INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A wide variety of pathogenic organisms can cause lower respiratory tract infections (1). In the management of this condition, specific antimicrobial therapy should be directed at the pathogen responsible in each individual case. However, clinical, laboratory, and radiographic features are seldom useful in discriminating between potential pathogens. It is thus accepted practice to use microbiologic methods in an attempt to determine the precise etiology in an individual patient.

The bacteriologic diagnosis of pulmonary infections in intubated patients is still a controversial issue (2, 3). The diagnostic challenge for the microbiology laboratory is complicated by the need to differentiate between organisms responsible for infection and colonizing flora. A consensus conference recommended bronchoscopic techniques to determine the true etiology in episodes of ventilator-associated pneumonia (VAP) (4), and these techniques have become the standard for research. However, some investigators have argued against routine use of these techniques and have suggested empiric therapy or less invasive techniques as more cost-effective approaches in clinical practice (2).

Woodhead and colleagues (5) reported that routine microbial investigation of all adults admitted to hospital with community-acquired pneumonia was unhelpful and probably unnecessary. However, the contribution of the different microbiologic investigations to the final antibiotic therapy in nosocomial episodes has not yet been assessed. Although different studies have reported that specific etiologies are associated with a worse survival (6), it is not certain that knowing the etiology of VAP to determine whether to use antibiotics improves outcome. Indeed, a recent report (7) suggested that when inadequate initial antibiotic therapy for VAP is modified because of BAL results, the outcome is no better than if inadequate therapy is given.

The aim of this study was to document the usefulness of the results obtained from microbiologic investigations used in clinical practice in an unselected group of adult patients in whom VAP was suspected. Our specific goals were: (1) to evaluate the impact of these investigations on modifying initial antibiotic therapy, and (2) to determine whether clinical outcome can be improved as a result of a guided antibiotic strategy policy.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Selection of Patients

During a 38-mo period, a prospective study of intubated patients with clinical suspicion of pneumonia (n = 250) was carried out in our medical-surgical intensive care department. Attending physicians were blind to the study (excepting J.R.), so as to avoid bias in the therapeutic approach. Our guidelines included a bronchoscopy (with protected specimen brushing and bronchoalveolar lavage), except when these procedures were contraindicated or in the presence of technical problems, in all intubated patients in whom pneumonia was suspected. Two blood cultures were carried out simultaneously in most patients, as were pleural fluid cultures if present. These procedures were performed before an empirical antibiotic regimen was started or before new antibiotics were prescribed in patients with prior antibiotic therapy. The initial therapeutic strategy was not standardized, and the regimen was selected by the attending physician. Episodes of recurrent pulmonary infection, developed at weekends (when immediate respiratory cultures could not be performed) or in patients who died within the first 24 h after pneumonia diagnosis were excluded, and no further follow-up was performed in these groups.

Ventilator-associated pneumonia was suspected when new and persistent pulmonary infiltrates not otherwise explained appeared on chest radiographs. Moreover, at least two of the following criteria were also required: (1) fever  38° C; (2) leukocytosis  10,000 mm³; (3) purulent respiratory secretions. A pneumonia was considered ventilator-associated when it occurred after 48 h of mechanical ventilation (MV) and was judged not to have been incubating before starting MV (6). Fiberoptic bronchoscopic examination using a protected specimen brush or bronchoalveolar lavage was performed within the first 12 h after the development of a new pulmonary infiltrate. The etiology was confirmed if the protected specimen brush yielded  1,000 CFU/ml or when the bronchoalveolar lavage yielded  10,000 CFU/ml of a pathogen microorganism. Positive qualitative cultures from pleural fluid or blood samples also confirmed the etiology.

Final Diagnosis

Of the initial 250 patients with suspected pneumonia, the study population consisted of 114 who were retrospectively considered to have true pneumonia. The determination of whether a patient had pneumonia was made by absolute consensus on clinical rounds in a daily meeting of all attendants in our department. Another 45 patients who developed pneumonia during a weekend and 10 with superinfections were not enrolled in this study. The final diagnosis of "no pneumo-nia" was established in 44 patients by absolute consensus on clinical grounds after the demonstration of either an alternative cause for any of the inclusion criteria or disappearance of radiographic opacities during the first 48 h of inclusion. Finally, in 37 patients, there was disagreement among the clinicians about the presence or absence of pneumonia, and these patients were considered indeterminate. Episodes in which the final diagnosis of pneumonia was rejected or considered indeterminate were excluded from study.

Definitions

Appropriate therapy was defined as the use of at least one antibiotic to which all isolates were susceptible in vitro from the moment in which bronchoscopy was performed. In presence of Pseudomonas aeruginosa, at least two active agents (combination therapy) were required. Clinical resolution was defined in patients who had complete resolution of all signs and symptoms of pneumonia along with improvement, or lack of progression, of all abnormalities on the chest radiograph (8). Patients were deemed clinically improved if fever disappeared and if pulmonary infiltrates and physical signs of pneumonia abated (9). Deaths were considered related to the pulmonary infection if occurring before any objective response to antimicrobial therapy or if the pulmonary infection was considered a contributing factor to death in patients with a comorbidity (10).

The excess mortality caused by inappropriate initial therapy was determined by subtracting the crude mortality rate when the patient was already receiving appropriate empiric therapy from the crude mortality rate of cases in which it was modified after bronchoscopy because of isolation of a resistant organism. If combination therapy was started because of the presence of P. aeruginosa, if an ineffective initial regimen was replaced, or if simplification for a more rational (lower spectrum) alternative to a prior effective treatment, the bronchoscopy was classified as relevant.

Microbiologic Management of Samples

In patients in whom ventilator-associated pneumonia was suspected, bronchoscopy was performed as previously described (6). After the protected specimen brush was transected into a sterile vial containing 1 ml of sterile lactated Ringer's solution, the vial was vigorously agitated for at least 60 s to suspend all the material from the brush. Specimens were immediately sent to the laboratory for quantitative cultures. Aliquots of 0.01 ml were taken from the original suspension and inoculated into blood agar, MacConkey agar, buffered charcoal yeast extract agar, and Sabouraud medium. One 0.001-ml aliquot was also inoculated into chocolate agar medium. Culture plates were incubated at 37° C under adequate aerobic and anerobic conditions; all plates except Sabouraud plates were evaluated for growth at 24 and 48 h. For the protected specimen brush, bacterial counts of 10³ CFU/ml or greater were used as the cutoff point to diagnose pneumonia. Two serial 10-fold dilutions were then done on the recovered bronchoalveolar lavage fluid, and 0.01-ml aliquots of the original suspension and each dilution were placed onto plates in the same way as for the protected specimen brush sample. All protected specimen brush and bronchoalveolar lavage isolates were identified by standard laboratory techniques (11).

Empiric Treatment of Patients

Immediately after the diagnostic procedure, an empirical antibiotic treatment was begun or was modified (if the patient was receiving a prior antibiotic therapy for a prior infection). The choice of antibiotics was left to the discretion of each attending physician. In patients with persistent fever > 39.0° C after 4 d of therapy, who developed worsening hypoxemia, or who required progressive increase of inotropic drugs ("clinical deterioration"), initial antibiotics were modified at the discretion of the attending physician. The standard antibiotic treatment in our institution included an antipseudomonal betalactam plus amikacin in all patients with suspected P. aeruginosa. Antibiotic therapy was changed based on culture and susceptibility studies when they were available (usually 48 to 72 h after starting empiric therapy). Because of the risk of false negative results, particularly in patients receiving prior antibiotic therapy, negative results were not taken into account in order to modify the empirical antibiotic therapy.

Statistical Analysis

Descriptive analysis was performed. Means were compared using Student's t test and the Mann-Whitney test. Proportions were compared using the chi-square test with Yates correction or Fisher's exact test when necessary. Confidence intervals (CI) for proportions were obtained assuming the binomial distribution. Fisher's exact test for unpaired samples and McNemar's test for paired ones were used to determine the statistical significance of differences. All p values and CI are two-sided. All interval estimates are 95% CIs.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

One hundred thirteen ventilated patients (77 male and 36 female) judged to have ventilator-associated pneumonia were prospectively followed during the study period. Their median age was 62 yr. One additional patient was excluded because he died within the first 24 h of diagnosis; 82 other patients were excluded because the diagnosis of pneumonia could not be confirmed. Underlying diseases of the study population are summarized in Table 1. The yield of microbiologic investigations is shown in Table 2. Pleural fluid culture was also positive in two episodes of pneumonia caused by P. aeruginosa. No major complications related to the bronchoscopic procedure were documented. The mean ± SD length of time of ventilation before pneumonia was suspected was 8.5 ± 8.2 d. Indeed, most episodes (69.3%) developed later than 5 d postintubation.

A causative pathogen was identified in 100 of 113 patients. A detailed list of the pathogens involved is shown in Table 3. All positive blood cultures were obtained in patients with positive protected specimen brush (PSB) samples. Isolates were: Pseudomonas aeruginosa (n = 5), Streptococcus viridans group (n = 1), Hemophilus influenzae (n = 1), Morganella morganii (n = 1), and polymicrobial (n = 1). Seventy patients were receiving prior antibiotic therapy for other infectious conditions (Table 4) and the causative pathogen was not identified in four of them. The empirical antibiotic choices used to treat VAPs are also listed in Table 4. Empiric monotherapy led to a directly related mortality rate of 21.8% in 53 patients, whereas combination therapy (all regimens in agreement with ATS guidelines [12]) in 58 patients did not prove (p > 0.20) any more successful (mortality rate = 24.1%). Antibiotics were changed (Table 5) in 51 of 100 (51.0%) episodes in which an etiology was identified and in four of 13 (30.8%) cases in which an etiology was not found (p = NS). These four were changed because of clinical deterioration compared with three in the other group (p < 0.05). Crude mortality was 54.0 and 92.3% (p < 0.05) for patients with episodes with and without etiologic diagnosis, respectively.

In 43 patients (38.0%) the antibiotics were changed because of information obtained from microbial tests (Table 5). In 27 (23.6%), therapy was replaced because the antimicrobial agents prescribed were ineffective against the microorganisms involved (Table 6), and this group showed a significantly greater increase in related mortality than did the adequate initial therapy group (37.0 versus 15.6%, p < 0.05). Crude mortality for both groups of patients was 63.0 and 41.5% (p = 0.06), respectively. The excess mortality caused by inappropriate initial therapy was estimated to be 21.4% (95% CI, 43.2 to 0.03). However, this change permitted clinical resolution in 17 (62.9%) of these 27, and 10 patients were discharged alive. In nine others (7.8%), combination therapy was started on the identification of P. aeruginosa isolates. In addition, bronchoscopic results permitted us to select a narrower and more rational therapy in seven (6.1%). No significant differences in frequency of changes were found in patients ventilated longer than 5 d (57.1 versus 41.7%, p = NS).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our study suggests that in patients with VAP, microbial investigations frequently lead to changes in antibiotic therapy. However, all positive blood cultures were obtained in patients with positive PSB samples. Consequently, bronchoscopic procedures were determinant in these changes. In addition, antibiotic treatment was modified because of inadequate clinical response in seven patients (6.1%) and to poor tolerance in two (1.8%). Simplified therapy was performed in seven others (6.1%). Simplifying therapy is of interest in economic terms, obviously, but also because administering unnecessary antibiotics may lead to superinfection with more resistant strains and also to the emergence of multiresistant pathogens in the hospital (6). These observations contrast with the findings from the study of Woodhead and colleagues (5) in which investigation of hospitalized patients with community-acquired pneumonia was unhelpful. However, the spectrum of pathogens responsible for community-acquired pneumonia, particularly in the ICU setting (13), varies with respect to nosocomial episodes, and different therapeutic principles must be applied (12, 14, 15). Indeed, our findings are similar to recent observations reported by Rodriguez de Castro and colleagues (16) in a population of ICU patients with severe pneumonia from different origins (community, hospital wards, and intubated patients).

Inadequate initial antibiotic therapy was identified by bronchoscopy and modified in around 25% of patients. An extra effective antibiotic was added because of isolation of P. aeruginosa in 7.8% of patients. In addition, patients in whom pneumonia was diagnosed but who had a negative etiologic diagnosis had a statistically significant higher crude mortality. We should clarify that this subgroup did not represent the whole group of patients with negative etiologic diagnoses and suspicion of pneumonia because most of them were classified as "uncertain" and excluded from analysis. This increase of mortality was associated with a greater percentage of changes in therapeutic regimens on the basis of clinical deterioration, and this points out that, at least in the study population, absence of diagnosis represents a risk factor for death.

Moreover, the outstanding finding was the observation of a statistically significant increase in related mortality as a result of inappropriate early antibiotic therapy despite a microbially guided change. This leads to a rise in crude mortality (caused by inappropriate initial therapy) of greater than 20%. Even with this problem, 62.9% showed clinical resolution, and 10 of these 27 patients were ultimately discharged alive from the ICU. In fact, a preliminary study by Luna and colleagues (7) suggested that bronchoalveolar lavage results were unable to modify the final outcome in a population of 109 intubated patients with pneumonia. These findings highlight that delay in the administration of effective therapy for intubated patients with pneumonia is associated with increased mortality and the need to develop guidelines to improve the initial antibiotic regimen in patients with nosocomial pneumonia.

Although all patients with pneumonia had fever, leukocytosis, or purulent tracheal secretions and a pulmonary infiltrate, the values of those criteria for the diagnosis of infection in ICU patients is somewhat doubtful. In our study, most episodes should be classified as "probable" pneumonia according to the American College of Chest Physicians guidelines (4), but this is the case of the most studies on VAP. In our opinion restricting the assessment to "definite" episodes alone represents an unacceptable bias. Indeed, critical to the validity of the study is to ensure that no patients without pneumonia were incorrectly included. As a result, a large number of episodes were classified as uncertain and were excluded from our analysis. This means that the excess of mortality caused by inappropriate initial antibiotic therapy might be different if all patients with clinical suspicion of pneumonia were included. Finally, it should be noted that sensitivity of microbiologic investigation has been overestimated compared with other studies (12) since no new antibiotics were prescribed after the onset of infection and before obtaining specimens for culture (9). This study was not specifically designed to investigate the yield (sensitivity or specificity) of bronchoscopic procedures in patients with clinical suspicion of pneumonia. As a result, the excellent sensitivity for PSB in this study should not be extrapolated to the whole population with suspicion of pneumonia.

Our study had several limitations. Most episodes in our series were "late" pneumonias, and P. aeruginosa was the most frequent pathogen. The effect of initial antibiotics on other pathogens may not be as critical, and the impact on outcome may be different in other institutions. Indeed, recurrent pulmonary infections were excluded from our study and this may underestimate the excess of mortality since multiresistant pathogens are responsible for most of these episodes. In the current study, the initial therapeutic strategy was not standardized (in fact, the recommendations for patients with suspected P. aeruginosa pneumonia in our institution depended on the class of antibiotics they had previously received) and the regimen was selected by the attending physician; this may represent a potential bias since "expert" opinion may be controversial. In addition, the excess of mortality caused by nosocomial infections in patients with other underlying diseases or different levels of severity may be different. Therefore, the generalizability of our results is unknown. Finally, these results are valid when bronchoscopic techniques are routinely performed in the etiologic investigation of pneumonia in intubated patients, but other findings may be obtained if less invasive techniques such as quantitative tracheal aspirates are performed, or if microbial investigation is restricted to patients with inadequate clinical response.

Despite these limitations, this study demonstrated that, at least in our institution, routine microbial investigation (specifically, bronchoscopic techniques) determines changes in antibiotic therapy in a high proportion of patients. Most importantly, our findings emphasize the critical importance of appropriate early antibiotic therapy. In our opinion, both observations should be taken into account in the design of future therapeutic guidelines on pneumonia in intubated patients.