess lung pathogens (score 8.0/10). Procalcitonin in their practice. White blood cell count (WBC), C-reactive (score 8.4/10) and CRP (score 8.0/10) were classified as protein (CRP), and procalcitonin (PCT) are used in all patients the most relevant biomarkers. Gram stain (8.3/10) was the by 7/10, 5/10, and 2/10 members of the panel, respectively. preferred microbiological technique. Eur J Clin Microbiol Infect Dis (2020) 39:281–286 283 Table 1 MCDA results Impact on Risk to the Impact on decision Time to Test Need of Cost Sum patient patient to start antibiotic clinical availability professionals score outcomes therapy relevant experience result Weight 7.7 7 6.9 6 5.7 4 2.7 Imaging techniques X-ray 55.8 49 53.5 54.2 23 16.2 251.7 Chest CT 69.3 47.3 55.2 32.8 20 6.8 231.4 Lung ultrasound 48.1 56 46.6 37 17 18.2 222.9 None 18.8 22.8 17.2 18.5 40 26.3 143.6 Microbiological tests Blood cultures 57.8 67.7 50.6 49.4 30 19.3 274.8 Endotracheal aspirate culture 33.4 59.5 42.6 53.2 33.3 24.8 246.8 Mini-bronchoalveolar lavage 47.5 46.7 40.3 30.4 26 17.1 208 culture Bronchoalveolar lavage 53.9 28 54.1 31.4 22.6 14.9 204.9 culture Telescopic catheter 51.3 31 35.7 25.7 20 13 176.7 Transbronchial biopsy 25.7 11.7 18.4 9.5 8 5.4 78.7 Biomarkers CRP 63.8 53.2 48 19.3 184.3 PCT 66 57.2 41.5 16.6 181.3 WBC 40.7 53.2 52.1 20.4 166.4 Pro-adrenomedullin 47.3 27.6 21.2 10 106.1 None 17.6 33.5 27.7 23.9 102.7 IL6 39.6 28.6 18.7 9.6 96.5 IL10 33 22.7 18.7 8.1 82.5 DIAGNOSTIC techniques Gram stain 52.9 51 49.2 17.5 21.9 192.5 MALDI-TOF 59.7 42.7 46.3 18.5 12.8 180 PCR tests 59.7 45.8 39.2 23 9.8 177.5 Gene Expert ® for ESBL 56.8 39.8 32.8 24 9.5 162.9 Multiplex PCR 52 34.5 28.5 22 9.8 146.8 Serological tests 25 21.7 29.9 25 16.5 118.1 Rapid automated microscopy 29.8 27 15.7 19 15.5 107 None 2.9 7.5 15 31 25.6 82 Discussion revealing concerns with final patient results and also the im- portance of risk evaluation before diagnostic tools use. Time This study represents the first comprehensive recommenda- to clinical relevant result is also a key decision factor, which tions for VAP diagnosis, enclosing image, microbiologic spec- had only been tested for microbiological techniques, in order imens, biomarkers, and lab tests using a formal decision sup- to rank these techniques as they are critical to assess rapidly port tool. According to this, VAP diagnosis should be based deteriorating patients. Surprisingly, in an era dominated by on chest X-ray, blood cultures, endotracheal aspirate, CRP (or health economics and healthcare resource limitation, experts PCT), and Gram stain. stated the cost as the least important characteristic, in spite of The clinical diagnostic criteria for VAP are subjective, lack being an important limitation in tool availability. specificity. For that, the use of diagnostic tools is widely var- Chest X-ray is the preferred imaging method for VAP diag- iable according to local practices and different beliefs of its nosis in this study. No expert stated to bypass an imaging meth- importance [12]. We evaluated some priority factors that can od to diagnose VAP in clinical practice, in contrast to the CDC affect the diagnostic approach. Impact on patient outcomes approach [13, 14], which excludes any imaging method for and risk were the characteristics of major concern to experts, ventilator-associated events (VAE) surveillance. Our results 284 Eur J Clin Microbiol Infect Dis (2020) 39:281–286 Table 2 Experts panel preference in VAP diagnosis (in proportion 0–10) Imaging techniques Microbiological tests Biomarkers Other laboratory techniques Chest X-ray 8.5 BAL culture 8.0 PCT 8.4 Gram stain 8.3 Chest CT 7.8 Mini-BAL culture 7.0 CRP 8.0 MALDI-TOF 7.5 Lung ultrasound 6.3 ET aspirate 5.8 WBC 6.3 Gene Expert® 6.8 None 2.5 Blood cultures 5.3 Pro-ADM 5.0 Multiplex PCR 6.5 Telescopic catheter 6.3 None 4.7 PCR tests 6.1 Transbronchial biopsy 2.5 IL6 3.9 Rapid automated microscopy 4.8 IL 10 3.7 Serological tests 3.9 None 1.3 BAL bronchoalveolar lavage, ET aspirate endotracheal aspirate, Pro-ADMI pro-adrenomedullin are in line with the ECDC strategy [15], which preserves chest sampling. Interestingly, both are qualified as “Weak recom- imaging as a key for VAP surveillance and also with studies mendation” and “Low-quality evidence”, exposing that this showing different prognosis of VAP and ventilator-associated concerning issue recommendation is more associated with di- tracheobronchitis [16]. In spite of this controversy, it is pertinent vergent interpretation than different recognized evidence. to keep in mind that VAP surveillance and diagnosis or clinical On the other hand, in the MCDA, mini-BAL scored management are different approaches. Important to note that, slightly better than BAL, mainly driven by risk profile despite lung ultrasound (LUS) use is increasing [17], in our (safer) and less necessity for professionals’ experience. study experts ranked as the least important imaging technique. This represents an important field, as in some places pro- This could be explained by the lack of training and access in fessionals lack experience in bronchoscopic BAL or if LUS, which can limit its use. experienced operators are not always present. Besides cost In IDSA/ATS 2016 [10] and ERS/ESICM/ESCMID/ and availability were out of the highest experts priorities, ALAT 2017 [9] guidelines, biomarkers used are not supported very specific tests as PCR tests and multiplex PCR were in VAP diagnosis as well as in decision for antibiotic with- not in the top priorities regarding microbiological tech- drawal (except in specific circumstances). Meanwhile, bio- niques. Timely results are of the highest importance in markers are abundantly used worldwide in VAP diagnosis, critically ill patient outcomes [21, 22] and also in reduc- despite clear evidence proving that their positive impact on ing antibiotic pressure; therefore, the need for timely re- outcomes had never been published [18]. Most of our panel sults is a relevant issue. In this subject, Gram stain, PCR declared the use of WBC, CRP, and PCT in clinical practice, tests, and Multiplex PCR that are able to produce results in spite of MCDA showed a preference by CRP and PCT. This within 1–2 h can have a special role. Notwithstanding, it could be explained by their low cost, and also, their use is is also important to take into account local microbiology generally intended as a good clinical practice (as some behav- because tools like Gene Expert ® for ESBL can be prior- iors in medicine that are not supported by high-quality evi- itized in some places. dence). Differently, IL6, IL10, and Pro-adrenomedullin re- This study has several limitations. First, the experts panel a main usefulness tools beyond investigational purposes. limited European panel that can impair study generalization Microbiological cultures and tests remain one of the most for other areas. Second, derived recommendations and data divergent issues in different guidelines [9–11]. Blood cultures, from patients did not take into account nor the number of despite infrequently positive, were previously recognized as patients treated in physicians ICUs, neither the physicians’ important tools in VAP [19], since bacteremic VAP is associ- experience. Third, specific patients admitted to each ICU (car- ated with higher mortality [20]. Although, in our MCDA ex- diac, burns, trauma, etc) were not addressed. Fourth, the avail- perts ranked blood cultures as the most important culture spec- ability of diagnostic tools in each hospital was not addressed, imen, when asked about their preferences in VAP diagnosis, which could limit physicians’ experience with some diagnos- blood cultures were only the fourth preferred. Between the tic techniques, influencing their beliefs. other methods of specimens’ collection for culture, endotra- The strengths include the study methodology (MCDA) and cheal aspirate score first in MCDA, at the same time as BAL the possible generalization of this data, as previously reported and mini-BAL were the preferred (by this order) invasive [23], and the possibility to update this information in the future techniques. This can be related to the controversy about this if new diagnostic tools becomes available in clinical practice. issue in international guidelines. While American guidelines Through MCDA analysis, this experts panel recommends [10] recommend noninvasive and semi-quantitative sampling, for VAP diagnosis a set of chest X-ray, blood cultures, endo- European guidelines [9] recommend invasive and quantitative tracheal aspirate, CRP (and PCT), and Gram stain. Eur J Clin Microbiol Infect Dis (2020) 39:281–286 285 References mechanical ventilation in European intensive care units. Crit Care Med. 2009. https://doi.org/10.1097/CCM.0b013e3181a037ac 1. Safdar N, Dezfulian C, Collard HR, Saint S. Clinical and economic 13. Magill SS, Klompas M, Balk R, et al. Developing a new, national consequences of ventilator-associated pneumonia: A systematic re- approach to surveillance for ventilator-associated events. Crit Care view. Crit Care Med. 2005. https://doi.org/10.1097/01.CCM. Med. 2013. https://doi.org/10.1097/CCM.0b013e3182a262db 0000181731.53912.D9 14. Klompas M. Complications of Mechanical Ventilation — The 2. Rello J, Ollendorf DA, Oster G, et al. Epidemiology and outcomes CDC’s New Surveillance Paradigm. N Engl J Med. 2013. https:// of ventilator-associated pneumonia in a large US database. Chest. doi.org/10.1056/nejmp1300633 2002. https://doi.org/10.1378/chest.122.6.2115 15. Plachouras D, Lepape A, Suetens C. ECDC definitions and 3. Klompas M, Branson R, Eichenwald EC, et al. Strategies to Prevent methods for the surveillance of healthcare-associated infections in Ventilator-Associated Pneumonia in Acute Care Hospitals: 2014 intensive care units. Intensive Care Med. 2018. https://doi.org/10. Update. Infect Control Hosp Epidemiol. 2014. https://doi.org/10. 1007/s00134-018-5113-0 1086/677144 16. Martin-Loeches I, Povoa P, Rodríguez A, et al. Incidence and prog- 4. Meduri GU. Diagnosis and differential diagnosis of ventilator-as- nosis of ventilator-associated tracheobronchitis (TAVeM): A sociated pneumonia. Clin Chest Med. 1995. multicentre, prospective, observational study. Lancet Respir Med. 5. Fagon JY, Chastre J, Hance AJ, Domart Y, Trouillet JL, Gibert C. 2015. https://doi.org/10.1016/S2213-2600(15)00326-4 Evaluation of clinical judgment in the identification and treatment 17. Wang G, Ji X, Xu Y, Xiang X. Lung ultrasound: A promising tool to of nosocomial pneumonia in ventilated patients. Chest. 1993. monitor ventilator-associated pneumonia in critically ill patients. https://doi.org/10.1378/chest.103.2.547 Crit Care. 2016. https://doi.org/10.1186/s13054-016-1487-y 6. Rello J, Kalwaje Eshwara V, Lagunes L, et al. A global priority list 18. Palazzo SJ, Simpson T, Schnapp L. Biomarkers for ventilator-asso- of the TOp TEn resistant Microorganisms (TOTEM) study at inten- ciated pneumonia: Review of the literature. Hear Lung J Acute Crit sive care: a prioritization exercise based on multi-criteria decision Care. 2011. https://doi.org/10.1016/j.hrtlng.2010.11.003 analysis. Eur J Clin Microbiol Infect Dis. 2019. https://doi.org/10. 19. Rello J, Gallego M, Mariscal D, Soñora R, Valles J. The value of 1007/s10096-018-3428-y routine microbial investigation in ventilator-associated pneumonia. 7. Tacconelli E, Carrara E, Savoldi A, et al. Discovery, research, and Am J Respir Crit Care Med. 1997. https://doi.org/10.1164/ajrccm. development of new antibiotics: the WHO priority list of antibiotic- 156.1.9607030 resistant bacteria and tuberculosis. Lancet Infect Dis. 2018. https:// 20. Agbaht K, Diaz E, Muñoz E, et al. Bacteremia in patients with doi.org/10.1016/S1473-3099(17)30753-3 ventilator-associated pneumonia is associated with increased mor- 8. Belton V, Stewart T (2002) Multiple Criteria Decision Analysis: An tality: A study comparing bacteremic vs. nonbacteremic ventilator- Integrated Approach. Kluwer Academic Publishers, Massachusetts associated pneumonia. Crit Care Med. 2007. https://doi.org/10. https://doi.org/10.1007/978-1-4615-1495-4 1097/01.CCM.0000277042.31524.66 9. Torres A, Niederman MS, Chastre J, et al. International ERS/ 21. Kollef MH, Sherman G, Ward S, Fraser VJ. Inadequate antimicro- ESICM/ESCMID/ALAT guidelines for the management of hospi- bial treatment of infections: A risk factor for hospital mortality tal-acquired pneumonia and ventilator-associated pneumonia. Eur among critically III patients. Chest. 1999. https://doi.org/10.1378/ Respir J. 2017. https://doi.org/10.1183/13993003.00582-2017 chest.115.2.462 10. Kalil, A. C., Metersky, M. L., Klompas, M., Muscedere, J., Sweeney, D. A., Palmer, L. B., … Brozek JL. Management of 22. Pulido MR, Moreno-Martínez P, González-Galán V, et al. Adults with Hospital-Acquired and Ventilator-Associated Application of BioFire FilmArray Blood Culture Identification Pneumonia: 2016 Clinical Practice Guidelines by the Infectious panel for rapid identification of the causative agents of ventilator- Diseases Society of America and the American Thoracic Society. associated pneumonia. Clin Microbiol Infect. 2018. https://doi.org/ Infect Dis Clin Pract. 2016. https://doi.org/10.1097/IPC. 10.1016/j.cmi.2018.06.001 0000000000000476 23. Rello J, Lode H, Cornaglia G, Masterton R. A European care bun- 11. Leone M, Bouadma L, Bouhemad B, et al. Hospital-acquired pneu- dle for prevention of ventilator-associated pneumonia. Intensive monia in ICU. Anaesth Crit Care Pain Med. 2018. https://doi.org/ Care Med. 2010. https://doi.org/10.1007/s00134-010-1841-5 10.1016/j.accpm.2017.11.006 12. Koulenti D, Lisboa T, Brun-Buisson C, et al. Spectrum of practice Publisher’s note Springer Nature remains neutral with regard to jurisdic- in the diagnosis of nosocomial pneumonia in patients requiring tional claims in published maps and institutional affiliations. Affiliations João Ferreira-Coimbra 1 & Carmen Ardanuy 2 & Emili Diaz 3 & Marc Leone 4 & Gennaro De Pascale 5,6 & Pedro Póvoa 7,8,9 & Cristina Prat-Aymerich 10 & Ricardo Serrano-Garcia 11,12 & Jordi Solé-Violan 13,14 & Rafael Zaragoza 15 & Jordi Rello 16 1 Internal Medicina Department, Centro Hospitalar Universitário do Sabadell, CIBERES Ciber de Enfermedades Respiratorias, Porto, Porto, Portugal Barcelona, Spain 2 4 Microbiology Department, Hospital Universitari de Bellvitge, Department of Anesthesiology and Critical Care Medicine, Aix University of Barcelona, CIBERES, Instituto Salud Carlos III, Marseille University, Assistance Publique Hôpitaux de Marseille, Madrid, Spain Nord hospital, Marseille, France 3 5 Department of Medicine, Universitat Autònoma de Barcelona Dipartimento di Scienza dell’Emergenza, Anestesiologiche e della (UAB), Critical Care Department, Corporació Sanitària Parc Taulí, Rianimazione–UOC di Anestesia, Rianimazione, Terapia Intensiva e 286 Eur J Clin Microbiol Infect Dis (2020) 39:281–286 11 Tossicologia Clinica, Fondazione Policlinico Universitario A. Intensive Medicine Service, General University Hospital of Gemelli IRCCS, Rome, Italy Alicante), Alicante, Spain 6 12 Università Cattolica del Sacro Cuore, Istituto di Anestesia e Department of Clinical Medicine, Miguel Hernández University of Rianimazione, Rome, Italy Elche, Alicante, Spain 7 13 NOVA Medical School, New University of Lisbon, Lisbon, Portugal Intensive Care Medicine Departament, Hospital Universitario de 8 GC De Negrín, Las Palmas de GC, Spain Center for Clinical Epidemiology and Research Unit of Clinical 14 Epidemiology, OUH Odense University Hospital, Odense, Denmark CIBERES, Instituto Salud Carlos III, Madrid, Spain 9 15 Hospital de São Francisco Xavier, Centro Hospitalar de Lisboa Intensive Care Unit, Hospital Universitario Dr. Peset, Ocidental, Lisbon, Portugal Valencia, Spain 10 16 Hospital Universitari Germans Trias i Pujol, Fundació Institut Clinical Research/Epidemiology In Pneumonia & Sepsis (CRIPS), d’Investigació en Ciències de la Salut Germans Trias i Pujol, Vall d’Hebron Institute of Research, CIBERES-Centro de Universitat Autònoma de Barcelona, CIBER Enfermedades Investigación en Red de Enfermedades Respiratorias, Respiratorias, Barcelona, Spain Barcelona, Spain