Clin Chest Med 26 (2005) 39 – 46
Pathophysiology of Pneumonia Amalia Alco´n, MD, PhDa, Neus Fa`bregas, MD, PhDa, Antoni Torres, MD, PhDb,* a
Surgical Intensive Care Unit, Anesthesiology Department, Hospital Clı´nic, Villarroel 170, Barcelona 08036, Spain Institut Clı´nic de Pneumologı´a I Cirurgı´a Tora´cica, IDIBAPS Universitat de Barcelona, Villarroel 170, Barcelona 08036, Spain
Pneumonia is an infectious process resulting from the invasion and overgrowth of microorganisms in lung parenchyma, breaking down defenses and provoking intra-alveolar exudates. The term community pneumonia refers to when the infection appears in a nonhospitalized population. The term hospitalacquired pneumonia or nosocomial pneumonia is used when there is no evidence that the infection was present or incubating at the time of hospital admission. Nosocomial pneumonia is most frequently found in mechanically ventilated patients; therefore, this term has been replaced by ventilator-associated pneumonia (VAP). With the development of noninvasive ventilation, a new term must be used when pneumonia occurs in nonintubated patients in the intensive care unit (ICU). VAP must be related to ‘‘intubation-associated pneumonia.’’ In the pathogenesis of nosocomial pneumonia, several ways of accessing the lung parenchyma have been described. The development of pneumonia requires that the pathogen reach the alveoli and that host defenses are overwhelmed by microorganism virulence or by the inoculum’s size. Intrusion of bacteria into the lower respiratory tract usually is the result of the aspiration of organisms from the upper
This article was supported by Red Gira 03/063, Red Respira 03/11, Instituto Carlos III and FIS non-responding PI020616. * Corresponding author. Servicio de Pneumologı´a, Hospital Clı´nic, Villarroel 170, Barcelona 08036, Spain. E-mail address: [email protected]
respiratory tract. Underlying disease, loss of mechanical respiratory defenses with the use of sedatives, tracheal intubation, and antibiotic treatment are determinant factors for change in the normal flora of the upper respiratory tract. Nasal, oropharynx, biofilm, and respiratory tract colonization have been related to the risk for pneumonia, especially in ‘‘late-onset’’ pneumonia. Aspiration of normal oropharynx flora in comatose patients and during intubation seems to be the pathogenesis of ‘‘early-onset’’ pneumonia. Less frequently, bacteremia, contaminated aerosols, tracheal aspiration maneuvers, or fibrobronchoscopes can introduce microorganisms directly into lung parenchyma. The relationship of the gastric chamber as the only source of colonization in VAP is more debatable. Postmortem studies show the complexity of histologic findings and suggest that quantitative cultures of lung samples cannot easily discriminate between the presence and absence of histologic pneumonia.
Nasal colonization The upper airway is usually colonized. In the study by Campbell et al  enrolling 776 trauma victims, 18.7% of nasal cultures were positive for Staphylococcus aureus on the day of admission. The extent of methicillin-resistant S aureus (MRSA) carriage within the community is not well known. Scudeller et al  studied 7640 consecutive patients admitted to their Italian hospital; the overall incidence of MRSA nasal carriers was 1.12%.
0272-5231/05/$ – see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ccm.2004.10.013
alco´n et al
Oropharyngeal and gastric colonization
Oropharyngeal and gastric aspiration
Fibronectin, a component of whole saliva, provides binding sites for the adhesion of oral streptococci while inhibiting adhesion of aerobic gram-negative bacilli. The reduction of normal inhibitory flora promotes the colonization of respiratory pathogens. In hospitalized patients, the oropharynx becomes a reservoir of infected secretions. A gastric pH under 4 prevents bacterial growth in the gastric chamber. In hospitalized patients, treatment with antacid drugs or ranitidine is commonly performed to prevent the appearance of stress ulcers. These drugs can increase the pH of gastric juice. De la Torre et al  found that, of 80 patients under mechanical ventilation studied serially during the first 2 weeks of admission, only 10 showed no microorganisms in gastric cultures. These patients had a lower mean gastric pH (3.3) than did patients with gastric colonization (mean gastric pH, 4.6). The type of microorganism that colonizes the stomach is determined by the germs present in the saliva or duodenum. In nonhospitalized patients treated with H2-blockers, gram-positive bacteria dominate the gastric flora. In hospitalized patients, aerobic gram-negative bacilli are predominant , reflecting the presence of these microorganisms in the duodenum or oropharyngeal saliva.
Normal adults frequently aspirate oropharyngeal secretions during sleep, but host defenses prevent lung infections, and the types of microorganisms aspirated are less virulent. The conditions in ill patients are different. First, aerobic gram-negative bacilli colonize the dental plaque and oropharynx. Second, the frequent presence of a nasogastric tube gives rise to lower esophageal sphincter incompetence. Ferrer et al  have demonstrated that small-bore nasogastric tubes in intubated patients do not reduce gastroesophageal reflux or microaspiration. Third, Torres et al  have recognized the importance of supine body position in gastric reflux and tracheal aspiration. Instilling a colloid with technetium via a nasogastric tube and placing patients in a semirecumbent position significantly reduced the radioactivity in tracheal secretions in a comparison with patients in a supine position (Fig. 1). Nevertheless, Girou et al  in a recent randomized study found no significant differences in the daily bacterial count in the oropharynx and trachea between patients placed in a semirecumbent position with continuous subglottic suctioning and patients who were in a supine position. It has not been possible to demonstrate the responsibility of gastric colonization as the initial source of microorganisms for the later development of VAP .
2000 1000 600 500 400
300 200 100 0
Time (min) Fig. 1. Semirecumbent position protects the patient from the aspiration of gastric contents to the lower airways as demonstrated using a radiolabeled technetium marker.
pathophysiology of pneumonia
Lower airway colonization The lower airways can be colonized in patients with chronic obstructive pulmonary disease and in hospitalized patients. Early tracheal colonization (within the first 24 hours of mechanical ventilation) has been described in intubated and mechanically ventilated patients. In this short time of intubation, 80% of the patients were colonized . The composition of the tracheal colonizing flora is worthy of special mention; the pattern of tracheal colonization changes over time among hospitalized and ventilated patients. Healthy patients may be chronically colonized. In the study by Ewig et al , the initial colonization rate at any site (nasal and pharyngeal, tracheobronchial, gastric juice, and protected-specimen brush sample) on ICU admission following brain injury was 83%. Streptococcus pneumoniae, Haemophilus influenzae, and S aureus were the predominant microorganisms in the upper airways. In the study by Sirvent et al  including 100 patients with head injury, 68% of the endotracheal aspirate samples taken within 24 hours of intubation were colonized. S aureus was found in 22% of patients, H influenzae in 20% of patients, S pneumoniae in 6% of patients, and gram-negative bacilli in 20% of patients. These microorganisms are responsible for most instances of early-onset pneumonia, suggesting that the aspiration of oropharynx secretions when the patient becomes unconscious or at intubation can have a role in the development of pneumonia. Pseudomonas spp have increased affinity to ciliated tracheal epithelial cells, and these microorganisms are not usually present in the oropharynx. Pseudomonas spp are probably not present in subglottic secretions. The adherence of Pseudomonas increases to desquamated epithelium, and following influenza virus infection, tracheostomy, or repeated tracheal suctions in intubated patients . In this way, tracheal colonization can be classified as precuff and post-cuff with two different behaviors.
cocci (21%), and yeast (3%). These microorganisms had frequently been isolated from previous sputum cultures. The endotracheal tube has been described as a reservoir for microorganisms, which can adhere to the surface of the foreign body. This biofilm is relatively insensitive to the effects of antibiotics and host defenses, and fragments of endotracheal tube biofilm can be dislodged by suction catheter or by ventilator gas flow (Fig. 2). In 1999, Adair et al  performed a study on 20 patients with VAP and 20 controls. They examined endotracheal tubes for the presence of biofilm after extubation, and the relation with microorganisms that caused VAP. Seventy percent of patients with VAP had identical pathogens isolated from endotracheal biofilm and tracheal secretions (electrophoresis, polymerase chain reaction technique, and susceptibility testing). No pairing of pathogens were observed in controls (P < .005). Feldman et al  described the sequence of endotracheal tube colonization. They studied 10 patients, who on admission showed no evidence of any infection, and cultured the oropharynx, gastric content, the interior of the airway tube (throat swab), and end tracheal secretions twice a day for 5 days. Nine patients became colonized. The oropharynx was the first site (at 36 hours), followed by the stomach (36 – 60 hours) and, thereafter, the lower respiratory tract (60 – 84 hours). Isolation of organisms from the endotracheal tube began at 48 hours but occurred in significant amounts later (60 – 96 hours). No grampositive isolates were found to colonize the endotracheal tube in significant amounts. Nosocomial pneumonia was diagnosed in 3 of the 10 patients. In two cases, Acinetobacter anitratus, the pathogen considered to be responsible for VAP, was first isolated from tracheal aspirates and then from the
Colonization of artificial airways Condensates of ventilator circuits can be a potential source of microorganisms. Craven et al  demonstrated that the inner parts of the ventilator circuit closest to the patient have the highest rates of contamination and the highest bacterial counts. After 24 hours in use, 80% of the ventilator circuits and the condensates were colonized, predominantly by aerobic gram-negative organisms (76%), gram-positive
Fig. 2. Microscopic electronic demonstration of biofilm formation in an endotracheal tube.
alco´n et al
interior of the endotracheal tube (between 60 to 84 hours), and clinical evidence of nosocomial pneumonia later developed (at 96 hours). The time at which colonization changes to infection is multifactorial and is a topic of debate.
Relationship between colonization and infection Performing a multivariate logistic regression analysis, Sirvent et al  found that tracheal colonization by S aureus, H influenzae, or S pneumoniae within 24 hours of intubation in head injury patients was an independent risk factor for early-onset pneumonia (odds ratio, 28.9%; 95% confidence interval, 1.59 – 52.5). Berrouane et al  investigated early-onset pneumonia in a neurosurgical ICU. They studied a cohort of patients over a 13-month period and compared neurotrauma patients with non – neurotrauma patients. A total of 565 adults were included; 57.9% had trauma, and 129 patients sustained 152 episodes of pneumonia. In both groups, the distribution of risk stratified by hospital days was bimodal, being highest during the first 3 days. The risk peaked again at days 5 and 6, and thereafter remained low. Pneumonia occurring during the first 3 days was associated with trauma (P = .036). Head injury may induce immunosuppression, perhaps explaining why neurotrauma patients are at higher risk for early-onset pneumonia. In the Berrouane study , early-onset VAP was caused by S aureus (33%), Haemophilus spp (23%), other gram-positive cocci (22%), and other gramnegative bacilli (19%). After the third day, gramnegative bacilli other than Haemophilus spp accounted for 45.4% of isolates, and the rate of MRSA isolates was 13% before the fourth day and 32% afterward. This change in causative organisms in late-onset VAP was confirmed in the study by Ewig et al . In follow-up cultures of respiratory samples, colonization rates with gram-negative bacilli and Pseudomonas spp increased significantly, with previous short-term antibiotics representing a risk factor.
Nevertheless, it remains unclear whether these are concomitant infections or if one favors the development of the other. The same group  tried to demonstrate a relationship between pneumonia and sinusitis by treating sinusitis in 199 patients and comparing them with a control group (200 patients). VAP was observed in 88 patients, 37 of who were in the study group and 51 in the control group (P = .02). Among the 80 patients with nosocomial sinusitis in the study group, 10 patients of 23 in whom VAP developed had the same organism isolated in the lung and sinus. These findings are of limited value, because the microorganisms that cause sinusitis and VAP are similar (Pseudomonas, S aureus, Streptococcus). The study did not perform chromosomal identification of microorganisms.
Is there a relationship between bacteremia and pneumonia? Bacteremia is not frequently considered a source of microorganisms producing VAP. Blood cultures in
SOURCE OF MICRO-ORGANISMS
Nasal carriers Sinusitis
Health care workers
Is there a relationship between sinusitis and pneumonia? Several researchers have tried to demonstrate a relationship between sinusitis and the development of pneumonia. Using a multivariate analysis, Holzapfel et al  demonstrated that sinusitis increased the risk for nosocomial pneumonia by a factor of 3.8.
Fig. 3. Pathogenesis of VAP. Microorganisms can reach lung parenchyma from different reservoirs.
pathophysiology of pneumonia
patients with VAP are clearly useful if there is a suspicion of another probable infectious condition, but the isolation of a microorganism in blood does not confirm that microorganism as the pathogen causing VAP. Fig. 3 shows the ways in which microorganisms enter the lung, divided into endogenous and external sources.
Histologic characteristics of ventilator-associated pneumonia The histology of VAP has been defined in recent years mainly through immediate postmortem studies. These studies have extensively investigated the lungs of patients mechanically ventilated for several days, and, along with experimental models of pneumonia, have allowed a description of the peculiar histologic and microbiologic characteristics and interactions of human VAP. From this information, important clinical implications have been concluded.
Histologic findings in ventilator-associated pneumonia Histologically, VAP has classically been accepted as the presence of foci of consolidation with intense leukocyte accumulation in bronchioles and adjacent alveoli. This definition is simplistic, because it does not take into account the severity and distribution of lesions. In a postmortem study with bilateral multiple biopsy sampling , the authors described four evolution stages of pneumonia (Fig. 4): An early phase (0 – 2 days of evolution) shows the presence of capillary congestion with an increased number of polymorphonuclear leukocytes. The alveolar spaces usually show a fibrinous exudate. An intermediate phase (3 – 4 days of evolution) is characterized by the presence of fibrin, a few erythrocytes, and several polymorphonuclear leukocytes within the alveoli. An advanced phase (5 – 7 days of evolution) shows polymorphonuclear leukocytes filling up most of the alveoli and macrophages incorporating cellular debris in the cytoplasm. A resolution phase (> 7 days of evolution) occurs when the inflammatory exudate is eliminated owing to phagocytic activity of mononuclear cells.
The authors also have defined two degrees of severity in relation to the lung extension of the lesion: mild and severe. Human and experimental postmortem histologic and microbiologic studies Chastre and colleagues  were the first to develop a postmortem human model based on critically ill patients. In the immediate postmortem period (within 30 minutes), they performed a left thoracotomy under surgical aseptic conditions and obtained six superficial small specimens from the anterior segment of the left lower lobe for culture. In addition, a 1-cm3 specimen was obtained for histologic analysis. They found a good association between histologic and bacteriologic findings (quantitative cultures). Rouby and colleagues  performed a more extensive approach and analyzed histologically two small lung specimens obtained from an area of consolidation of the lower lobe. Another small specimen was cut from the same area and bacteriologically examined. The entire lung was then surgically removed, and a complete lung autopsy was done. They were the first to describe that pneumonia in ventilated patients is a multifocal process disseminated within each pulmonary lobe. These foci of pneumonia were predominantly distributed in lower lobes and in dependent zones of the lung. The histologic lesions of bronchopneumonia were always located within large zones of altered lung parenchyma. Rouby and colleagues demonstrated that a single lung specimen would miss the histologic pneumonia in approximately 30% of cases. The latter finding must be taken into account when interpreting the results of earlier studies, which limited histologic examination to a single sample. Marquette et al  studied the histologic characteristics of entire fixed lungs and confirmed the findings of the previous studies. Pneumonia was found in 50% of the dependent segments and in 37% of the nondependent segments. One of the major characteristics of the lesions was their typically scattered pattern of distribution within normal or damaged lung parenchyma. Only 14 of 83 examined lobes (16.8%) had all of their segments involved by the infectious process. The scattering of the lesions was even more prominent at the segmental level, where the infectious alveolar damage ranged from limited foci of pneumonia to large areas of confluent pneumonia. A distinctive finding in several cases was the absence of pneumonia from the peripheral lung samples, whereas central areas of the same segment displayed typical foci.
alco´n et al
Fig. 4. Histopathology phases of VAP (hematoxylin and eosin, 200). (A) Early phase (0 – 2 days of VAP evolution). Capillary congestion is seen with increased numbers of polymorphonuclear leukocytes. Alveolar spaces show fibrinous exudates. (B) Intermediate phase (3 – 4 days of VAP evolution). Presence of fibrin, erythrocytes, and several polymorphonuclear leukocytes are seen within the alveoli. (C) Advanced phase (5 – 7 days of VAP evolution). Polymorphonuclear leukocytes fill up most of the alveoli, and macrophages incorporate cellular debris in the cytoplasm. (D) Resolution phase (> 7 days). Inflammatory exudates are eliminated owing to phagocytic activity of mononuclear cells.
Quantitative cultures of lung samples cannot easily discriminate the presence or absence of histologic pneumonia. In the authors’ previously mentioned postmortem study , patients were included with and without antibiotic treatment (at least 48 hours free of antibiotic treatment). Several specimens from each lobe of the two lungs were aseptically obtained (average of 16 samples per patient). When the evolution classification of VAP was applied, the disseminated multifocal heterogeneous pattern of VAP was confirmed to involve predominantly the lower lobes. Interestingly, all of the phases in this classification coexisted in the same patient and in the same lung, exhibiting a pleomorphic histologic pattern. Overall, the intermediate phase and advanced phase were the most common stages of pneumonia observed. Similar to other studies, nonspecific alveolar damage and bronchiolitis also were frequent findings. Papazian and coworkers  analyzed one entire lung of 38 patients. They found no sign of bronchopneumonia in 20 cases. Conversely, in the remaining 18 cases, they confirmed bronchopneumonia histologically. There was no relationship between the
results of the various cultures and the pathology results. Bronchiolitis was noted in eight patients, five of who had concomitant histologic signs of pneumonia. The remaining three patients had negative lung cultures. Additional histologic findings were fibrosis in nine cases and diffuse alveolar damage in seven cases. These investigators examined the significance of the isolation of Candida spp in their samples. Despite frequent lung colonization by Candida, only two patients exhibited histologic signs of Candida pneumonia. Lung tissue cultures were positive for Candida albicans in these two patients. This finding agrees with the results of the study by El-Ebiary et al  who found that the incidence of Candida pneumonia was 8%. Nevertheless, the incidence of Candida isolation from pulmonary biopsies in critically ill, mechanically ventilated, nonneutropenic patients who die is high (40%), indicating that Candida is a frequent colonizing agent in critically ill patients. To avoid confounding factors such as antibiotic presence or lung injury described in other studies, Marquette et al  induced pneumonia in pigs free from antibiotics and previous concomitant lung
pathophysiology of pneumonia
disease secondary to tracheobronchial stenosis. They found that the histologic lesions of pneumonia, as well as the lung bacterial burden, were unequally distributed within the lungs and even within the lung segments. Moreover, specimens showing histologic evidence of pneumonia had significantly higher bacterial burden than specimens with bronchial infections and specimens with neither bronchial nor lung infection. Nevertheless, the authors could not define a clear threshold for quantitative cultures to discriminate the presence or absence of pneumonia. This study provides experimental insights into the relationship between microbiologic and histologic features in bacterial pneumonia and confirms previous findings in humans. In humans, there is compelling evidence to suggest that quantitative biopsy cultures cannot reliably discriminate between patients with and without evidence of histologic pneumonia. The use of a specific threshold to define the presence of pneumonia does not take into account the fact that lung infection occurs along a bacteriologic continuum. When pneumonia begins, or if infectious bronchiolitis is present, the diagnostic threshold may not be met. The same scenario occurs when prior antibiotics have been given. Another explanation for low qualitative lung cultures in the presence of histologic pneumonia is the normal functioning of antibacterial lung defenses, which clear lung bacterial burden. The specificity of lung cultures is low, with a high rate of false-positive cultures. In postmortem studies, falsepositive lung cultures (without pneumonia) may result from bacterial colonization and bronchiolitis. It has been suggested that in critically ill patients who are close to death, the lung may have massive bacterial colonization, which could explain the frequent presence of bacteria in distal airways without histologic pneumonia. From all of these findings, it is clear that, owing to the poor relationship between quantitative lung cultures and histologic examination, quantitative lung biopsy cultures alone cannot be used to validate in vivo diagnostic techniques to investigate microbiologically VAP. Clinical implications of histologic and microbiologic findings in postmortem studies of ventilator-associated pneumonia The histologic findings of human postmortem studies have the following clinical implications: The initial phases of VAP that probably need to be treated with antibiotics cannot be detected at the time of portable chest radiography.
Because VAP is a multifocal process, techniques that explore broad lung regions, such as bronchoalveolar lavage, are clearly preferred to those that explore only a segment (protectedbrush specimen). Blind diagnostic methods that can sample lung dependent zones are probably as accurate as visually guided methods. The microbiologic findings have the following clinical implications: As mentioned previously, quantitative postmortem lung cultures cannot be used as a gold standard to validate microbiologic diagnostic techniques. When interpreting quantitative bacteriology at the bedside, the clinician should weight a number of factors that can modify bronchial and alveolar bacterial burden: the presumed stage of bronchopneumonia, the administration of antibiotics, the technique of distal sampling, natural host bacterial defenses, the duration of mechanical ventilation, and the presence of acute lung injury. The microbiologic complexity of VAP does not support the concept of a standard threshold for the diagnosis of pneumonia. Treatment algorithms based on definite thresholds of quantitative cultures may lead to undertreating patients. Because early and adequate initial antibiotic treatment is one of the major factors related to the prognosis of VAP, the strict execution of treatment based on quantitative thresholds without clinical judgment may be hazardous to patients.
Summary Healthy patients may be chronically colonized. More than 50% of patients who are admitted to ICUs have already been colonized at the time of admission with the microorganisms responsible for subsequent infections. The development of pneumonia requires that the pathogen reach the alveoli and that host defenses are overwhelmed by microorganism virulence or by the inoculum size. Endogenous sources of microorganisms are nasal carriers, sinusitis, oropharynx, gastric, or tracheal colonization, and hematogenous spread. Other external sources of contamination, such as ICU workers, aerosols, or fibrobronchoscopy, must be considered as accidental.
alco´n et al
Histologic findings in postmortem studies show the complexity and the heterogeneity of the distribution of VAP. Quantitative biopsy cultures cannot reliably discriminate between patients with and without evidence of histologic pneumonia.
References  Campbell W, Hendrix E, Schwalbe R, et al. Headinjured patients who are nasal carriers of Staphylococcus aureus are at risk for Staphylococcus aureus pneumonia. Crit Care Med 1999;27:798 – 801.  Scudeller L, Leoncini O, Boni S, et al. MRSA carriage: the relationship between community and healthcare setting A study in an Italian hospital. J Hosp Infect 2000;46:222 – 9.  De la Torre FJ, Pont T, Ferrer A, et al. Pattern of tracheal colonization during mechanical ventilation. Am J Respir Crit Care Med 1995;152:1028 – 33.  Driks MR, Craven DE, Celli BR, et al. Nosocomial pneumonia in intubated patients given sucralfate as compared with antacids or histamine type 2 blockers: the role of gastric colonization. N Engl J Med 1987; 317:1376 – 82.  Ferrer M, Bauer TT, Torres A, et al. Effect of nasogastric tube size on gastroesophageal reflux and microaspiration in intubated patients. Ann Intern Med 1999;130:991 – 4.  Torres A, Serra-Batlles J, Ros E, et al. Pulmonary aspiration of gastric contents in patients receiving mechanical ventilation: the effect of body position. Ann Intern Med 1992;116:540 – 3.  Girou E, Buu-Hoy A, Stephan F, et al. Airway colonization in long-term mechanically ventilated patients. Intensive Care Med 2004;30:225 – 33.  Carden˜osa JA, Sole-Violan J, Bordes A, et al. Role of different routes of tracheal colonization in the development of pneumonia receiving mechanical ventilation. Chest 1999;116:462 – 70.  Ewig S, Torres A, El-Ebiary M, et al. Bacterial colonization patterns in mechanically ventilated patients with traumatic and medical head injury: incidence, risk factors, and association with ventilator-associated pneumonia. Am J Respir Crit Care Med 1999;159:188 – 98.  Sirvent JM, Torres A, Vidaur L, et al. Tracheal colonisation within 24 h of intubation in patients with head trauma: risk factors for developing early-onset ventilator-associated pneumonia. Intensive Care Med 2000;26:1369 – 72.  Estes RJ, Meduri GU. The pathogenesis of ventilatorassociated pneumonia. I. Mechanisms of bacterial transcolonization and airway inoculation. Intensive Care Med 1995;21:365 – 83.
 Craven DE, Goularte TA, Make BJ. Contaminated condensate in mechanical ventilator circuits: a risk factor for nosocomial pneumonia? Am Rev Respir Dis 1984;129:625 – 8.  Adair CG, Gorman SP, Feron BM, et al. Implications of endotracheal tube biofilm for ventilator- associated pneumonia. Intensive Care Med 1999;25:1072 – 6.  Feldman C, Kassel M, Cantrell J, et al. The presence and sequence of endotracheal tube colonization in patients undergoing mechanical ventilation. Eur Respir J 1999;13:546 – 51.  Berrouane Y, Daudenthun I, Riegel B, et al. Early onset pneumonia in neurosurgical intensive care unit patients. J Hosp Infect 1998;40:275 – 80.  Holzapfel L, Chevret S, Madinier G, et al. Influence of long-term oro- or nasotracheal intubation on nosocomial maxillary sinusitis and pneumonia: results of a prospective, randomised, clinical trial. Crit Care Med 1993;21:1132 – 8.  Holzapfel L, Chastang C, Demingeon G, et al. A randomised study assessing the systemic search for maxillary sinusitis in nasotracheally mechanically ventilated patients: influence of nosocomial maxillary sinusitis on the occurrence of ventilator-associated pneumonia. Am J Respir Crit Care Med 1999;159: 695 – 701.  Fa´bregas N, Torres A, El-Ebiary M, et al. Histopathologic and microbiologic aspects of ventilator-associated pneumonia. Anesthesiology 1996;84:757 – 9.  Chastre J, Viau F, Brun P, et al. Prospective evaluation of the protected catheter brush for the diagnosis of pulmonary infections in ventilated patients. Am Rev Respir Dis 1984;130:924 – 39.  Rouby JJ, Rossignon MD, Nicolas MH, et al. A prospective study of protected bronchoalveolar lavage in the diagnosis of nosocomial pneumonia. Anesthesiology 1989;71:679 – 85.  Marquette CH, Copin MC, Wallet F, et al. Diagnostic tests for pneumonia in ventilated patients: prospective evaluation of diagnostic accuracy using histology as a diagnostic gold standard. Am J Respir Crit Care Med 1995;151:1878 – 88.  Papazian L, Thomas P, Garbe L, et al. Bronchoscopic or blind sampling techniques for the diagnosis of ventilator associated pneumonia. Am J Respir Crit Care Med 1995;152:1982 – 91.  El-Ebiary M, Torres A, Fa`bregas N, et al. Significance of the isolation of Candida species from respiratory samples in critically ill, non-neutropenic patients: an immediate post-mortem histologic study. Am J Crit Care Med 1997;156:583 – 90.  Marquette CH, Wallet F, Copin MC, et al. Relationship between microbiologic and histologic features in bacterial pneumonia. Am J Respir Crit Care Med 1996;154:1784 – 7.