VENTILATOR-ASSOCIATED pneumonia (VAP) is the

Transcription

1 Ventilator-Associated Pneumonia: Incidence, Risk Factors, Outcome, and Microbiology Mandakini Pawar, BSc, Yatin Mehta, MD, DNB, FRCA, FAMS, Poonam Khurana, MD, Anshumali Chaudhary, MD, Vinay Kulkarni, MD, and Naresh Trehan, MD Objective: To determine the incidence, risk factors, outcome, and pathogens of ventilator-associated pneumonia (VAP) in a cardiac surgical intensive care unit (ICU). Design: Prospective study. Setting: Escorts Heart Institute and Research Centre, New Delhi, India. Participants: Nine hundred fifty-two consecutive patients undergoing cardiac operations who received intermittent positive-pressure ventilation (IPPV). Interventions: All patients were assigned into VAP (n 25) and non-vap (n 927) groups. Measurement and Main Results: Risk factors and other variables were analyzed with univariate and multivariate analysis. Of the 952 patients studied, 25 (2.6%) had VAP. On univariate analysis, significant risk factors were emergency surgery, chronic obstructive pulmonary disease (COPD), reintubation, coma, steroid treatment, intra-aortic balloon counterpulsation (IABC), enteral feedings, tracheostomy, acute physiology, age, and chronic health evaluation (APACHE II) score, prior antibiotics, and IPPV hours. On multivariate analysis, IPPV hours ( v VENTILATOR-ASSOCIATED pneumonia (VAP) is the most common nosocomial infection in the intensive care unit (ICU). It is a pulmonary infection that occurs after at least 48 hours of intermittent positive-pressure ventilation (IPPV), and is a leading cause of morbidity and mortality. The incidence of VAP ranges from 10% to 65% of intubated patients depending on the risk factors. It is associated with an attributable mortality of up to 13% and approaches 55% when VAP is commonly caused by antibiotic-resistant nosocomial organisms like Psuedomonas aeruginosa, Escherichia coli, Klebsiella pneumoniae, Acinetobacter species, and Staphylococcus species. 1 Most episodes of VAP are thought to develop from the aspiration of oropharyngeal secretions containing potentially pathogenic organisms. Aspiration of gastric secretions may also contribute but to a lesser degree. Tracheal intubation interrupts the body s anatomic and physiologic defenses against aspiration, making IPPV a major risk factor for VAP. 2 Patients who develop VAP after cardiac surgery require prolonged IPPV and hence longer ICU and hospital stays and further increase the cost of health care. Prevention of this nosocomial infection (variation in patient positioning, continuous aspiration of subglottic secretions, selective digestive tract decontamination, the use of stress ulcer prophylaxis) 3,4 and understanding the risk factors of VAP would help in decreasing the incidence of VAP. Good management strategies for VAP From the Department of Anaesthesiology and Microbiology, Escorts Heart Institute and Research Centre, New Delhi, India. Address reprint requests to Yatin Mehta, MD, DNB, FRCA, FAMS, Department of Anaesthesiology, Escorts Heart Institute and Research Centre, New Delhi , India. yatinmehta@hotmail.com Copyright 2003, Elsevier Science (USA). All rights reserved /03/ $30.00/0 doi: /jcan ; p < 0.001) and steroids (20% v 0%; p < 0.001) were independent predictors of VAP. The most common pathogens isolated were Pseudomonas aeruginosa (22), Escherichia coli (10), Klebsiella pneumoniae (4), Staphylococcus species (4), and Acinetobacter species (2). The mortality rate in VAP was 16% as compared with 0.2% in non-vap cases (p < 0.001). Conclusion: These data suggest that by univariate analysis the risk factors for VAP were emergency surgery, COPD, reintubation, coma, steroid treatment, IABC, enteral feedings, tracheostomy, APACHE II score, prior antibiotics, and IPPV hours. On multivariate analysis, only IPPV hours and steroids were independent predictors of VAP. Pseudomonas aeruginosa is the most common pathogen associated with VAP, and the mortality is increased with VAP. Copyright 2003, Elsevier Science (USA). All rights reserved. KEY WORDS: Ventilator-associated pneumonia (VAP), intermittent positive-pressure ventilation (IPPV), steroids, cardiac surgery like early and accurate diagnosis and more specific antimicrobial use may significantly improve patients outcome. 1,5 MATERIALS AND METHODS Approval of the institutional review board and informed consent from the patient were obtained for this prospective study conducted at Escorts Heart Institute and Research Centre between July 27, 2001, and October 31, 2001, to determine the incidence, risk factors, outcome, and pathogens of VAP in the cardiac surgical ICU. During this period, 952 adult patients who were undergoing cardiac operations and were admitted to the ICU on IPPV were included in this prospective study. They consisted of VAP (n 25) and non-vap (n 927) patient groups. Pulmonary infection was defined as VAP if the onset was recorded after at least 48 hours of IPPV and if the following criteria were present. 1. Chest radiography showing a new and persistent infiltrate consistent with pneumonia. 2. Purulent sputum. 3. Elevated temperature 38 C or leukocyte count 10,000/mL or 1,000/mL. 4. Micro-organisms isolated from at least one of the following samples: (1) bronchoalveolar lavage, (2) endotracheal aspirate, or (3) sputum. Patients were considered not having VAP when at least one clinical sign of VAP was absent with no positive culture and no curative antibiotic was prescribed. The patients were routinely administered stress ulcer prophylaxis (ranitidine, 50 mg IV 3 times a day). All patients received intravenous antibiotic prophylaxis of cefazolin, 1 g twice a day, and diabetic patients received gentamicin, 1 mg/kg 3 times a day, along with cefazolin, 1 g twice a day for 36 hours. During the study, the following risk factors were recorded. Patients age, gender, IPPV hours, reintubation, underlying illness, acute physiology, chronic health (APACHE II) score, type of surgery (elective/emergency), redo operations, enteral nutrition, total ICU stay, steroid treatment, total postoperative stay, intra-aortic balloon counterpulsation (IABC) used, tracheostomy, and antibiotic usage were recorded. 22 Journal of Cardiothoracic and Vascular Anesthesia, Vol 17, No 1 (February), 2003: pp 22-28

2 VENTILATOR-ASSOCIATED PNEUMONIA 23 Table 1. Comparison of VAP and Non-VAP Cases (Univariate Analysis) Variable VAP (n 25) Non-VAP (n 927) p Value Age (y) NS Gender (%) Male 21 (84) 818 (88.2) Female 4 (16) 109 (11.8) NS Diabetes (%) 12 (48) 370 (39.9) NS Smoking (%) 4 (16) 164 (17.7) NS Bronchial Asthma (%) 2 (8) 24 (2.6) NS Obesity (%) 5 (20) 86 (9.3) NS Chronic Bronchitis (%) 7 (0.8) NS Note. p value indicates comparison between VAPs and non-vaps. Abbreviation: NS, not statistically significant Univariate analysis was used to identify factors with significant unadjusted effects on VAP. Multiple logistic regression then was applied to determine the significance or independent effects of such variables on VAP. All statistical tests were 2-tailed, and the level of signficance was set at 0.05%. RESULTS Of 952 patients in the study, 25 (2.6%) had VAP. The mean interval between admission to the ICU and the diagnosis of VAP was days. The median interval was 4 days. More than 90% of VAP cases occurred within 10 days of admission to the ICU. Table 1 shows the results from the univariate analysis for 8 variables that were not significantly associated with VAP. Mean age, gender, diabetes, smoking, bronchial asthma, obesity, and chronic bronchitis were not significantly different between VAP and non-vap groups. However, the univariate analysis (Table 2) showed emergency surgery; COPD; tracheostomy; reintubation; coma; steroid treatment; APACHE II score; number of antibiotics before VAP, IPPV, and IABC; and enteral feedings were significantly associated with VAP cases. Total ICU stay and total postoperative stay were significantly higher among patients with VAP than those without VAP (p 0.001). Out of 952 operations, 25 were redo operations. Two redo operations had VAP (n 8%), and 23 redo operations did not develop VAP (2.5%; p 0.089). Multivariate analysis took into account the variables that could be the risk factors for VAP. This stepwise logistic regression showed that steroids and IPPV were independently associated with VAP (Table 3). A total of 45 respiratory tract specimens were received for culture and sensitivity from the 25 VAP cases. These specimens included endotracheal aspirate (15), bronchoalveolar lavage (12), and sputum (18). Infection was monomicrobial in 48% of patients and polymicrobial in 52% of patients. Gramnegative organisms accounted for 84% of the cases of pneumonia and a combination of gram-positive and gram-negative accounted for 16%. The organism recovered from the specimens were identified on mini-api Biomeriux (A-10; Green Park, New Delhi, India) identification system. The common pathogens isolated were P aeruginosa, E coli, K pneumoniae, Acinetobacter species, and Staphylococcus species (Tables 4 and 5). Antimicrobial susceptibility was performed on mini-api Biomeriux for all the organisms isolated. Antibiograms of grampositive bacteria are shown in Fig 1. The majority of staphylococci were oxacillin resistant (75%) but susceptible to vancomycin and teicoplanin (100%). Of all staphylococci, 75% of isolates were susceptible to clindamycin and rifampicin. Most of the isolates had poor susceptibility toward penicillins, -lactams, cephalosporins, quinolones, and aminoglycosides. Figs 2 to 4 show the susceptibility pattern of gram-negative isolates. Imipenem (91.89%) and amikacin (86.48%) showed the best activities against these isolates. E coli (91.3%) and K pneumoniae (100%) were highly susceptible to piperacillin plus Tazobactum, but P aeruginosa (45%) showed poor susceptibilty to it. Ceftazidine showed the best activity against P aeruginosa (80%) and K pneumoniae (75%), but was ineffective against E coli (0% susceptibility). Acinetobacter was resistant to all the routinely used antibiotics (0% susceptibility). DISCUSSION The authors investigated the incidence, risk factors, outcome, and pathogens of VAP in the cardiac surgical ICU. The estimated prevalence of nosocomial pneumonia in intensive care units ranges from 10% to 65% with mortality rates of 13% to 55%. 1 Nosocomial pneumonia was diagnosed in 25 (2.6%) of the 952 patients in this study. The incidence of 2.6% in cardiothoracic patients is comparable to that reported in another study. 6 The supine position of mechanically ventilated patients increases the risk of VAP 3-fold as compared with patients in a semirecumbent position. 7 In this series, 44% of patients with VAP had IABC, in whom it is difficult to change the position of the patient. This was 5.6% in non-vap cases. This could partly explain the higher incidence in patients with IABC. Table 2. Comparison of VAP and Non-VAP Cases (Univariate Analysis) Variable VAP (n 25) (%) Non-VAP (n 927) (%) p Value Emergency surgery 7 (28) 21 (2.3) Redo surgery 2 (8.0) 23 (2.5) (NS) COPD 3 (12) 21 (2.3) 0.05 Reintubation 9 (36) Coma 1 (4) 0.05 Steroid treatment 5 (20) 0.001) IABC 11 (44) 52 (5.6) Enteral feedings 9 (36) 2 (0.2) Tracheostomy 9 (36) APACHE II score Prior antibiotic (nos) Total ventilation hrs Total ICU stay Hospital stay Mortality 4 (16) 2 (0.2) Note. Data are shown as mean SD or number of patients; p value indicates comparison between VAPs and non-vaps. Abbreviation: COPD, chronic obstructive pulmonary disease; IABC, intra-aortic balloon counterpulsation; APACHE, acute physiology, age and chronic health evaluation; NS, not significant.

3 24 PAWAR ET AL Table 3. Predictors for VAP on Multivariate Logistic Regression Analysis Variables B SE WALD 2 p Value Exp (B) 95% CI of Exp (B) Steroid Ventilation (d) Note. p value indicates comparison between VAPs and non-vaps. Abbreviations: B, Beta; SE, Standard error; WALD 2, Chi-square; Exp, Exponent; CI, Confidence interval. The supine position can be used as a disease severity marker and directly contributes to mortality by increasing the incidence of VAP. 8 Selective digestive decontamination has been proposed in critical patients as class A treatment to avoid oropharyngeal colonization by the American College of Chest Physicians and the Society of Critical Care Medicine consensus conference. 3,4 Pneumonia develops in up to 40% of patients in ICUs, 9 but debate exists over whether the high mortality associated with such pneumonia is because of the pneumonia (attributable mortality) or whether the pneumonia is simply a part of the general downhill course of seriously ill patients. In a cardiothoracic ICU, the crude mortality rate of VAP was calculated as 30%. 6 In this series of 952 patients, 6 patients died. Out of 6 patients, 4 patients had VAP, which was a contributory factor of mortality (out of a total 25 VAP cases). The mortality rate in VAP was 16% as compared with 0.2% (p 0.001) in non-vap cases. This finding agrees with the study done by Fagon and colleagues 9 who have concluded that nosocomial pneumonia independently contributed to ICU-related mortality. Other studies, however, cast some doubt on this conclusion. For example, in a case-control study, patients with VAP (many of whom had experienced trauma) were matched for other variables with 85 controls. Mortality was 40% for the patients with pneumonia and 38.8% for controls. Therefore, the question of whether nosocomial pneumonia is an attributable cause of death in the ICU remains unanswered; moreover, the answer may differ among different types of patients in the ICU. In this study, 48% patients had monomicrobial infection, and 52% patients had polymicrobial infection. The VAP was more with gram-negative bacteria (84%) than with mixed grampositive and gram-negative bacteria infections (16%). It is well documented that the higher rates of infection and mortality among ICU patients are mostly related to factors such as exposure to invasive procedures, underlying disease conditions, duration of stay in the ICU, infection sites, and association with nosocomial multidrug-resistant pathogens This study included types of organisms and their susceptibility to commonly used antibacterial agents. As reported in other studies, 6,15 the most common gram-negative isolates in VAP patients (n 25) were P aeruginosa (88%), E coli (40%), Klebseilla species (16%), and Acinetobacter species (80%). Numerous studies have shown that multidrug-resistant bacteria, in particular aerobic gram-negative bacteria, easily colonize the gastrointestinal tract and respiratory tract of hospitalized patients. 13,16 In addition, it is well known that multidrugresistant bacteria are becoming increasingly prevalent in the hospital environment as a result of the extensive use of antibiotics. 11,12,17,18 In VAP cases, the usage of antibiotics before the diagnosis of VAP was significantly higher as compared with the non-vaps, and justification for this lies in the fact that the patients were critically ill (as APACHE II score for VAPs) necessitating empiric therapy before the results of culture and sensitivity are known. Antimicrobial susceptibility is a hospital- and region-dependent matter (eg, P aeruginosa is many regions in Europe and the United States is not sensitive enough to permit aztreonam as the first choice). The results of this study revealed that a few types of multidrug-resistant, gram-negative bacteria were Acinetobacter species, P aeruginosa, and E coli. However, excellent activity against all gram-negative isolates was shown for imipenem (91.89%) and amikacin (86.48%). Neither of these drugs has been extensively used in the treatment of these patients. These findings suggest that the most important strategies for controlling the problem of multidrug-resistant organisms in the ICU should be directed toward continuously monitoring the presence of these organisms and the avoidance of excessive or continued use of any single drug over a long period of time. Table 4. Micro-organisms Number (%) Recovered From VAP Cases Micro-organism recovered Total No. of Organisms Recovered 42 (%) of Organisms Recovered Gram-negative Pseudomonas aeruginosa 22 (52.3) Escherichia coli 10 (23.8) Klebsiella pneumoniae 4 (9.5) Acinetobacter species 2 (4.7) Gram-positive Staphylococcus species 4 (9.5) (Staphylococus epidermidis Ox-S-1, S haemolytics Ox-R-3) Abbreviations: Ox, Oxacillin; S, sensitive; R, Resistant. Table 5. Patients Number (%) Organism Recovered Micro-organism Recovered Total No. of VAP 25 (%) of Patients From Whom Organism Was recovered Gram-negative 22 (88) Pseudomonas aeruginosa Escherichia coli 10 (40) Klebsiella pneumoniae 4 (16) Acinetobacter species 2 (8) Gram-positive Staphylococcus sp 4 (16) (Staphylococcus epidermidis Ox-S-I, S haemolytics Ox-R-3) Abbreviations: Ox, Oxacillin; S, sensitive; R, Resistant.

4 VENTILATOR-ASSOCIATED PNEUMONIA 25 Fig 1. Percentage of susceptible gram-positive isolates (Staphylococcus species) from VAP cases. Fig 2. Percentage of susceptible gram-negative isolates (K pneumoniae) from VAP cases.

5 26 PAWAR ET AL Fig 3. Percentage of susceptible gram-negative isolates (P aeruginosa) from VAP cases. Fig 4. Percentage of susceptible gram-negative isolates (E coli) from VAP cases.

6 VENTILATOR-ASSOCIATED PNEUMONIA 27 On univariate analysis variables like age, gender, diabetes, smoking, bronchial asthma, obesity, chronic bronchitis, and redo operations did not have any significant association with VAP. However, some studies 19,20 have shown a significant association of age and VAP. In this study, the mean age for VAP was and, for non-vap, (p 0.238). This may have accounted for differences in findings. Some studies have suggested that tracheostomy is associated with an increased risk of VAP This study has shown a significant relation between tracheostomy and VAP, but it is not clear whether it is the occurrence of VAP that leads to tracheostomy or vice versa. However, because in this study most cases of VAP (88.88%) occur (ie, mean 3.5 days of IPPV) before the tracheostomy, it is concluded that tracheostomy could be a result of VAP rather than a risk factor of VAP. Since the early clinical reports of IABC, the frequency of use has increased in the course of cardiac surgery for treatment of low cardiac output and cardiogenic shock. 24 The patients who are on IABC are hemodynamically unstable and may require long-term ventilatory support. In this study, it was found that longer duration IPPV is a risk factor for developing VAP, which is comparable to other studies. 5,24,25 Emergency operations for acute myocardial infarction or percutaneous transluminal coronary angioplasty failure also contribute to postoperative morbidity 8 because the patients who undergo emergency operations are hemodynamically unstable and require longer ventilatory support. This explains its significant association with VAP. After cardiac surgery, immunosuppression caused by extracorporeal circulation, anesthetics, transfusion of blood products, body position, and mechanical ventilation favors the development of pulmonary infection as a potential risk. 20,26 In this study, it was found that 20% of patients with VAP were on preoperative steroids as compared with none in non-vap patients. On multivariate analysis, steroids had significant independent (or adjusted) effect on VAP. So, it can be concluded that patients receiving steroids are more prone to develop VAP. Enteral feeding was found to be significantly associated with VAP in this study, which is comparable to another study. 5 It has been suggested that an early initiation of enteral feeding can help to maintain the gastrointestinal epithelium and prevent bacterial translocation, but it may also increase the risk of gastric distention, colonization, aspiration, and pneumonia. Accurate assessment of nutritional status and avoidance of unnecessary enteral nutrition may help to reduce the risk of nosocomial pneumonia. Patients who develop VAP require longer ICU and hospital stays. 5,27-30 In this study, total ICU stay for VAP was significantly higher as compared with non-vap ( v ; p 0.001). Patients who were reintubated and were on prolonged IPPV are at a higher risk for VAP. 5,25,30 In this study univariate analysis has shown that COPD is a significant indicator of VAP. The risk factor was similarly identified by Celis et al 28 in a multivariate analysis of nosocomial pneumonia in nonmechanically ventilated patients and in a study of pneumonia in a a respiratory ICU and by Rello et al. 29 The loss of mucosal clearance predisposes patients with COPD to infection, probably by all gram-negative bacilli. 29 In summary, ventilator-associated pneumonia is a leading cause of morbidity and mortality in ICU patients, leading to lengthened ICU and hospital stays and higher health care costs. The mortality caused by VAP increases if it is caused by resistant bacteria. On univariate analysis significant risk factors were emergency surgery, COPD, reintubation, coma, steroid treatment, IABC, enteral feedings, tracheostomy, APACHE II score, prior antibiotic number, and IPPV hours. Other significant variables were mortality, total ICU stay, and total hospital stay. Multivariate analysis showed IPPV hours and steroids had independent effects on VAP. Prolonged IPPV is a critical risk factor for VAP. P aeruginosa is one of the most difficult to treat of those pathogens responsible for VAP. Good management strategies for VAP like adequate infection control practices, early and accurate diagnosis, and more specific antimicrobial use may significantly improve patients outcome. ACKNOWLEDGMENT The authors thank S. Shekhawat for statistical analysis and R. Mathew for secretarial assistance. 1. Kollef MH, Schuster DP: Ventilatior-associated pneumonia: clinical consideratins. AJR Am J Roentgenol 163: , Kollef MH, Skubas MJ, Sundi TM: A randomized clinical trial of continuous aspiration of subglottic secretions in cardiac surgery patients. Chest 116: , American College of Chest Physicians Society of Critical Care Medicine Consensus Conference: Definition for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Crit Care Med 20: , Baxby D, Saene HKF, Stoutenbeck CP, et al: Selective decontamination of the digestive tract: 13 years on, what it is and what it is not. Intensive Care Med 22: , Ziad AM, Cawen C, Gybolahan O, et al: The incidence and risk factors of ventilator associated pneumonia in a Riyadh Hospital. Infection control and hospital epidemiology. 21: , Sim S, Yek A, Yortseven N, et al: Ventilator-associated pneumonia in a cardiothoracic surgery centre, postoperative intensive care unit. J Hosp Infect 47: , 2001 REFERENCES 7. Kolleff MH: Ventilator-associated pneumonia: A multivariate analysis. JAMA 270: , Castelli P, Condemi A, Munari M, et al: Intra-aortic balloon counterpulsation: outcome in cardiac surgical patients. J Cardiothorac Vasc Anesth 15: , Fagon JY, Chastre, Vuagnat A, et al: Nosocomial pneumonia and morbidity among patients in intensive care units. JAMA 275: , Papazian L, Bregeon F, Jhizion X, Chaster J, et al: Effect of ventilator associated pneumonia on mortality and morbidity. Am J Respir Crit Care Med 154:91-97, New HL: Infection problems for the 1990 s. Do we have an answer? Scand J Infect Dis 91:7-13, Emori TG, Gaynes RP: An overview of nosocomial infection including the role of microbiology laboratory. Clin Microbiol Rev 6: , Weinstein RA: Epidemology and control of nosocomial infections in adult intensive care units. Am J Med 91: , 1991

7 28 PAWAR ET AL 14. Daschner FD, Frey P, Wolff G, et al: Nosocomial infections in intensive care wards: A multicentre prospective study. Intensive Care Med 8:5-9, Apostolopoulou E, Sparoo L, Soumilas A, Mantas I: Determinative factor in nosocomial pneumonia in patients with mechanical ventilation in the respiratory care unit. Noseleutike 29: , Verhoef J, Verhage EAE, Visser MR: A decade of experience with selective decontamination of the digestive tract as prophylaxis for infections in patients in the intensive care unit: what have we learned? Clin Infect Dis 17: , Verbisl L: Epidemiology and sensitivity of 8625 ICU haematology/oncology bacterial isolates in Europe. Scand J Infect Dis 91:14-24, Shehabi AA, Zobi J, Zabalawi G: Emergence of multiple drug resistance among salmonella species in Jordon. Eur J Clin Microbiol 14: , Nicoton D, Ulrich C: Process improvement plan for the reduction of nosocomial pneumonia in patients on the ventilator. J Nurs Care Qual 10:18-23, Kollef MH: Ventilation-associated pneumonia: A multivariate analysis. JAMA 270: , Celis R, Torres A, Galell J, et al: Nosocomical pneumonia: A multivariate analysis of risk and prognosis. Chest 93: , Cross AS, Reap B: Role of respiratory assistance devices in endemic nosocomial pneumonia. Am J Med 70: , Kollef MH, Van Horz B, Prentice D, et al: Patient transport from intensive care increases the risk of developing ventilator associated pneumonia. Chest 112: , Aksnes J, Abdelnoor M, Platou ES, et al: Mortality in patients supported by intra-aortic balloon pump in the course of cardiac surgery was related to perioperative myocardial infarction. Eur J Cardiothorac Surg 10: , Bauyer TT, Ferror R, Angrill J, et al: Ventilator-associated pneumonia: Incidence, risk factors and microbiology. Semin Respir Infect 15: , Jansen NJ, Van Oeveren W, Van Vliet M, et al: The role different types of corticosteroids on the inflammatory mediators in cardiopulmonary bypass. Eur J Cardiothorac Surg 5:211, Joseph L: Hospital acquired pneumonia. Risk factors, microbiology and treatment. Chest 119:373S-384S, Celis R, Torres A, Gatell JM, et al: A nosocomical pneumonia in a multivariate analysis of risk and prognosis. Chest 93: , Rello J, Ausina V, Ricart M, et al: Risk factors for infection by pseudomonus aeruginosa in patients with ventilator-associated pneumonia. Intensive Care Med 20: , Daniel T, Blandine M, Claude R, et al: Risks and routes for ventilator associated pneumonia with pseudomonas aeroginosa. Am J Respir Crit Care Med 157: , 1998