clinical investigations in critical care Impact of Ventilator-Associated Pneumonia on Outcome in Patients With COPD*

Transcription

1 clinical investigations in critical care Imact of Ventilator-Associated Pneumonia on Outcome in With COPD* Saad Nseir, MD; Christohe Di Pomeo, PhD; Stéhane Soubrier, MD; Béatrice Cavestri, MD; Elsa Jozefowicz, MD; Fabienne Saulnier, MD; and Alain Durocher, MD Purose: The aim of this study was to determine the imact of ventilator-associated neumonia (VAP) on outcome in atients with COPD. Methods: Prosective, observational, case-control study conducted in a 30-bed ICU during a 5-year eriod. All COPD atients who required intubation and mechanical ventilation (MV) for > 48 h were eligible. VAP diagnosis was based on clinical, radiograhic, and quantitative microbiologic criteria. with unconfirmed VAP were excluded, as well as atients with ventilator-associated tracheobronchitis without subsequent VAP. Matching (1:1) criteria included MV duration before VAP occurrence, age 5 years, simlified acute hysiology score II on ICU admission 5, and ICU admission category. Variables associated with ICU mortality were determined using univariate and multivariate analyses. Results: A total of 1,241 atients were eligible; 181 atients (14%) were excluded, including 133 atients for VAT and 48 atients for unconfirmed VAP. VAP develoed in 77 atients (6%), and all were successfully matched. Pseudomonas aeruginosa was the most frequently isolated bacteria (31%). ICU mortality rate (64% vs 28%), duration of MV (24 15dvs13 11d[ SD]), and ICU stay (26 17dvs15 13 d) were significantly (< 0.001) higher in case atients than in control atients. VAP was the only variable indeendently associated with ICU mortality (odds ratio [OR], 7.7; 95% confidence interval [CI], 3.2 to 18.6; < 0.001). In VAP atients who received corticosteroids during their ICU stay comared with those who did not receive corticosteroids, mortality rate (50% vs 82%; OR, 1.8; 95% CI, 1.2 to 2.7; 0.002), duration of MV (21 14 d vs d, 0.043), and ICU stay (22 16dvs31 18 d, 0.006) were significantly lower. Conclusion: VAP is associated with increased mortality rates and longer duration of MV and ICU stay in COPD atients. (CHEST 2005; 128: ) Key words: clinical outcome; COPD; corticosteroid treatment; critical care; mechanical ventilation; nosocomial neumonia; ventilator-associated neumonia Abbreviations: CI confidence interval; MRB multidrug-resistant bacteria; MV mechanical ventilation; NIV noninvasive ventilation; OR odds ratio; SAPS simlified acute hysiology score; VAP ventilator-associated neumonia; VAT ventilator-associated tracheobronchitis *From the Intensive Care Unit (Drs. Nseir, Soubrier, Cavestri, Jozefowicz, Saulnier, and Durocher), Calmette Hosital, Regional University Centre; and Medical Assessment Laboratory (Dr. Di Pomeo), EA 3614, Lille II University, Lille, France. Presented in art at the 101st American Thoracic Society Conference, May 20 25, 2005, San Diego, CA. Ventilator-associated neumonia (VAP) is the second-most-frequent nosocomial infection among ICU atients. This infection is associated with high mortality and morbidity rates. 1 However, mortality attributable to VAP is controversial. The largest casecontrol study, 2 conducted in the United States on 842 VAP atients, found no significant difference in mortality rates between VAP atients and matched control Manuscrit received December 15, 2004; revision acceted February 14, Reroduction of this article is rohibited without written ermission from the American College of Chest Physicians ( org/misc/rerints.shtml). Corresondence to: Saad Nseir, MD, Réanimation Médicale, Hôital Calmette, CHRU, boulevard du Pr Leclercq, Lille cedex, France; s-nseir@chru-lille.fr 1650 Clinical Investigations in Critical Care

2 subjects, suggesting that severely ill atients died with VAP and not of VAP. However, other well-conducted case-control studies 3 5 concluded that VAP is associated with an attributable mortality rate (range, 27 to 50%). These conflicting results could be exlained by differences in atient characteristics, adequacy of initial antimicrobial treatment, microorganisms resonsible for VAP, severity of illness, comorbid factors, and host resonse factors. 6 COPD is a risk factor for nosocomial lower resiratory tract infections. 7,8 Rello et al 9 demonstrated that COPD was associated with higher mortality rates in VAP atients. However, after adjustment for confounding factors, COPD was not indeendently associated with mortality in VAP atients. To our knowledge, no study has evaluated the imact of VAP on mortality and morbidity in COPD atients. However, bronchoulmonary infection could result in increased mortality and morbidity rates when structural lung disease is resent. Therefore, we erformed a rosective casecontrol study to determine the imact of VAP on mortality and duration of mechanical ventilation (MV) and ICU stay in COPD atients. Materials and Methods This rosective, observational, case-control study was conducted in a 30-bed ICU from January 1996 to January Because it was observational, Institutional Review Board aroval was not required in accordance with Institutional Review Board Regulation. All COPD atients who required intubation and MV for 48 h were eligible. without COPD, trauma atients, atients who did not receive MV or received MV for 48 h, atients who only received noninvasive ventilation (NIV), atients with solid or hematology malignancy, and atients with tracheotomy at ICU admission were not eligible. with ventilator-associated tracheobronchitis (VAT) 7 without subsequent VAP were excluded, as well as atients with clinically susected VAP without microbiologic confirmation. All data were rosectively collected. VAP eisodes were identified by rosective surveillance of nosocomial infections. Only first eisodes of VAP occurring 48 h after the initiation of MV were included. Study Poulation were intubated either via the oral or nasal routes according to the clinical status and the habits of the hysician in charge. The oroharyngeal cavity was cleaned qid with chlorhexidine solution. Continuous subglottic suctioning was not utilized. The ventilator circuit was not changed routinely. In all atients, a heat and moisture exchanger was ositioned between the Y iece and the atient; the heat and moisture exchangers were changed every 48 h or more frequently if visibly soiled. were ket in semirecumbent osition during most of their eriod of MV. There was no systematic stress ulcer rohylaxis and no selective digestive decontamination. Infection control olicy included isolation techniques in atients with multidrug-resistant bacteria (MRB), written antibiotic treatment rotocol, and continuous surveillance of nosocomial infections. Corticosteroid treatment (methylrednisolone, 0.5 to 1 mg/kg/d) was at the hysicians discretion. This treatment was administered for acute exacerbation of COPD or bronchosasm. In the absence of contraindications, NIV was used in all COPD atients with acute hyercanic resiratory failure. Intubation was erformed in atients with NIV failure or contraindications for NIV. Definitions COPD was defined according to American Thoracic Society criteria. 10 VAP was defined by the resence of two of the following criteria: temerature 38.5 C or 36.5 C, leukocyte count 10,000/ L or 1,500/ L, urulent tracheal asirate, associated with new or rogressive radiograhic infiltrate and a ositive ( 10 6 cfu/ml) tracheal asirate culture result. VAP eisodes occurring 5 d after starting MV were considered to be early onset. Late-onset VAP was defined as VAP diagnosed 5d after starting MV. Prior antibiotic use was defined as any antibiotic treatment during the 2 weeks receding ICU admission. Antimicrobial theray was deemed adequate when at least one antibiotic active in vitro on all organisms causing VAP was administrated in aroriate dosage within the first 24 h of VAP occurrence. MRB were defined as follows: methicillin-resistant Stahylococcus aureus, ceftazidime or imienem-resistant Pseudomonas aeruginosa, Acinetobacter baumannii, extendedsectrum -lactamase-roducing Gram-negative bacilli, and Stenotrohomonas maltohilia. Long-term oral corticosteroid use was defined as administration of corticosteroids ( 20 mg/d) during at least 1 month over the last 3 months. Corticosteroid treatment during ICU stay was defined as any corticosteroid treatment for 2 days. Outcomes evaluated included ICU mortality, duration of MV, and ICU stay. Matching Criteria Each case atient was matched to one control atient according to the following criteria: (1) duration of MV before VAP occurrence; a control atient had to have at least the same duration of MV before VAP occurrence as a case atient; (2) age 5 years; (3) simlified acute hysiology score (SAPS) II on ICU admission 5 oints; (4) category of ICU admission (medical/ surgical); and (5) date of ICU admission when more than one otential control atient was well matched to a case atient. Statistical Analysis Statistical software (SPSS version 9; SPSS; Chicago, IL) was used for data analysis. Case atients were comared with control subjects using 2 test or Fisher Exact Test when aroriate for qualitative variables, and Mann-Whitney U test for quantitative variables. Results are resented as frequency (%) for qualitative variables or mean SD for quantitative variables. Univariate and multivariate analyses were erformed to determine variables associated with ICU mortality. The following variables were included in the univariate analysis: age, gender, SAPS II on ICU admission, surgery, transfer from other wards, diabetes mellitus, rior antibiotic use, long-term oral corticosteroid use, organ failures, rimary diagnosis, corticosteroid treatment during ICU stay, VAP, VAP related to MRB, and adequacy of initial antimicrobial treatment for VAP. Only significant ( 0.05) variables were included in the stewise logistic regression model. Results During the study eriod, 1,241 atients were eligible; 181 atients (14%) were excluded, including CHEST / 128 / 3/ SEPTEMBER,

3 Figure 1. Study rofile. 133 atients (10%) for VAT without subsequent VAP, and 48 atients (3%) for clinically susected VAP without bacteriologic confirmation. VAP develoed in 77 atients (6%), and were all successfully matched (Fig 1). The rate of rior antibiotic use was higher in case atients than in control atients. During ICU stay, the duration of antibiotic treatment was longer in case atients than in control atients. Other atient characteristics were similar in case atients and control atients (Table 1). The mean time between starting MV and VAP occurrence was 12 9d( SD). VAP was late in onset in 55 atients (71%). Initial antimicrobial treatment was adequate in 58 atients (75%) with VAP. The rate of inadequate initial antibiotic treatment was higher in atients with VAP related to MRB than in atients with VAP related to other bacteria (18 of 32 atients [56%] vs 1 of 45 atients [2%]; odds ratio [OR], 2.3; 95% confidence interval [CI], 1.5 to 3.6; 0.001). The mean dose of corticosteroids re- Table 1 Patient Characteristics* Excluded Characteristics Cases (n 77) Controls (n 77) Value Unconfirmed VAP (n 48) Values VAT (n 133) Values At ICU admission Age, yr / /0.049 Male gender 62 (80) 60 (77) (77) 0.337/ (75) 0.167/0.319 SAPS II / /0.041 Surgical atients 8 (10) 8 (10) (10) 0.519/ (9) 0.438/0.438 Transfer from other wards 67 (87) 68 (88) (87) 0.552/ (86) 0.539/0.438 Diabetes mellitus 10 (12) 14 (18) (20) 0.238/ (14) 0.576/0.288 Prior antibiotic use 53 (68) 33 (42) (70) 0.488/ (57) 0.062/0.032 Long-term oral corticosteroid use 34 (44) 28 (36) (35) 0.220/ (40) 0.492/0.120 Organ failure Resiratory 72 (93) 74 (96) (93) 0.634/ (98) 0.064/0.260 Cardiac 16 (20) 11 (14) (18) 0.486/ (18) 0.377/0.307 Renal 8 (10) 6 (7) (4) 0.184/ (9) 0.461/0.487 Neurologic 3 (3) 2 (2) (12) 0.041/ (5) 0.298/0.292 Digestive 3 (3) 4 (5) (6) 0.422/ (3) 0.614/0.433 Primary diagnosis Acute exacerbation of COPD 57 (74) 61 (79) (79) 0.333/ (71) 0.405/0.139 Pneumonia 6 (7) 5 (6) (8) 0.582/ (13) 0.150/0.086 Congestive heart failure 3 (3) 2 (2) (2) 0.502/ (3) 0.507/0.614 Others 11 (14) 9 (11) (10) 0.367/ (12) 0.394/0.564 During hositalization Corticosteroid use 42 (54) 38 (49) (43) 0.161/ (51) 0.410/0.417 Duration of antibiotic treatment, d / /0.038 *Results by univariate analysis. Data are resented as mean SD or No. (%) unless otherwise indicated. with unconfirmed VAP vs case atients/atients with unconfirmed VAP vs control atients. VAT atients vs case atients/vat atients vs control atients Clinical Investigations in Critical Care

4 ceived by VAP atients during ICU stay was mg/kg/d (equivalent to mg/d of hydrocortisone in a 60-kg atient); mean duration of corticosteroid treatment was days. Microbiologic Results One hundred one bacteria were associated ( 10 6 cfu/ml) with the 77 first VAP eisodes. P aeruginosa (31%), A baumannii (19%), and S aureus (14%) were the most frequently isolated microorganisms (Table 2). VAP was olymicrobial in 23 atients (29%). VAP was related to MRB in 32 atients (41%). Tracheal asirate culture findings were ositive at 10 4 to 10 5 cfu/ml in 7 of 48 atients (14%) with unconfirmed VAP. Outcome of Case and Control Mortality rate, duration of MV, and length of ICU stay were significantly higher in case atients than in control atients (Table 3). No significant relationshi was found between outcome of VAP atients and adequacy of initial antimicrobial treatment, MRB, or surgery (Table 4). However, mortality rate, duration of MV, and ICU stay were significantly lower in VAP atients who received corticosteroids during their ICU stay comared with those who did not receive corticosteroids (Table 4). Among the 42 VAP atients who received corticosteroids in the ICU, 34 atients received long-term corticosteroid treatment, and 8 atients received only short-term corticosteroid treatment. In VAP atients with long-term and short-term corticosteroid treatment comared with VAP atients who did not receive corticosteroids, the mortality rate was significantly lower (21 of 34 atients [61%] vs 29 of 35 atients [82%]; 0.045), and durations of MV (23 16dvs27 16 d, 0.182) and ICU stay (25 18dvs31 18 d, 0.109) were similar. In VAP atients with only short-term corticosteroid treatment comared with VAP atients who did not receive Table 2 Bacteria Associated With 77 Eisodes of VAP Microorganisms No. (%) Gram-negative 82 (81) P aeruginosa 32 (31) A baumannii 20 (19) Serratia secies 8 (7) Escherichia coli 6 (5) Enterobacter secies 5 (4) Proteus secies 4 (3) S maltohilia 4 (3) Klebsiella secies 3 (2) Gram-ositive 19 (18) Methicillin-resistant S aureus 8 (7) Methicillin-sensitive S aureus 7 (6) Stretococcus neumoniae 4 (3) corticosteroids, the mortality rate (0 of 8 atients [0%] vs 29 of 35 atients [82%], 0.001) and duration of MV (14 11dvs27 16 d, 0.039) and ICU stay (15 11dvs31 18 d, 0.006) were significantly lower. In VAP atients with only short-term corticosteroid treatment comared with VAP atients receiving long-term and short-term corticosteroids, resectively, the mortality rate (0 of 8 atients [0%] vs 21 of 34 atients [61%], 0,002) and duration of MV (14 11dvs23 16 d, 0.008) and ICU stay (15 11dvs25 18 d, 0.001) were significantly lower. No significant difference was found between VAP atients who received corticosteroids during their ICU stay comared with those who did not receive corticosteroids with regard to adequacy of initial antibiotic treatment (31 of 42 atients [73%] vs 27 of 35 atients [77%], 0.473) and MRB rates (19 of 42 atients [42%] vs 13 of 35 atients [37%], 0.314). Factors Associated With ICU Mortality Renal, cardiac, and neurologic failures on ICU admission, VAP, and VAP related to MRB were associated with ICU mortality in univariate analysis (Table 5). Only VAP was indeendently associated with ICU mortality (OR, 7.76; 95% CI, 3.2 to 18.6; 0.001). Characteristics and Outcome of Excluded excluded for unconfirmed VAP were significantly younger than case atients and control atients, as well as atients excluded for VAT comared with case atients and control atients. SAPS II was significantly lower in VAT atients comared with control atients. The rate of rior antibiotic use was significantly higher in atients with unconfirmed VAP comared with control atients, and in VAT atients comared with control atients. Neurologic failure was more frequent in atients with unconfirmed VAP comared with case atients and control atients. The duration of antibiotic treatment during ICU stay was longer in atients with unconfirmed VAP comared with control atients, and in VAT atients comared with control atients. Other atient characteristics were similar in excluded atients comared with case atients and control atients (Table 1). No significant difference in outcome was found between atients excluded for unconfirmed VAP and case atients. Mortality rate, duration of MV, and ICU stay were significantly higher in atients excluded for unconfirmed VAP comared with control atients (Table 3). In atients with unconfirmed VAP who had ositive tracheal asirate culture findings (10 4 to 10 5 cfu/ml) comared with those who had negative culture findings, mortality rates (4 of 7 atients [57%] vs 26 of 41 atients [63%], CHEST / 128 / 3/ SEPTEMBER,

5 Table 3 Outcomes of Study and Excluded * Excluded Variables Cases (n 77) Controls (n 77) Value Unconfirmed VAP (n 48) Value VAT (n 133) Value MV duration, d / / Duration of ICU / / stay, d ICU mortality 50 (64) 22 (28) (62) 0.123/ (45) 0.001/0.013 *Results by univariate analysis. Data are resented as mean SD or No. (%) unless otherwise indicated. with unconfirmed VAP vs case atients/atients with unconfirmed VAP vs control atients. VAT atients vs case atients/vat atients vs control atients ), duration of MV (23 13dvs22 12 d, 0.491), and duration of ICU stay (25 14dvs d, 0.489) were similar. In atients with unconfirmed VAP who had ositive tracheal asirate culture findings (10 4 to 10 5 cfu/ml) comared with VAP atients, resectively, mortality rates (4 of 7 atients [57%] vs 50 of 77 atients [64%], 0.389), duration of MV (23 13dvs24 15 d, 0.461), and length of ICU stay (25 14dvs d, 0.459) were similar. Although the duration of MV and ICU stay were similar in VAT atients comared with case atients, the mortality rate was significantly lower in atients with VAT comared with case atients. Mortality rate, duration of MV, and ICU stay were significantly higher in VAT atients comared with control atients (Table 3). Table 4 Outcomes of Subgrous of With VAP* Variables, No. ICU Mortality MV Duration, d ICU Length of Stay, d Initial antimicrobial treatment Adequate (63) Inadequate (68) VAP related to MRB Yes (68) No (62) Surgical atients Yes 8 5 (62) No (65) Corticosteroid use during ICU stay Yes (50) No (82) *Data are resented as No. (%) of mean SD unless otherwise indicated by univariate analysis (atients who received corticosteroid treatment during ICU stay vs atients who did not receive corticosteroids). Discussion Our results demonstrate that VAP is associated with higher mortality rates and longer duration of MV and ICU stay in COPD atients. In addition, VAP is indeendently associated with ICU mortality. Our results also suggest that low doses of corticosteroids are associated with a lower mortality rate and shorter duration of MV and ICU stay in COPD atients with VAP. To our knowledge, this case-control study is the first to evaluate the imact of VAP on mortality and morbidity in COPD atients. Previous studies 8,11 13 conducted in heterogeneous ICU oulations, including some atients with COPD, highlighted the link between VAP and increased mortality and morbidity. However, VAP may increase mortality in severely ill atients but may be more likely to develo in sicker atients, with inherently higher Table 5 Risk Factors for ICU Mortality in With and Without VAP Variables (n 154), No. ICU Mortality, No. (%) Value* OR (95% CI) Renal failure Yes ) (1 14) No (45) VAP due to MRB Yes (71) ( ) No (40) VAP Yes (64) ( ) No (28) Cardiac failure Yes (59) (1 6.2) No (44) Neurologic failure Yes 5 5 (100) ( ) No (44) *Results by univariate analysis Clinical Investigations in Critical Care

6 mortality rates. 14 Several studies 2 5,12,13,15 27 have investigated the comlex relationshi between VAP and mortality using different diagnostic criteria and different statistical methods in different oulations. Multivariate analyses, case-control studies, and randomized controlled studies of an effective revention of VAP were used to determine the imact of VAP on mortality and morbidity. VAP has been identified as an indeendent risk factor for mortality by numerous studies. 13,15,16 However, other studies 12,17 failed to identify VAP as an indeendent risk factor for mortality. The selection and exclusion of otential risk factors, and the different atient characteristics may exlain these conflicting results. Three of eight case-control studies 3 5 concluded that VAP was associated with significantly increased mortality. However, the mortality rates did not differ significantly between atients with VAP and matched control subjects in the five other case-control studies. 2,18 21 In the largest study 2 on VAP conducted in the United States, 842 atients with VAP were matched with 2,243 atients without VAP according to duration of MV, severity of illness on hosital admission, tye of hosital admission, and age. The mortality rates were similar in case atients and control atients. However, atients with VAP occurring 24 h after intubation were included. In addition, the mean interval between intubation and the identification of VAP was 3.3 days. Moreover, rates of COPD atients included in that study could not be determined. Randomized studies showed a reduction in VAP rates using continuous asiration of subglottic secretions, selective digestive decontamination, semirecumbent osition, and chest hysiotheray. A reduction in VAP rates in these studies was not associated with reduction in mortality rates, suggesting the absence of mortality attributable to VAP. However, a randomized trial 27 assessed the efficacy of NIV in atients with weaning failure, and found a significant reduction in nosocomial neumonia and mortality rates in atients with NIV comared with the conventional weaning grou. These results suggest that invasive ventilation withdraw is associated with a reduction in VAP incidence, resulting in lower mortality rate. The relationshi between VAP and mortality is influenced by several factors: adequacy of initial antimicrobial treatment, microorganisms resonsible for VAP, severity of illness, comorbid factors, and host resonse factors. In our study, the mortality rate attributable to VAP was 34%. This high rate could be exlained by the following: (1) advanced age, (2) high rates of VAP related to MRB and inadequate initial antibiotic treatment, (3) high rate of late-onset VAP, and (4) antecedent of COPD. Advanced age, VAP related to MRB, inadequate initial antibiotic treatment, and late-onset VAP have reviously been reorted to be associated with higher mortality rates in ICU atients. 3,28 30 COPD is associated with increased mortality rates in VAP atients. 9 In atients with structural lung disease, nosocomial bronchoulmonary infection may result in increased mortality and morbidity. However, in atients with community-acquired neumonia, structural lung disease was not associated with increased mortality. 31 In this study, VAP was associated with longer duration of MV and ICU stay. These results are consistent with those of other case-control studies. 2,3,5,18 21 In order to reduce mortality and morbidity in COPD atients with VAP, romt and adequate initial antibiotic treatment should be used. 6 In addition, revention measures should be alied in intubated COPD atients to reduce VAP incidence. 32 In COPD atients with VAP, corticosteroid use during ICU stay, esecially short-term use in atients without long-term corticosteroid use, was associated with lower mortality rate and shorter duration of MV and ICU stay. These results are lausible given that increased inflammatory resonse is the main redictor of nonresonse and mortality in atients with VAP. 33 In addition, a recent randomized controlled study 34 conducted in atients with severe community-acquired neumonia found a significant reduction in mortality rate in atients who received low doses of hydrocortisone comared with those who received lacebo. However, corticosteroid use is the major indeendent risk factor for ICUacquired aresis. 35 Furthermore, corticosteroid use could be harmful in atients with immunoaralysis and uncontrolled sources of infection. The beneficial effect of corticosteroids found in our study was a secondary outcome, and further randomized trials are necessary to determine the effect of corticosteroids on outcome of COPD atients with VAP. Our study has several limitations. First, this was a single-center study. Therefore, our results may not be generalizable to atients in other ICUs. Second, some of the trends observed in VAP subgrou analyses could have reached statistical significance if the number of included atients had been larger. Third, invasive rocedures were not used to diagnose VAP. However, quantitative tracheal asirate was erformed in all atients with a threshold of 10 6 cfu/ml. Postmortem studies 36 demonstrated an accetable overall diagnostic accuracy of quantitative tracheal asirate as comared with BAL or rotected secimen brush. Fourth, information on severity of COPD was not available. Finally, exclusion of atients with unconfirmed VAP and atients with VAT may be debatable. Because clinical and radiograhic signs of VAP are not secific, we used clinical, radiowww.chestjournal.org CHEST / 128 / 3/ SEPTEMBER,

7 grahic, and quantitative microbiologic criteria to define VAP. A high rate of rior antibiotic use (70%) may have resulted in negative culture findings of tracheal asirate in some atients with unconfirmed VAP. However, the characteristics and outcomes of all excluded atients were carefully evaluated. We conclude that VAP is associated with increased mortality and longer duration of MV and ICU stay in COPD atients. Efforts should be made to reduce VAP incidence in these atients and to initiate romt and adequate antibiotic treatment when VAP is susected. References 1 Richards MJ, Edwards JR, Culver DH, et al. Nosocomial infections in medical intensive care units in the United States: National Nosocomial Infections Surveillance System. Crit Care Med 1999; 27: Rello J, Ollendorf DA, Oster G, et al. Eidemiology and outcomes of ventilator-associated neumonia in a large US database. Chest 2002; 122: Bercault N, Boulain T. Mortality rate attributable to ventilator-associated nosocomial neumonia in an adult intensive care unit: a rosective case-control study. Crit Care Med 2001; 29: Cunnion KM, Weber DJ, Broadhead WE, et al. Risk factors for nosocomial neumonia: comaring adult critical-care oulations. Am J Resir Crit Care Med 1996; 153: Fagon JY, Chastre J, Hance AJ, et al. Nosocomial neumonia in ventilated atients: a cohort study evaluating attributable mortality and hosital stay. Am J Med 1993; 94: Niederman MS. Can otimal management revent mortality in ventilator-associated neumonia? Crit Care Med 2002; 30: Nseir S, Di Pomeo C, Pronnier P, et al. Nosocomial tracheobronchitis in mechanically ventilated atients: incidence, aetiology and outcome. Eur Resir J 2002; 20: Torres A, Aznar R, Gatell JM, et al. Incidence, risk, and rognosis factors of nosocomial neumonia in mechanically ventilated atients. Am Rev Resir Dis 1990; 142: Rello J, Ausina V, Ricart M, et al. Imact of revious antimicrobial theray on the etiology and outcome of ventilator-associated neumonia. Chest 1993; 104: American Thoracic Society. Standards for the diagnosis and care of atients with chronic obstructive ulmonary disease. Am J Resir Crit Care Med 1995; 152(sul):S77 S Tejada Artigas A, Bello DS, Chacon VE, et al. Risk factors for nosocomial neumonia in critically ill trauma atients. Crit Care Med 2001; 29: Kollef MH. Ventilator-associated neumonia: a multivariate analysis. JAMA 1993; 270: Vincent JL, Bihari DJ, Suter PM, et al. The revalence of nosocomial infection in intensive care units in Euroe: results of the Euroean Prevalence of Infection in Intensive Care (EPIC) Study; EPIC International Advisory Committee. JAMA 1995; 274: Carlet J. Dying from or with a nosocomial neumonia in the intensive care unit? Crit Care Med 2001; 29: Fagon JY, Chastre J, Vuagnat A, et al. Nosocomial neumonia and mortality among atients in intensive care units. JAMA 1996; 275: Timsit JF, Chevret S, Valcke J, et al. Mortality of nosocomial neumonia in ventilated atients: influence of diagnostic tools. Am J Resir Crit Care Med 1996; 154: Craven DE, Kunches LM, Kilinsky V, et al. Risk factors for neumonia and fatality in atients receiving continuous mechanical ventilation. Am Rev Resir Dis 1986; 133: Leone M, Bourgoin A, Giuly E, et al. Influence on outcome of ventilator-associated neumonia in multile trauma atients with head trauma treated with selected digestive decontamination. Crit Care Med 2002; 30: Paazian L, Bregeon F, Thirion X, et al. Effect of ventilatorassociated neumonia on mortality and morbidity. Am J Resir Crit Care Med 1996; 154: Baker AM, Meredith JW, Haonik EF. Pneumonia in intubated trauma atients: microbiology and outcomes. Am J Resir Crit Care Med 1996; 153: Heyland DK, Cook DJ, Griffith L, et al. The attributable morbidity and mortality of ventilator-associated neumonia in the critically ill atient: The Canadian Critical Trials Grou. Am J Resir Crit Care Med 1999; 159: Valles J, Artigas A, Rello J, et al. Continuous asiration of subglottic secretions in reventing ventilator-associated neumonia. Ann Intern Med 1995; 122: D Amico R, Pifferi S, Leonetti C, et al. Effectiveness of antibiotic rohylaxis in critically ill adult atients: systematic review of randomised controlled trials. BMJ 1998; 316: Drakulovic MB, Torres A, Bauer TT, et al. Suine body osition as a risk factor for nosocomial neumonia in mechanically ventilated atients: a randomised trial. Lancet 1999; 354: Ntoumenooulos G, Presneill JJ, McElholum M, et al. Chest hysiotheray for the revention of ventilator-associated neumonia. Intensive Care Med 2002; 28: Kirschenbaum L, Azzi E, Sfeir T, et al. Effect of continuous lateral rotational theray on the revalence of ventilatorassociated neumonia in atients requiring long-term ventilatory care. Crit Care Med 2002; 30: Ferrer M, Esquinas A, Arancibia F, et al. Noninvasive ventilation during ersistent weaning failure: a randomized controlled trial. Am J Resir Crit Care Med 2003; 168: Esteban A, Anzueto A, Frutos F, et al. Characteristics and outcomes in adult atients receiving mechanical ventilation: a 28-day international study. JAMA 2002; 287: Kollef MH, Sherman G, Ward S, et al. Inadequate antimicrobial treatment of infections: a risk factor for hosital mortality among critically ill atients. Chest 1999; 115: Kollef MH, Silver P, Murhy DM, et al. The effect of late-onset ventilator-associated neumonia in determining atient mortality. Chest 1995; 108: Arancibia F, Bauer TT, Ewig S, et al. Community-acquired neumonia due to gram-negative bacteria and Pseudomonas aeruginosa: incidence, risk, and rognosis. Arch Intern Med 2002; 162: Dodek P, Keenan S, Cook D, et al. Evidence-based clinical ractice guideline for the revention of ventilator-associated neumonia. Ann Intern Med 2004; 141: Ioanas M, Ferrer M, Cavalcanti M, et al. Causes and redictors of nonresonse to treatment of intensive care unit-acquired neumonia. Crit Care Med 2004; 32: Confalonieri M, Urbino R, Potena A, et al. Hydrocortisone infusion for severe community-acquired neumonia: a reliminary randomized study. Am J Resir Crit Care Med 2005; 171: De Jonghe B, Sharshar T, Lefaucheur JP, et al. Paresis acquired in the intensive care unit: a rosective multicenter study. JAMA 2002; 288: Nseir S, Marquette CH. Diagnosis of hosital-acquired neumonia: ostmortem studies. Infect Dis Clin North Am 2003; 17: Clinical Investigations in Critical Care