SUMMARY
SUMMARY While critically ill patients experience a life-threatening illness, they commonly contract ventilator-associated pneumonia. This nosocomial infection increases morbidity and likely mortality as well as the cost of health care. This article reviews the literature with regard to diagnosis, treatment, and prevention. It provides conclusions that can be implemented in practice as well as an algorithm for the bedside clinician and also focuses on the controversies with regard to diagnostic tools and approaches, treatment plans, and prevention strategies.
INTRODUCTION
Patients in the intensive care unit (ICU) are at risk for dying not only from their critical illness but also from secondary processes such as nosocomial infection. Pneumonia is the second most common nosocomial infection in critically ill patients, affecting 27% of all critically ill patients (170). Eighty-six percent of nosocomial pneumonias are associated with mechanical ventilation and are termed ventilator-associated pneumonia (VAP). Between 250,000 and 300,000 cases per year occur in the United States alone, which is an incidence rate of 5 to 10 cases per 1,000 hospital admissions (134, 170). The mortality attributable to VAP has been reported to range between 0 and 50% (10, 41, 43, 96, 161). Studies have provided different results when determining attributable mortality, in part because of very different populations (less-acute trauma patients, acute respiratory distress syndrome [ARDS] patients, and medical and surgical ICU patients) and in part as a result of variances in appropriate empirical medical therapy during the initial 2 days. Furthermore, the organisms recovered have an impact on outcome, with higher mortality rates seen in VAP caused by Pseudomonas aeruginosa, Acinetobacter spp., and Stenotrophomonas maltophilia (109). Beyond mortality, the economics of VAP include increased ICU lengths of stays (LOS) (from 4 to 13 days), and incremental costs associated with VAP have been estimated at between $5,000 and $20,000 per diagnosis (20, 206, 211).
Ventilator-associated pneumonia is defined as pneumonia occurring more than 48 h after patients have been intubated and received mechanical ventilation. Diagnosing VAP requires a high clinical suspicion combined with bedside examination, radiographic examination, and microbiologic analysis of respiratory secretions. Aggressive surveillance is vital in understanding local factors leading to VAP and the microbiologic milieu of a given unit. Judicious antibiotic usage is essential, as resistant organisms continue to plague intensive care units and critically ill patients. Simple nursing and respiratory therapy interventions for prevention should be adopted. Over the past several decades our understanding of VAP has grown significantly with regard to pathogenesis, risk factors, diagnostic testing, therapies, and prevention by modifying risk factors. This paper is designed for the practicing clinician in addressing diagnosis, treatment, and prevention of VAP.
DIAGNOSIS
Clinical DiagnosisVentilator-associated pneumonia is usually suspected when the individual develops a new or progressive infiltrate on chest radiograph, leukocytosis, and purulent tracheobronchial secretions. Unfortunately, and unlike for community-acquired pneumonia, accepted clinical criteria for pneumonia are of limited diagnostic value in definitively establishing the presence of VAP. In a postmortem study by Fabregas et al., when findings on histologic analysis and cultures of lung samples obtained immediately after death were used as references, a new and persistent (>48-h) infiltrate on chest radiograph plus two or more of the three criteria (i) fever of >38.3°C, (ii) leukocytosis of >12 × 109/ml, and/or (iii) purulent tracheobronchial secretions had a sensitivity of 69% and a specificity of 75% for establishing the diagnosis of VAP (60). When all three clinical variables were required for the diagnosis, the sensitivity declined further (23%); the use of a single variable resulted in a decrease in specificity (33%). The poor accuracy of clinical criteria for diagnosing VAP should not be surprising considering that purulent tracheobronchial secretions are invariably present in patients receiving prolonged mechanical ventilation and are seldom caused by pneumonia. In addition, the systemic signs of pneumonia such as fever, tachycardia, and leukocytosis are nonspecific; they can be caused by any state that releases the cytokines interleukin-1, interleukin-6, tumor necrosis factor alpha, and gamma interferon (33, 34, 63, 135). Examples of such conditions include trauma, surgery, the fibroproliferative phase of ARDS, deep vein thrombosis, pulmonary embolism, and pulmonary infarction. Reasonable clinical criteria for the suspicion of VAP include a new and persistent (>48-h) or progressive radiographic infiltrate plus two of the following: temperature of >38°C or <36°C, blood leukocyte count of >10,000 cells/ml or <5,000 cells/ml, purulent tracheal secretions, and gas exchange degradation (5, 103).
The sensitivity of the clinical criteria for VAP outlined above is even lower in patients with ARDS, where it may be difficult to detect new radiographic infiltrates. In the setting of ARDS, Bell et al. reported a false-negative rate of 46% for the clinical diagnosis of VAP (11). Consequently, suspicion for VAP in the setting of ARDS should be high. The presence of even one of the clinical criteria for VAP, unexplained hemodynamic instability, or an unexplained deterioration in arterial blood gases should prompt consideration of further diagnostic testing (129).
When purulent sputum, a positive sputum culture, fever, and leukocytosis are present without a new lung infiltrate, the diagnosis of nosocomial tracheobronchitis should be entertained. In mechanically ventilated patients, nosocomial tracheobronchitis has been associated with a longer ICU stay and time on the ventilator, without increased mortality (158). In one randomized trial of intubated patients with community-acquired tracheobronchitis, antibiotic therapy resulted in a decreased incidence of subsequent pneumonia and mortality (156). However, prospective, randomized, controlled trials are required before antibiotic therapy can be recommended for the routine treatment of nosocomial tracheobronchitis. Furthermore, differentiation of tracheobronchitis from pneumonia is dependent upon the radiograph, which in the ICU is portable and often of poor quality. Hence, the clinician should utilize a clinical pulmonary infection score (CPIS) (see below) to direct therapy.
Radiologic DiagnosisWhile the portable chest radiograph still remains a mandatory component in the diagnosis of ventilated patients with suspected pneumonia, as with clinical criteria for diagnosing VAP, it too has problems with both sensitivity and specificity. Poor-quality films further compromise the accuracy of chest X rays. Although a normal chest radiograph makes VAP unlikely, in one study of surgical patients, 26% of opacities were detected by computed tomography (CT) scan but not by portable chest X ray (25). In addition, asymmetric pulmonary infiltrates consistent with VAP can be caused by numerous noninfectious disorders, including atelectasis, chemical pneumonitis, asymmetric cardiac pulmonary edema, pulmonary embolism, cryptogenic organizing pneumonia, pulmonary contusion, pulmonary hemorrhage, drug reaction, and asymmetric ARDS. The overall radiographic specificity of a pulmonary opacity consistent with pneumonia is only 27% to 35% (116, 216).
Nonetheless, because of their high specificity, certain chest radiograph findings can be useful in establishing the diagnosis of pneumonia when present. Based on several studies, including an autopsy study by Wunderink et al., these useful findings include rapid cavitation of the pulmonary infiltrate, especially if progressive; an air space process abutting a fissure (specificity, 96%); and an air bronchogram, especially if single (specificity, 96%). Unfortunately, such radiographic abnormalities are uncommon (216).
Blood and pleural fluid cultures.Although VAP spreads to the blood or pleural space in <10% of cases, if an organism known to cause pneumonia is cultured in the setting of clinically suspected pneumonia, treatment is warranted. Consequently, most experts recommend that two sets of blood cultures and a thoracentesis for nonloculated pleural effusions of ≥10 mm in diameter on a lateral decubitus chest radiograph should be part of the evaluation of suspected VAP (30). If the effusion is loculated, ultrasound guidance may be required. However, it is important to keep in mind not only that the sensitivity of blood cultures for the diagnosis of VAP is less than 25% but also that when positive, the organisms may originate from an extrapulmonary site of infection in as many as 64% of cases and even when VAP is present (23, 124).
Nonquantitative or semiquantitative airway sampling.Gram staining and nonquantitative and semiquantitative cultures of tracheal secretions have the advantages of reproducibility and of requiring little technical expertise and no specialized equipment or technique. However, these studies add little to the sensitivity and specificity of the clinical diagnosis of VAP, as the upper respiratory tract is rapidly, within hours of intubation, colonized by potential pulmonary pathogens, even when pneumonia is not present (57, 91). Thus, if an organism is cultured or noted on Gram stain, one does not know if it is the cause of the pneumonia or simply colonization. In a study of 48 patients with respiratory failure, concordance between tracheal nonquantitative cultures and cultures of lung tissue from open lung biopsy was only 40% (82). In that study, of those patients with pneumonia on lung histology, endotracheal aspirate (ETA) had a sensitivity of 82% but a specificity of only 27%. In addition, routine surveillance cultures of ETAs to anticipate the etiology of a subsequent pneumonia can be misleading in a significant percentage of patients, though recent data indicate that quantitative ETAs may be helpful (see below) (78, 146).
Only 15% of ETAs are adequate specimens when strict definitional criteria (organisms on Gram staining and fewer than 10 squamous epithelial cells per low-power field [magnification, ×100]) are followed (153). Furthermore, the number of polymorphonuclear leukocytes is not predictive of an interpretable specimen in patients with VAP (153). Nonquantitative and semiquantitative cultures of ETAs for the diagnosis of VAP are most useful if the specimen is adequate and antimicrobial therapy has not been added or changed in the prior 72 h. The negative predictive value of these cultures in this setting is high (94%) (15). Alternative causes for the patient's presentation, including nonpulmonary sites of infection, should be investigated. In addition, the absence of growth of multidrug-resistant organisms in this circumstance provides strong evidence that these bacteria are not causative. Antibiotics should be adjusted accordingly. Overall, the presence of prior antibiotics results in a false-negative rate of 10 to 40% (200).
Because of the poor specificity of the clinical diagnosis of VAP and of qualitative evaluation of ETAs, Pugin et al. developed a composite clinical score, called the clinical pulmonary infection score (CPIS), based on six variables: temperature, blood leukocyte count, volume and purulence of tracheal secretions, oxygenation, pulmonary radiography, and semiquantitative culture of tracheal aspirate. The score varied from 0 to 12. A CPIS of >6 had a sensitivity of 93% and a specificity of 100% (164). However, there were several limitations of that investigation, including that only 28 patients with 40 episodes of pneumonia were studied and that the diagnosis was based upon a “bacterial index” that has not been a well-accepted reference test for pulmonary infection. The “bacterial index” is the sum of the logarithm of the number of bacteria cultured per milliliter of bronchoalveolar lavage (BAL) fluid. Two subsequent studies evaluated the accuracy of the CPIS by using both histology and lung tissue cultures as the reference tests (60, 162). In these investigations, the sensitivity was 72% to 77%, and the specificity was 42% to 85%. However, the study with the greater diagnostic accuracy used the tracheal aspirate culture recorded within the 48 to 72 h preceding the investigation, information that may not be available routinely. Moreover, with a sensitivity of 72 to 77%, a CPIS of ≤6 is still insufficient to withhold antibiotic therapy safely in patients with suspected VAP.
Using quantitative culture of BAL fluid as the diagnostic criteria for VAP and a CPIS of >5 as the diagnostic cutoff, the sensitivity and specificity of the CPIS were 83% and 17%, respectively. The addition of Gram staining via blind or bronchoscopically directed BAL or PTC (see below) did improve the overall sensitivity and specificity of the CPIS (65). However, the false-negative rate was still 16 to 25% (65).
Due to poor specificity and poor positive predictive value, reliance on clinical parameters, chest X-ray findings, and non- and semiquantitative sputum analysis will result in overdiagnosis and therefore overtreatment of VAP. Such an approach will result in excess antibiotic use with its attendant cost, potential toxicity, and selection of drug-resistant organisms. A recent decision analysis suggested that more deaths occurred if patients were treated with antibiotics on the basis of only clinical suspicion of VAP than if antibiotics were withheld (190).
Overreliance on the clinical diagnosis of VAP may also result in undertreating alternative infectious and noninfectious causes of fever and pulmonary infiltrates in mechanically ventilated patients. Meduri et al. prospectively studied 50 patients with clinically suspected VAP (139). Twenty-two patients had ARDS. Based on quantitative cultures of bronchoscopic protected specimen brush (PSB) and bronchoalveolar lavage, pneumonia was diagnosed in 42%. Of the infectious causes of fever and pulmonary infiltrates on chest radiography, 84% were pneumonia, sinusitis, urinary tract infection, or catheter-related infection. Less frequent infectious causes included intra-abdominal abscess, peritonitis, acalculous cholecystitis, Clostridium difficile colitis, empyema, wound infection, primary bacteremia, and candidemia. Twenty-four percent of fevers were secondary to noninfectious causes, including deep venous thrombosis, pulmonary embolism, pancreatitis, chemical aspiration, fibroproliferative stage of ARDS, and drugs. Fifty-six percent of the chest X-ray abnormalities were due to noninfectious causes. Concomitant infections were found in 62% of cases, with 60% of these being caused by a different pathogen. On average, there were 1.7 causes of fever per patient.
Quantitative cultures of airway specimens.To potentially improve the specificity of the diagnosis of VAP and the consequent unnecessary antibiotic use and its associated problems, numerous studies have investigated the role of quantitative cultures of respiratory secretions. These have included nonbronchoscopic methods such as quantitative cultures of ETAs (QEAs) and sampling of secretions from distal airways “blindly” via an endobronchial catheter. Blind bronchial sampling (BBS), PSB, protected telescoping catheter (PTC), BAL, and protected BAL (mini-BAL) samples can be obtained via the latter method. Bronchoscopic sampling methods permit quantitative cultures of PSB, PTC, and protected and nonprotected BAL specimens.
The PSB and PTC are double-sheathed catheters with a biodegradable plug occluding the distal end of the inner catheter to minimize bacterial contamination. The PSB and PTC procedures involve placing the tip of the bronchoscope or “blindly placed” catheter next to an involved bronchial segmental orifice. With bronchoscopy, direct visualization is possible. With a “blind” procedure, the catheter is advanced until resistance is met and then retracted a few centimeters. The inner catheter is then advanced 2 or 3 cm beyond the outer catheter, ejecting the plug. With PSB, a brush is further advanced and rotated several times; with PTC, a 10-ml syringe is used to perform three brief aspirations of secretions. BAL involves the infusion and aspiration of sterile saline through a flexible fiber-optic bronchoscope or “blindly placed” catheter wedged into a bronchial segmental orifice. Protected BAL involves a specialized balloon-tipped catheter with a distal ejectable plug. When performing a BAL to diagnose VAP, instillation of at least 140 ml of saline is required to maximize diagnostic yield (70, 138, 145).
If a bronchoscopically directed quantitative culture is chosen, the patient should receive adequate sedation, with consideration of a short-acting paralytic agent to prevent coughing during the procedure. The endotracheal tube must be ≥1.5 mm larger than the external diameter of the flexible bronchoscope. The patient should receive a fraction of inspired oxygen (FiO2) of 100%, and positive-end expiratory pressure should be reduced as much as tolerated. To maximize ventilation and minimize air trapping, the peak inspiratory flow should be decreased to ≤60 liters/min, the respiratory rate set between 10 and 20 breaths/min, and the peak inspiratory pressure alarm increased. The patient should be carefully monitored throughout the procedure, with particular attention to exhaled tidal volume, peak inspiratory pressure, oxygen saturation, the electrocardiogram, and vital signs. Secondary hypotension should be anticipated, and appropriate intravenous fluids and vasopressors should be available for immediate administration (70).
The sampling area should be chosen based on the location of the infiltrate on chest X ray or CT scan. This typically corresponds to the bronchial segment with purulent secretions and/or where endobronchial abnormalities are maximal, which can be clues in the setting of diffuse pulmonary infiltrates or minimal changes in a previously abnormal chest X ray (137). When in doubt, sample the posterior right lower lobe, since autopsy studies have indicated that VAP frequently involves this area (61, 92, 130, 173). Multiple specimens are no more accurate than single specimens (136).
As with nonquantitative and semiquantitative cultures, only adequate specimens should be processed. The presence of more than 1% epithelial cells or 10 epithelial cells per low-power field (magnification, ×100) in bronchoscopic or “blind” BAL, PSB, PTC, or bronchial sampling suggests heavy oropharyngeal colonization. Returns of <10% of the instilled BAL fluid are probably not representative of the lower respiratory tract (70). Since interpretation of such specimens is unreliable, they should not be cultured. For QEAs, the same criteria mentioned above for nonquantitative and semiquantitative cultures of an ETA should be utilized.
For each of the quantitative culturing methods, threshold values have been derived and are expressed in CFU per milliliter. If the number of CFU/ml is equal to or exceeds the threshold values for the particular technique, a diagnosis of pneumonia is made. Threshold values often employed for diagnosing pneumonia by quantitative cultures are ≥105 to 106, ≥104, and ≥103 CFU/ml for QEA, bronchoscopic BAL, and PSB, respectively, with ≥105 CFU/ml being the most widely accepted value for QEA (12, 30). For “blind” distal sampling, the thresholds are ≥103 CFU/ml for PSB and mini-BAL and ≥104 CFU/ml for cultures obtained with BBS and unprotected BAL (Table 1) (30).
Diagnostic accuracy of quantitative culturing techniques
These “cutoff” values for diagnosing VAP are based in part on the findings of quantitative cultures obtained from infected lung tissue and the volume and dilution of the respiratory secretions retrieved by the technique. For instance, BAL collects approximately 1 ml of secretions in 10 to 100 ml of effluent. This corresponds to a dilution factor of 1/10 to 1/100. Several investigators have confirmed that with pneumonia, pathogens are present in lower respiratory tract inflammatory secretions at concentrations of at least 105 to 106 CFU/ml; contaminants are generally present at less than 104 CFU/ml. Consequently, for BAL, the threshold value of 104 CFU/ml corresponds to 105 to 106 CFU/ml in the pneumonia (30).
Numerous factors can influence the results of quantitative cultures, including the timing of the pneumonia, the skill and experience of the operator, the adequacy of the specimen, technical aspects such as appropriate processing and delays in transport to the laboratory, special populations such as those with chronic obstructive pulmonary disease (who may have relatively high bacterial counts without pneumonia), and prior or concurrent antibiotic therapy (149).
Because of these potential limitations, it is important to bear in mind that a quantitative culture that exceeds a threshold value is not diagnostic of VAP by itself (9). False-positive quantitative cultures could be secondary to bronchiolitis, colonization, or oropharyngeal contamination (9). Likewise, a result below these threshold values does not rule out the presence of pneumonia, particularly in the setting of prior antibiotic therapy (see below). While higher bacterial counts correlate with a higher likelihood of VAP, lower counts are associated with a lower probability. Consequently, rather than interpreting a quantitative culture as either “positive” or “negative,” it is clinically more useful to utilize the exact number of CFU/ml (see below) (9).
(i) Diagnostic accuracy.Before discussing and comparing the diagnostic accuracies of the different quantitative culturing methods and their advantages and disadvantages, it is important to understand that there exists a significant amount of controversy in the medical literature regarding the use of quantitative cultures in general as well as the use of bronchoscopic (invasive) versus nonbronchoscopic (noninvasive) methods. At least part of the reason for this controversy is that the studies evaluating the accuracies of ETA and of bronchoscopic and nonbronchoscopic PSB, PTC, and BAL to diagnose VAP have shown a significant degree of variability in sensitivity, specificity, and positive and negative predictive values for each of the techniques. This variability has resulted from the use of different “gold standards” for the diagnosis of VAP, the use of different cutoff thresholds for quantitative cultures, differences in equipment and protocols, and differences between the populations studied, in particular, the use of antibiotics. Even the most accepted “gold standard,” histopathologic examination and culture of lung tissue obtained by biopsy or at autopsy, has inherent problems. Among other things, patients included in autopsy studies may not be representative of most patients with VAP. Moreover, the recognition of histologic pneumonia varies among pathologists. In a study by Corley et al., the prevalence of pneumonia in postmortem open lung biopsies determined by each of four pathologists varied from 18% to 38% (40). Nonetheless, histopathology and lung tissue culture remain our best “gold standard” for the diagnosis of VAP.
Numerous studies have demonstrated that prior and concurrent antibiotic therapy decrease the accuracy, sensitivity, and negative predictive value of Gram staining, including the percentage of cells containing intracellular organisms (ICOs), as well as quantitative, semiquantitative, and nonquantitative cultures (184). In a study of 76 patients with VAP by Montravers et al., PSB quantitative cultures obtained after the administration of effective antibiotic therapy showed complete eradication of the causative organisms after only 3 days of treatment in 67% of patients (150). Even 24 h of administration of an antibiotic can affect culture results (189). This effect of prior antibiotics on the false-negative rate of microbiologic studies is of great concern, particularly since VAP is a potentially lethal disorder. However, if antibiotics have not been changed in the last 72 h, the diagnostic yield of any culture technique is unaffected (189, 197).
(ii) Bronchoscopic protected specimen brush and BAL.Chastre and Fagon pooled the results of 18 studies that evaluated the bronchoscopically directed PSB technique for diagnosing VAP (30). A total of 795 critically ill patients were included in the analysis. The overall diagnostic accuracy of this method was very good, with a sensitivity of 89% (95% confidence interval [CI], 87 to 93%) and a specificity of 94% (95% CI, 92 to 97%). However, studies investigating the reproducibility and variability of bronchoscopic PSB have raised concerns about this technique. Timsit et al. and Marquette et al. repeated PSB sampling in the same lung subsegment and noted that results of quantitative cultures were on each side of the 103-CFU/ml threshold in 16.7% to 13.6% of cases, respectively, with 59 to 67% of samples having CFU/ml counts varying by more than 10-fold (131, 196). These investigators concluded that, as with all quantitative culturing techniques, borderline PSB quantitative culture results should be interpreted with caution. In such a circumstance, one should consider repeating the test if suspicion of VAP persists and antibiotics have not yet been started.
Torres and El-Ebiary reviewed 23 studies that evaluated the accuracy of bronchoscopic BAL in diagnosing VAP. A total of 957 patients were included in the analysis. Sensitivity ranged from 42 to 93%, with a mean ± standard deviation of 73% ± 18%. The specificity ranged from 45 to 100%, with a mean specificity ± standard deviation of 82% ± 19% (199). In 12 studies, the detection of ICOs in 2 to 5% of recovered cells had a sensitivity of 69% ± 20% and a specificity of 75% ± 28% for diagnosing VAP (199). Reproducibility of BAL is excellent when the culture is sterile. However, for positive cultures, the quantitative repeatability was only 53% in one study (73).
Chastre et al. compared PSB and BAL to the “gold standard” histopathologic findings and quantitative tissue culture results from the same areas of lungs of patients in the terminal phase of their illness. Patients were included in the study only if they never had pneumonia or had acquired it during the terminal phase of their illness. Antibiotics had not been changed or added in the 3 days prior to the sampling. In this investigation, PSB had a sensitivity of 82%, a specificity of 77%, a positive predictive value of 74%, and a negative predictive value of 85%; BAL had a sensitivity of 91%, a specificity of 78%, a positive predictive value of 83%, and a negative predictive value of 87%; and the presence of ≥5% ICOs had a sensitivity, specificity, positive predictive value, and negative predictive value of 91%, 89%, 91%, and 89%, respectively (31).
(iii) Quantitative endotracheal aspirate.Using a threshold value of ≥106 CFU/ml, the sensitivity and specificity of QEA have varied widely from study to study. Sensitivity ranged from 38% to 82% with a mean of 76% ± 9%; specificity ranged from 72% to 85% with a mean of 75% ± 28%. When the diagnosis of VAP was based on postmortem lung examination, the sensitivity/specificity for 105-CFU/ml and 106-CFU/ml thresholds were 63%/75% and 55%/85%, respectively (130). Patients on antibiotics were included in that study, which may have decreased the sensitivity of the procedure.
Jourdain et al. found the QEA to have a sensitivity and a specificity of as high as 68% and 84%, respectively, and a false-negative rate of as high as 32% compared to bronchoscopic quantitative PSB and BAL (95, 133). The diagnostic thresholds for QEA, PSB, and BAL in that study were ≥106 CFU/ml, ≥103 CFU/ml and ≥5% ICOs, respectively. In addition, only 40% of the organisms isolated from QEAs were concomitantly isolated from PSB specimens. Strengths of this study included a well-defined “gold standard,” which is as close as one can get to the “true gold standard” of histopathology and lung tissue culture, and the absence of an addition to or change in antibiotics in the 3 days prior to the appearance of the new pulmonary infiltrate.
(iv) Blind BBS, PSB, and BAL.Campbell reviewed 15 studies evaluating the accuracy of blinded sampling methods (26). A total of 654 episodes of pneumonia were included in the analysis. Sensitivities for BBS, mini-BAL, and PSB were 74 to 97%, 63 to 100%, and 58 to 86%, respectively. Specificities ranged from 74 to 100% for BBS, from 66 to 96% for mini-BAL, and from 71 to 100% for PSB. Marik and Brown compared blind PSB to PSB performed by bronchoscopy. In that study both diagnostic techniques were performed in the absence of antibiotic therapy and blind PSB preceded bronchoscopy, to minimize contamination of the lower respiratory tract. In addition, the study used a well-defined and reasonable “gold standard” for the diagnosis of VAP, though not histopathologic and lung tissue culture. In that investigation, blind PSB had a sensitivity of 86%, a specificity of 85%, a positive predictive value of 80%, and a negative predictive value of 90% (128, 133).
(v) Comparisons among the different quantitative culturing techniques: bronchoscopic versus nonbronchoscopic techniques.Inherent advantages of nonbronchoscopic techniques include less invasiveness; less compromise of oxygenation, ventilation, and respiratory mechanics during the procedure; less likelihood of increasing intracranial pressure; less likelihood of inducing arrhythmias; availability where there is no bronchoscopist; lack of contamination presented by the bronchoscopic channel; availability to patients with small endotracheal tubes; and lower cost. Of the quantitative techniques, QEA is least invasive, most readily available, and least expensive, and it requires the least experience and is easily repeatable.
Where comparisons have been made, the authors of most studies have concluded that the diagnostic accuracies of nonbronchoscopic and bronchoscopic techniques are similar. Nonetheless, and although not noted by all studies, certain generalizations regarding the overall medical literature can be made.
(i) In some studies, the concordance between the sensitivity of bronchoscopic versus nonbronchoscopic quantitative cultures has been only approximately 80% (30, 94, 140). Consequently in some patients, particularly if the pneumonia is not diffuse and involves the left lung or upper lobes, the diagnosis of VAP could be missed by blind sampling.
(ii) Compared to nonbronchoscopic sampling methods, bronchoscopic quantitative cultures have greater specificity (104, 162).
(iii) Because BAL samples larger areas of lung, it as at least as sensitive as PSB and PTC (135). The sensitivity and negative predictive value of a culture for pneumonia are affected by the size of the sampling area and the amount of retrieved secretions. Bronchoalveolar lavage, which samples approximately 1 million alveoli, is estimated to recover 5 to 10 times the number of organisms obtained by PSB. Quantitative endotracheal aspirates would likewise be expected to provide more representative samples than PSB and PTC. Combining the results of PSB and BAL may increase sensitivity (187).
(iv) Other “technical” advantages of BAL over PSB are that the technique of smear preparation for direct microscopic examination of BAL is better established and that BAL is less likely to cause bleeding (28).
(v) Protected sampling methods such as PSB, PTC, and protected BAL, because they “bypass” the oropharyngeal and upper airway bacterial contamination/colonization, have superior specificity and positive predictive values (135).
(vi) Protected specimen brush is more specific than sensitive for the diagnosis of VAP. Consequently, a positive result increases the likelihood of pneumonia being present (199).
(vii) Direct visualization of the airways by bronchoscopy permits sampling from the airway that corresponds to the abnormal region on chest X ray or CT scan, to purulent secretions, and/or to maximal endobronchial abnormalities. Consequently, bronchoscopy should theoretically improve the sensitivity of the procedure, particularly for pneumonias involving the upper lobes and the left lung (94, 115).
(a) Timsit et al. reported that the presence of two or more of the following had a sensitivity of 78% and a specificity of 89% for diagnosing VAP: a decrease in the partial pressure of arterial oxygen (PaO2)/FiO2 ratio of ≥50 mmHg, distal purulent secretions, or persistence of distal secretions surging from distal bronchi during exhalation (195).
(b) In a study with autopsy serving as the gold standard to diagnose VAP, a bronchoscopic BAL with <50% neutrophil differential had a 100% negative predictive value for the diagnosis of VAP (98).
(viii) Compared to QEA and BBS, blind and bronchoscopic PSB, PTC, and BAL provide additional information that may be clinically useful.
(a) In one study of PSB, having fewer than 10% neutrophils on direct examination was uniformly associated with negative cultures, a finding that would contribute significantly to the specificity and positive predictive value of the procedure (142).
(b) Immediate performance of a direct microscopic examination, BAL, PBC, and PTC also enable a search for ICOs. If ≥2 to 5% of recovered cells contain ICOs, this result can potentially serve as a guide for the initial selection of empirical therapy. Unfortunately, as indicated above, the accuracy of this procedure is too low to be clinically useful in most circumstances. Not surprisingly, concomitant antibiotic administration increases the likelihood of false-negative results (53, 205). Moreover in one study, one-third of episodes of VAP caused by Pseudomonas aeruginosa were associated with negative direct stainings (205). Consequently, in the majority of cases, a negative direct staining still requires initial broad-spectrum antibiotics until culture results are returned.
(c) In a recent study by Michel et al., QEAs performed twice a week anticipated the etiology of a subsequent pneumonia in 83% of cases (146). This contrasts with the results of similar studies utilizing nonquantitative cultures of ETAs (78).
The key question, however, is not which quantitative culturing technique is more accurate but whether these techniques affect outcomes from VAP. Four studies have prospectively evaluated the impact of an invasive diagnostic strategy on the morbidity of, the use of antimicrobial drugs in, and the mortality of VAP. Unfortunately, the four studies had different designs as well as methodological flaws, including an inappropriate selection of diagnostic techniques for comparison, insufficient control for previous antimicrobial treatment, inconsistent ways of managing antibiotic treatment in patients who had negative microbiologic assays, and insufficient power to detect clinically important differences among alternative strategies.
In the three, randomized, controlled Spanish studies, no differences in mortality and morbidity were found when either invasive (PSB and/or BAL) or QEA techniques were used to diagnose VAP (175, 176, 186). However, these studies contained relatively few patients (51, 76, and 88 patients) and therefore were not powered sufficiently to demonstrate a difference in mortality. Moreover, antibiotics were continued in all patients, thereby negating one of the major potential advantages of any diagnostic test in patients clinically suspected of having VAP. In other studies, it has been shown that antibiotics can be safely stopped in patients with negative quantitative cultures (30). Although these studies did not demonstrate any difference in clinical outcomes, they did confirm that invasive tools are associated with a greater ability to narrow or discontinue antibiotics.
In a large, multicenter, randomized, unblinded French study of 413 critically ill patients with a clinical suspicion of pneumonia, bronchoscopy with quantitative cultures of PSB or BAL was compared to nonquantitative endotracheal aspirates (64). Patients with recent changes in antibiotic therapy were excluded, limiting the ability to generalize the results. Patients in the invasive diagnostic group had more antibiotic-free days in a 28-day period (11.4 versus 7.5 days), fewer antibiotics per day (1.0 versus 1.3), and less organ dysfunction at day 3 and 7. The mortality rate at 14 days was significantly lower in the invasive diagnostic group (16.2% versus 25.8%; P = 0.022); there was no difference at 28 days. However, when a multivariate analysis was performed, there was an improvement in mortality at 28 days (hazard ratio, 1.54; 95% confidence interval, 1.10 to 2.16; P = 0.01).
One problem with this study that could have confounded results was that the invasive diagnostic group had a much lower rate of inappropriate initial antibiotics (1 patient [0.5%] versus 24 patients [13%]; P < 0.001). Of the cohort that received inappropriate antibiotics, 33% died, including all in the noninvasive diagnostic group before day 14. Numerous studies have documented the importance of appropriate and early antimicrobial therapy for VAP (2, 90, 125). Therefore, the improved outcome at 14 days in the invasive diagnostic group may have been secondary to higher use of inappropriate antibiotics in the noninvasive diagnostic group and not to the actual invasive procedures.
A recent meta-analysis of the impact of invasive approaches on the diagnosis of VAP concluded that invasive lower airway sampling does not alter hospital mortality but consistently results in changes to the antibiotic regimen (182). In that study, the odds ratio for change in antibiotic management after invasive testing was 2.85 (95% CI, 1.45 to 5.59). In another outcomes study, the invasive bronchoscopic evaluation of VAP was also shown to allow de-escalation or narrowing of antibiotics to occur once organisms and their susceptibilities were identified (169). In contrast, in a decision analysis of antibiotic and diagnostic strategies for VAP, Ost et al. concluded that from the perspective of minimizing cost, minimizing antibiotic use, and maximizing survival, the best strategy was employing mini-BAL and treating with three antibiotics (160). While mini-BAL did not improve survival, it did decrease cost and antibiotic use (160).
At the present time, based on the available data, the optimal strategy for diagnosing VAP remains to be defined. The American Thoracic Society (ATS)/Infectious Disease Society of America guidelines do provide expert opinion supporting quantitative or semiquantitative cultures of respiratory specimens, although the panel favors invasive quantitative techniques (5). However, a large, better matched, multicenter, randomized study comparing quantitative cultures of endotracheal aspirates to quantitative cultures of bronchoscopic specimens to a clinical strategy using scoring systems and nonquantitative and semiquantitative cultures is still needed. Potential confounding variables such as antibiotic regimens and antibiotic discontinuation protocols must be controlled. Until such evidence exists, the use of invasive bronchoscopic techniques cannot be required for the routine diagnosis of VAP. Therefore, which diagnostic approach for VAP should be undertaken is up to the discretion of the clinician. Factors to consider include local experience, expertise, availability, and cost.
TREATMENT
Principles to apply when choosing appropriate therapy for VAP include knowledge of organisms likely to be present, local resistance patterns within the ICU, a rational antibiotic regimen, and a rationale for antibiotic de-escalation or stoppage. Although the clinician could know the organisms and sensitivities prior to the development of VAP (see “Antibiotic Management” below), this is often not the case. In the latter situation, empirical choices that provide adequate coverage are critical. Early effective therapy for VAP is associated with reduced mortality. Luna et al. demonstrated that inadequate therapy during the initial 48 h, despite provision of adequate therapy after BAL results, was associated with a mortality rate of 91% (125). When empirical therapy was appropriate, mortality rates were much lower (38%). Delays in the administration of appropriate antibiotic therapy for VAP have been associated with excess mortality (2, 90, 125). In one study, a delay in appropriate therapy for 24 h or more was associated with a 69.7% mortality, compared to 28.4% in patients treated without the delay (P < 0.001) (90). Consequently, once VAP is considered, cultures must be obtained quickly and treatment initiated without delay. VAP should be considered with a CPIS score of >6, as illustrated in Table 2, or alternatively, with a new pulmonary infiltrate and at least two of the following: fever, leukocytosis, and purulent secretions.
CPIS
As multiple etiologies may explain why patients develop a fever and pulmonary infiltrates while receiving mechanical ventilation, we often search for other infectious and noninfectious etiologies concurrently with evaluation for VAP. The extent of this investigation is dictated by the clinical circumstances, including physical examination, laboratory findings, and the severity of illness (Tables 3 and 4). In patients with sepsis, a definite site of infection cannot be found in 20 to 30% (212). As delays in treating severe sepsis significantly increase mortality, we are very hesitant to discontinue antibiotics in patients with severe sepsis, even if initial respiratory and other cultures are negative. In such critically ill individuals, we usually continue broad-spectrum antibiotics as we continue to aggressively pursue other infectious and noninfectious causes of the patient's presentation (Tables 3 and 4). However, because VAP is rarely occult, we direct our antibiotic coverage and diagnostic efforts at non-VAP causes of sepsis.
Routine care of patients suspected of having ventilator-associated pneumonia
Evaluation for infectious (other than VAP) and noninfectious causes of fever
There is a general consensus that VAP is very likely in certain situations. These circumstances are outlined in Table 5 (198, 215). However, either such scenarios are uncommon or the procedures required are undesirable or contraindicated in critically ill patients on mechanical ventilators. Therefore, in most cases the clinician has a choice of two strategies for managing suspected VAP (Table 6). One strategy is based on clinical criteria and nonquantitative or semiquantitative cultures of tracheal aspirates. The other strategy utilizes quantitative cultures of respiratory specimens. The quantitative culture approach can be further divided into bronchoscopic (invasive) and nonbronchoscopic (noninvasive) strategies. As outlined above, each strategy has its own advantages and disadvantages.
High probability of VAP
Evaluation of fever in patients suspected of having ventilator-associated pneumonia
Bronchoscopy allows the direct examination of respiratory secretions from BAL, PSB, and PTC to determine the percentage of cells containing ICOs. Some experts cite this potential for early guidance of antibiotic management as a factor favoring the bronchoscopic approach to the management of VAP over other strategies (29, 30). However, and as outlined above, the false-negative rate of direct staining is alarmingly high, particularly with concomitant antibiotic use and with VAP caused by Pseudomonas. Consequently, we contend that a negative direct staining still requires initial broad-spectrum antibiotics until culture results are returned, particularly if antibiotics have been added or changed in the previous 72 h.
Quantitative Culture StrategyAlthough there is no definitive evidence that quantitative cultures clearly improve patient outcomes, we favor a quantitative culture strategy for the management of suspected VAP. The superior specificity of quantitative compared to nonquantitative and semiquantitative culture techniques permits us to more confidently discontinue antibiotics and thereby avoid the attendant complications, including the potential for increased bacterial resistance. In addition, a negative quantitative culture compels us to more aggressively search for other noninfectious and nonpulmonary infectious causes of the patient's presentation.
Our ICU practice is to rely on QEAs, as most studies have concluded that the sensitivities of nonbronchoscopic and bronchoscopic quantitative techniques are comparable. However, the overall concordance in some studies has been only approximately 80% (30, 94, 140). That is, in some patients, the diagnosis of VAP could be missed by blind, nonbronchoscopic sampling, particularly if the pneumonia involves the left lung or upper lobes. Moreover, the additional information obtained from direct visualization of the airways, percentage of neutrophils on BAL, and percentage of epithelial cells and neutrophils on direct staining of PSB and BAL aid in our decision making. When we utilize bronchoscopy, we favor BAL over PSB because of its better safety profile and because it samples a greater area of the lung, which should theoretically improve sensitivity and negative predictive value. Performing both BAL and PSB may increase sensitivity further (see above).
If the quantitative culture strategy is employed, it is essential to interpret quantitative culture results in the clinical context. Consider a quantitative BAL culture yield of 103 CFU/ml from a mechanically ventilated patient obtained 48 h after administration of broad-spectrum antibiotics. This is below the 104-CFU/ml threshold, but antibiotics given or changed within the 72 h prior to obtaining a quantitative culture can decrease the bacterial burden and result in a false-negative quantitative culture. In the appropriate clinical context, such a result can be interpreted as consistent with the presence of VAP. In contrast, the same culture result obtained for an individual on no antibiotics or without a change in the previous 72 h would be less indicative of VAP. Moreover, if available, the percentage of neutrophils on BAL differential, the percentage of neutrophils and epithelial cells on direct staining of BAL and PSB, the percentage of organisms containing ICOs, and visual inspection of the airways can also be useful in individual cases (see above).
Clinical StrategyThe primary advantage of a clinical strategy for diagnosing VAP is that it does not require specific expertise or specialized equipment or techniques and is noninvasive. Therefore, such an approach can be utilized anywhere. However, because of the poor specificity of clinical signs and symptoms of VAP and of nonquantitative or semiquantitative cultures of tracheal secretions, relying on the clinical approach would be expected to result in treating noninfectious processes with broad-spectrum antibiotics as well as potentially failing to recognize and pursue noninfectious mimics of VAP and nonpulmonary infections.
Antibiotic ManagementThe ATS has recently published guidelines to guide empirical antibiotic choices (5). These guidelines are divided into those for patients at risk for VAP caused by multidrug-resistant organisms and those for patients without such risk. Risk factors for multidrug-resistant organisms include prior antimicrobial therapy in the preceding 90 days, current hospitalization exceeding 5 days (not necessarily ICU days), high frequency of resistance in the community or local hospital unit, and immunosuppressive disease and/or therapy. In addition, the clinician must consider risk factors for health care-associated pneumonia, as such a pneumonia may present with multidrug-resistant organisms even upon hospital admission (5). Such risk factors for the intubated patient include a hospitalization for >2 days within the preceding 90 days, residence in a long-term care facility, chronic dialysis within 30 days, home wound care, home infusion therapy (inclusive of antibiotics), and a family member with a multidrug-resistant pathogen.
In the absence of risk factors for multidrug-resistant bacteria, the clinician should choose empirical therapy for Streptococcus pneumoniae, Haemophilus influenzae, methicillin-sensitive Staphylococcus aureus, and antibiotic-sensitive gram-negative enteric organisms. Antibiotic choices include ceftriaxone, quinolones (levofloxacin, moxifloxacin, or ciprofloxacin), ampicillin/sulbactam, or ertapenem (Fig. 1). When risk factors for multidrug-resistant organisms are present, the clinician must consider not only the organisms listed above but also Pseudomonas aeruginosa, Klebsiella, Enterobacter, Serratia, Acinetobacter, Stenotrophomonas maltophilia, Burkholderia cepacia, and methicillin-resistant S. aureus. Empirical therapy is broadened to include (i) either an antipseudomonal cephalosporin (cefepime or ceftazadime), an antipseudomonal carbepenem (imipenem or meropenem), or a β-lactam/β-lactamase inhibitor (pipercacillin-tazobactam) plus (ii) an antipseudomonal fluoroquinolone (ciprofloxacin or levofloxacin) or an aminoglycoside (amikacin, gentamicin, or tobramycin) plus linezolid or vancomycin.
Algorithm for diagnosis and treatment of VAP. (©American Thoracic Society. Adapted from the American Journal of Respiratory and Critical Care Medicine [5] with permission.) Antibiotic choice can be tailored to the pathogens’ last sensitivity report should QEA surveillance cultures be obtained twice weekly and should the growth level exceed 100,000 CFU/ml (146.)
While the complex regimen outlined above is appropriate, creating a milieu of further resistant organism must be a concern, as it will lead to fewer opportunities to choose effective empirical therapy. As noted in “DIAGNOSIS” above, there is considerable controversy over the use of quantitative cultures and which quantitative culturing technique to use. Michel et al. applied QEA as a surveillance tool and routinely obtained samples twice weekly (146). Sensitivities were determined for microorganisms present at a concentration of ≥103 CFU/ml. When VAP occurred, the most recent QEA preceding VAP was used to direct antibiotic therapy, and a BAL was obtained to assess the appropriateness of the antibiotic regimen. Those authors also compared results of BAL to empirical regimens that would have been chosen by the classification of Trouillet et al. and the 1996 ATS consensus guidelines (6, 203). The antibiotic regimen as guided by QEA was appropriate in 95% of cases. This was not statistically different from the appropriateness of the empirical regimens chosen by the strategy of Trouillet et al. (83% appropriate) but was superior to that of the empirical choices suggested by the 1996 ATS guidelines (68% appropriate). This approach is very new, and the cost is that of culturing and determining sensitivities (if the threshold is exceeded). The benefit is that it appears to provide a high likelihood for appropriate initial therapy. Furthermore, it will likely reduce the application of overly broad antibiotic regimens, hence reducing the likelihood of inducing more multidrug-resistant organisms.
Considerable controversy surrounds monotherapy versus combination therapy for patients with VAP. The primary reasons for combination therapy are to prevent the development of resistance, improve outcomes, provide synergy, and provide sufficient antibiotic coverage should the pathogen be resistant to the agent that would have been chosen as single therapy. The former two arguments, while logical, have yet to be proven (36, 209). In fact, a meta-analysis suggested that clinical failure was more common with combination therapy, as was nephrotoxicity; aminoglycosides were the second agent, and combination therapy did not prevent new resistance patterns (209). However, given that mortality is higher when therapy is inappropriate during the first 48 h, we favor initiating combination therapy for patients at risk for multidrug-resistant organisms until sensitivities are known. This is consistent with an approach suggested by Gruson et al. (75).
Commonly employed methods to reduce the development of resistance include de-escalation therapy, truncated courses of antibiotics, dosing regimens that account for patient-antibiotic pharmacokinetics and pharmacodynamics (PK/PD), antibiotic cycling, and surveillance cultures. Most intensivists have embraced the former two; however, the latter two remain controversial. The ATS has put forth a management strategy to address de-escalation and early stoppage of antibiotics (5). Upon suspicion of VAP, empirical antibiotics are initiated and lower respiratory tract cultures obtained. At 48 to 72 h, if the patient is improving and cultures are negative, strong consideration should be given to stopping antibiotics. Rello et al. have suggested truncating the course at <5 days provided the patient has been afebrile for >48 h (169). Should the culture results be positive and the patient has improved at 48 to 72 h, then the ATS guidelines suggest de-escalation (reduction in antibiotics to be administered, including potential for monotherapy) and treating patients without P. aeruginosa, Acinetobacter, or Stenotrophomonas maltophilia for 7 to 8 days. A longer course is indicated for P. aeruginosa, Acinetobacter, and Stenotrophomonas maltophilia.
The antibiotic regimen (choice and dosing) should be reevaluated for change or prolongation in patients with poor clinical responses, which may be assessed by a rising CPIS. A rising CPIS has been associated with higher mortality (183). These recommendations are based on the results from studies by Dennesen et al., Luna et al., Singh et al., and Ibrahim et al. (49, 88, 122, 183). Such strategies are dependent upon clear evidence of patient improvement as defined by reduction in serial CPISs or improvement of the PaO2/FiO2 ratio at days 3 to 5 (122). The technique chosen in obtaining microbiologic data may indeed affect the clinical decision to de-escalate therapy. Heyland et al. reported that the choice of bronchoscopic BAL and PSB resulted in increased physician confidence in the diagnosis and management of VAP; this resulted in a greater tendency to limit or discontinue antibiotics, an outcome that was echoed in a recent meta-analysis (79, 182).
Singh et al. proposed another potential clinical strategy to minimize unnecessary antibiotic use for VAP and the potential consequences (183). In this study, patients with a modified CPIS of ≤6 on day 1 (Table 2) were randomized to receive either standard antimicrobial therapy or ciprofloxacin monotherapy, with reevaluation at 3 days. In the ciprofloxacin monotherapy group, if the CPIS remained at ≤6 at day 3, antibiotics were discontinued. Continuation of antibiotics in the standard therapy group was left up to the discretion of the attending physician but occurred in 96% of patients. Despite monotherapy with ciprofloxacin, a shorter duration of treatment (P = 0.0001) and lower cost (P = 0.003), mortality, and length of ICU stay did not differ. In addition, antimicrobial resistance, superinfections, or both were less in the experimental group than in the standard therapy group (15% versus 36%; P = 0.017). Such an approach recognizes that a gold standard for diagnosing VAP does not exist, and consequently the approach does not attempt to discern whether the patient did or did not have pneumonia. Rather, the goal was to identify patients for whom a shorter course of antibiotic therapy would suffice. In a subsequent study that incorporated the modified CPIS, 41% of patients with a score of ≤6 did not have pneumonia by quantitative BAL culture (65). Therefore, one potential explanation for the good outcome in the study by Singh and colleagues is that many of the patients did not have pneumonia.
The antibiotic regimen may be appropriate but the dose or frequency not appropriate. This is important not only in treating resistant organisms but in preventing the development of resistance, by means of eradications. Three tools to predict antibiotic efficacy are to assess the peak concentration achieved, the time the antibiotic exceeds the MIC, and the extent to which the area under a concentration-time plot exceeds the MIC (Fig. 2). Coupling an understanding of the mechanisms of how bacteria are eradicated and PK/PD parameters results in reduced mortality and morbidity (3, 4, 18, 84, 120, 151, 152, 177, 217).
Pharmacodynamic and pharmokinetic approach to antibiotic therapy.
The ATS guidelines also address the nonresponding patient with negative and positive cultures (5). A change in antibiotic coverage is warranted should the culture results indicate that empirical therapy was inappropriate. However, if culture results are negative or appropriate antibiotic regimens were chosen and the patient has not improved, then the clinician should consider other organisms, other diagnoses, or a complication of the disease or therapy (Fig. 1). Such a decision is generally made after 72 h of therapy, as most patients respond within this time frame (122, 123). At this juncture, quantitative cultures are warranted. Montravers et al. have demonstrated that clinical failure rarely occurred when protected brush samples recovered organisms at <103 CFU/ml (7% failure rate). Conversely, higher failure rates were seen when culture results exceeded 103 CFU/ml (55.8% failure rate) (150). Persistent fever or failure to improve with antibiotic therapy may indicate that the inciting process is noninfectious. Other diagnoses include atelectasis, congestive heart failure, venous thromboembolic disease, pancreatitis, chemical pneumonitis from aspiration, proliferative phase of acute respiratory distress syndrome, drug fever, or pulmonary hemorrhage (5). Alternatively, the process may be infectious but not VAP. The clinician should consider empyema, lung abscess, Clostridium difficile colitis, urinary tract infection, and sinusitis (139, 172) (Table 4).
Candida species, while commonly cultured from patients, rarely cause invasive pulmonary disease, even when quantitative thresholds are exceeded (58, 166, 210). However, Candida may be a marker that the patient is more likely to develop VAP with P. aeruginosa (a causal relationship has not been demonstrated) (8). Azoulay et al., in a surveillance study, noted that patients colonized with Candida species were 1.58 times more likely to develop VAP and 2.22 times more likely to develop P. aeruginosa (8). The three most common Candida species recovered were C. albicans (2/3), C. glabrata (1/5), and C. tropicalis (1/8). Those colonized were older and were more likely to have respiratory failure as a reason for ICU admission. These patients had longer courses of mechanical ventilation, received more antibiotics, and experienced higher hospital mortality. However, there are no data to support the routine administration of antifungal therapy when Candida species are found in pulmonary secretions of mechanically ventilated patients.
After considering VAP, the clinician should promptly institute therapy. Choosing an appropriate antibiotic regimen, defined by sensitivities of the organism cultured and by dosing regimen ordered, is paramount, as the first 48 h is critical to patient survival. Surveillance cultures may provide better guidance than empirical strategies. Truncated courses of antibiotics are indicated for culture-negative improving patients and for VAP not caused by P. aeruginosa, Acinetobacter, and Stenotrophomonas maltophilia. Such a practice may reduce the likelihood of colonization with multidrug-resistant organisms or creating a local environment of resistant organisms.
PREVENTION
Clinicians must focus on eliminating or minimizing the incidence of VAP through preventive techniques (Fig. 3). While little has affected the incidence of late-onset VAP, the occurrence of early-onset VAP can be reduced by simple measures. Data have accumulated to support interventions and establish guidelines, yet translation into practice is lacking. Health care team compliance rates vary between 30 and 64% (39). The focus should be addressing modifiable risk factors such as endotracheal and nasogastric tubes, tracheotomy, reintubation, enteral nutrition, corticosteroid administration, gastric pH-modifying agents, supine positioning, prior antibiotic usage, poor infection control practice, and contaminated respiratory equipment, medications, or water (24, 42, 100, 102, 155, 191).
Potential strategies to prevent VAP.
Noninvasive mechanical ventilation (NIV) has been associated with more favorable outcomes (mortality and morbidity) in comparison to endotracheal tube placement in patients with acute exacerbations of chronic obstructive pulmonary disease or acute pulmonary edema (7, 21, 117, 118). The incidence of nosocomial pneumonia was reduced in the group randomized to NIV (7, 27, 74, 76, 157). Furthermore, immunocompromised patients with bilateral infiltrates also benefited from NIV over invasive mechanical ventilation (IMV) with regard to both mortality and morbidity (81). Yet clinicians have significant reluctance to initiate NIV, perhaps because of patient intolerance or increased resource consumption (nursing and respiratory therapy).
Once the decision to intubate is made, the practice of VAP prevention should be directed at reducing colonization and aspiration (volume of organisms presented to the lungs). This begins with choosing the oral route of intubation and focusing on minimizing the duration of mechanical ventilation (DOMV). Oral intubation is preferred over nasal intubation, as the latter has been associated with both VAP and sinusitis, with the same bacteria identified in both. Rouby et al., demonstrated a significant reduction in nosocomial sinusitis when patients are orally cannulated with endotracheal and gastric tubes (172). Holzapfel et al. have linked the reduction in nosocomial sinusitis to a reduction in VAP (83). Furthermore, the clinician must give careful attention to the mundane and seemingly small interventions, such as regularly assessing endotracheal cuff pressure, performing endotracheal suctioning, draining ventilator tube condensate, avoiding gastric overdistention, avoiding the supine position, avoiding unnecessary ventilator circuit changes, application of heat and moisture exchangers (HMEs) when appropriate, minimizing out-of-ICU transports, and regular hand cleaning with soap or alcohol disinfectant. Maintaining cuff pressure of endotracheal tubes at ≥20 mm Hg reduces nosocomial pneumonia, presumably by minimizing the passage of oropharyngeal contents into the trachea (192).
The duration of intubation directly affects the likelihood of VAP, which is more evident in patients with ICU LOS exceeding 5 days. Fagon et al. suggested that the incidence of VAP increases by 1% per day of IMV (62). However, Cook et al. found that the incidence per day varies over time, with 3% per day during first 5 days of IMV, 2% for the second 5 days, and 1% for the subsequent 5-day period (39). This observation is supported by Ibrahim et al., who identified an incidence rate of VAP of 11.5%, 56% of which were early onset (≤5 day) (87). Hence, the greatest attack rates appear to be during the initial days of mechanical ventilation. Additionally, significant risk factors for early-onset VAP include cardiopulmonary resuscitation and continuous sedation (167).
Continuous sedation is more often administered in the acute phase of an illness. In addition to treating the primary cause of respiratory failure, the DOMV can be reduced through judicious use of sedatives and analgesics. Studies by Brook et al. and Kress et al. have demonstrated that protocols for sedative and analgesic administration with the goal of minimizing constant infusions led to reduced DOMV (22, 113). Furthermore, daily interruption of sedation results in a reduced incidence of intensive care unit complications, in which VAP was included (127, 180). Weaning protocols have also resulted in reduced DOMV, whether respiratory therapist initiated or not (59, 67).
Patients should be cared for in the semirecumbent position to reduce the extent of aspiration, especially when receiving enteral feeds. Radionuclide studies reveal increased tracheal penetration of gastric contents when intubated patients are supine (85, 159, 201). Drakulovic et al. found that the simple elevation of the head of bed to 45° results in dramatic reductions in VAP incidence and a trend toward reduced mortality (54). Nonetheless, a recent survey by the University Hospital Consortium revealed that compliance with the simple and no-cost intervention of elevating the head is woefully low, and a study by Heyland et al. revealed that the head of bed is on average elevated to 29° and not 45° (80). Kinetic bed therapy has also led to a reduction in the incidence of VAP (46, 48, 69, 72, 97, 202, 213). However, this is costly and has not been directly compared to head-of-bed elevation, a no-cost option.
Some VAP is contracted from inhalation of bacteria through the ventilator circuit and may be a result of contaminated aerosols, condensate, or suction catheters. Traditionally, ventilator circuit changes have been on a regular schedule and often daily. However, the data examining this practice reveal that there is no benefit to changing the circuit on a regular basis, and the present recommendations are to change the circuit when soiled (56, 108, 121). Such a practice would likely reduce the rate of accidental spillage of condensate into the airway. As heated humidifiers enhance the amount of condensate, attention has been focused on HMEs. These devices have led to a reduction in VAP, albeit small, and should be used in patients without significant secretions or concern over the risk of obstruction (17, 55, 99, 107, 132, 141, 174). While changing the HME less frequently than every 48 h may lead to further reductions in VAP, care must be taken to carefully monitor for trapped secretions and subsequent airway obstruction or increments in the work of breathing (45, 193).
Endotracheal suctioning of intubated patients can be performed through an open or closed system. In theory the closed system could reduce the incidence of VAP, but in practice this has not been demonstrated (35, 50, 93, 106, 219). Cost analysis favors the closed system, as the enveloped catheter can be reused for suctioning and needs to be changed only when dysfunctional (52). However, respiratory therapists have voiced concerns over residue buildup within the lumen of the endotracheal tube.
As most VAP follows from aspiration of oropharyngeal secretions, attention to proper cuff inflation pressures and endotracheal suctioning can affect the volume presented to the trachea. The application of continuous suction of subglottic secretions through specialized endotracheal tubes will reduce the incidence of VAP (110, 126, 181, 185, 204). Surprisingly, this was not associated a reduction in mortality, ICU LOS, or duration of mechanical ventilation. While studying the application of continuous subglottic suctioning, Rello et al. noted a trend of increased VAP in patients with endotracheal cuff pressures of <20 cm H2O (168). Hence, it is recommended not only to assess cuff pressure for tracheal ischemia (which occurs when pressure exceeds 30 cm H2O) but also to ensure that adequate cuff pressure (>20 mm Hg) is present.
The endotracheal tube itself is a reservoir for gram-negative bacteria. The buildup of a biofilm within endotracheal tubes occurs frequently. One study demonstrated that 84% of endotracheal tubes examined had a biofilm (188). As documented by Inglis et al., this biofilm is heavily laden with bacteria, usually gram-negative organisms (66, 89). At present, ongoing studies are directed at either eliminating this biofilm or reducing the bacterial load associated with it.
Oral decontamination with chlorhexidine has been shown to reduce the incidence of VAP in patients undergoing cardiac surgery, presumably by reducing oropharyngeal colonization (51). Furthermore, numerous studies with oral decontamination antibiotic pastes alone or coadministered with systemic antibiotics have shown a reduction in early VAP (1, 13, 47, 163, 171). Two meta-analyses have suggested better results with oral decontamination alone than with the combination of oral and systemic prophylaxis (44, 154). With either approach, however, concern over the emergence of antibiotic-resistant organisms has tempered use, as has the labor intensity required to apply these regimens at the bedside. This is particularly true in ICUs housing organisms with high antibiotic resistance rates (68, 71, 77, 119, 147, 148, 208, 214). While often recommended, it appears not to be routinely practiced. Two recent studies will further the debate, as they demonstrated significant reductions in VAP and mortality with selective decontamination of the digestive tract (47, 114). These two studies were performed under conditions where selective decontamination of the digestive tract is most effective, i.e., surgical intensive care units housing patients less likely to be colonized with resistant bacteria.
Gastric volumes and acidity affect the incidence of VAP. Reducing the acidity of gastric secretions and feeding will reduce bacterial overgrowth. However, in high-risk patients (ventilated for >48 h and coagulopathic), the risk of bleeding outweighs the risk of VAP from pH-modifying agents (52). Hence, it is difficult to recommend against H2 blockers or proton pump inhibitors. Sucralafate may indeed be superior from the viewpoint of VAP, but it is less effective with regard to prophylaxis of gastrointestinal bleeding, and thus it use is not warranted over H2 blockers or proton pump inhibitors (16, 37, 143).
Multiple studies have examined postpyloric versus gastric feedings with regard to incidence of aspiration and development of VAP. These studies were small and inconclusive. In a meta-analysis, postpyloric feedings reduce the incidence of VAP and increased the nutrition delivered (80). However, no single trial demonstrated that postpyloric tube feedings prevent VAP. The improved delivery of nutrition was likely the result of decreased gastric residual assessments and consequently fewer stoppages in continuous tube feedings. A recent publication favored a delay of greater than 5 days before initiating tube feedings, as the incidence of VAP was reduced (86). Further data are needed to unconditionally embrace this practice.
Preventing Multidrug ResistanceAntibiotic cycling remains controversial. Employing a rotational schedule for empirical antibiotic administration for suspected VAP may indeed lead to a reduced incidence of resistant organisms (75, 111, 165). While such a strategy may not reduce the incidence of VAP, reductions in mortality may be seen (165). This is likely a result of changes in resistance patterns resulting in a higher likelihood of choosing appropriate antibiotic regimens (112). Because rotational schedules have primarily targeted reducing the resistance of gram-negative organisms, we do not know the impact of rotating antibiotics against gram-positive organisms, such as methicillin-resistant S. aureus. Furthermore, the frequency with which to rotate antibiotics remains unclear, as monthly and quarterly regimens have been assessed with documented successes (75, 165). Furthermore, the probability of antibiotic cycling leading to a reduction in antimicrobial resistance is low as determined through mathematical modeling (14). At this juncture, it is premature to recommend rotating antibiotics or a rotational schema.
Multidrug resistance can also be reduced when patient-antibiotic PK/PD characteristics are accounted for. Early eradication minimizes the opportunity for a population of organisms to develop resistance. Peak concentrations for aminoglycosides 10-fold greater than MIC appear to inhibit the emergence of resistant organisms (178, 207). When choosing fluoroquinolones, resistant organisms are less likely to be seen when the 24-h area-under-the-curve/MIC levels are >100 for gram-negative bacteria and >40 for gram-positive bacteria (177, 179, 194). Changes in medication frequency or infusion rates can increase the time that the antibiotic concentration exceeds the MIC. For β-lactams, monobactams, glycopeptides, and cabapenems this can be important in enhancing bactericidal activity, again reducing opportunities for resistant organisms to emerge (18, 19, 105, 120, 217).
In summary, several opportunities to reduce the incidence of VAP are available to the clinician. Many are no-cost or minimal-cost interventions and should be implemented as part of routine care protocols. Care of the critically ill should be directed at applying interventions that reduce mortality, minimize morbidity, shorten the length of stay, and reduce cost. Reducing VAP through the simple measures outlined does exactly that. We recommend that the clinician's practice include noninvasive mechanical ventilation over intubation when appropriate, oral intubation when an endotracheal tube is necessary, orogastric over nasogastric tubes, elevation of the head to at least 30°, minimization of sedation, administration of a proton pump inhibitor when prophylaxis is indicated, a frequency of ventilator tubing changes at 7 days or when soiled, avoidance or elimination of endotracheal tube leak, good technique in removal of condensate, and of course excellent hand hygiene. At this time we do not support the routine use of endotracheal tubes with subglottic suction capabilities, rotational beds, in-line suction systems, rotational antibiotic schemes, or selective gut decontamination.
Strategies and a more thorough discussion on prevention are within the ATS/Infectious Disease Society of America statement and papers by Kollef and by Dodek et al. (5, 52, 101). Zack et al. have demonstrated that a multifaceted and multidisciplinary approach to VAP prevention can indeed reduce the incidence (218). Success is dependent upon persistent attention to detail, high compliance rates, and a champion.
CONCLUSION
A low threshold for suspicion of VAP is needed when a patient's clinical course deteriorates. The day 1 CPIS can be useful, especially when combined with quantitative cultures. The choice of which quantitative culture methodology to use is an open debate. However, diagnostic cost favors QEA, which can also be implemented as a surveillance technique. However, the clinician is more likely to stop antibiotics with a more invasive quantitative culture, resulting in increased savings.
Antibiotic administration should be promptly initiated when VAP is suspected and quantitative cultures obtained and should be broad in coverage. Knowledge of local antibiograms should guide the choice of antibiotics, in addition to likelihood of organisms (early- or late-onset VAP). For patients already on antibiotics at the time of suspected VAP, the clinician should choose antibiotics from different classes, as it is likely that resistance to “in-use” antibiotics has developed.
Assessment of the likelihood of VAP by day 3 is needed to decide whether antibiotics should be continued. The assessment should include a repeat CPIS, as the change in CPIS can guide clinical decisions, even stoppage of antibiotics. Assessment of quantitative culture results and sensitivities at this juncture is prudent, as it may permit early antibiotic de-escalation by choosing a more narrowly focused agent(s). Monotherapy may be appropriate in many instances of VAP and should reduce the incidence of drug resistance. A change to monotherapy may be possible in a responding patient where organism sensitivity results permit. A short course (6 to 8 days) can be administered to patients with VAP but is dependent on the patient physiologic response to treatment along with which organisms have been recovered (see above) (32, 144).
Simple and effective preventive measures can be instituted easily and at minimal costs. Such measures might include NIV, diligent respiratory care, hand hygiene, elevation of head, oral and not nasal cannulation, minimization of sedation, institution of weaning protocols, judicious use of antibiotics, de-escalation, and leveraging PK/PD characteristics for antibiotics administered. More costly interventions should be reserved for appropriate situations.
Utilizing the preventive, diagnostic, and treatment recommendations outlined in this paper should allow for improved outcomes for a common and serious medical complication seen in ICU mechanically ventilated patients.
- Copyright © 2006 American Society for Microbiology
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.↵
- 102.↵
- 103.↵
- 104.↵
- 105.↵
- 106.↵
- 107.↵
- 108.↵
- 109.↵
- 110.↵
- 111.↵
- 112.↵
- 113.↵
- 114.↵
- 115.↵
- 116.↵
- 117.↵
- 118.↵
- 119.↵
- 120.↵
- 121.↵
- 122.↵
- 123.↵
- 124.↵
- 125.↵
- 126.↵
- 127.↵
- 128.↵
- 129.↵
- 130.↵
- 131.↵
- 132.↵
- 133.↵
- 134.↵
- 135.↵
- 136.↵
- 137.↵
- 138.↵
- 139.↵
- 140.↵
- 141.↵
- 142.↵
- 143.↵
- 144.↵
- 145.↵
- 146.↵
- 147.↵
- 148.↵
- 149.↵
- 150.↵
- 151.↵
- 152.↵
- 153.↵
- 154.↵
- 155.↵
- 156.↵
- 157.↵
- 158.↵
- 159.↵
- 160.↵
- 161.↵
- 162.↵
- 163.↵
- 164.↵
- 165.↵
- 166.↵
- 167.↵
- 168.↵
- 169.↵
- 170.↵
- 171.↵
- 172.↵
- 173.↵
- 174.↵
- 175.↵
- 176.↵
- 177.↵
- 178.↵
- 179.↵
- 180.↵
- 181.↵
- 182.↵
- 183.↵
- 184.↵
- 185.↵
- 186.↵
- 187.↵
- 188.↵
- 189.↵
- 190.↵
- 191.↵
- 192.↵
- 193.↵
- 194.↵
- 195.↵
- 196.↵
- 197.↵
- 198.↵
- 199.↵
- 200.↵
- 201.↵
- 202.↵
- 203.↵
- 204.↵
- 205.↵
- 206.↵
- 207.↵
- 208.↵
- 209.↵
- 210.↵
- 211.↵
- 212.↵
- 213.↵
- 214.↵
- 215.↵
- 216.↵
- 217.↵
- 218.↵
- 219.↵