American Journal of Respiratory and Critical Care Medicine

Except for bronchoalveolar lavage, the value of the Gram stain examination of respiratory tract samples for the diagnosis of hospital-acquired pneumonia (HAP) and their potential impact on empiric antibiotic treatment have rarely been assessed. During a 14-mo period, both plugged telescoping catheter (PTC) and endotracheal aspirate (EA) were performed when an HAP was suspected in mechanically ventilated patients. The results of Gram stain examinations and cultures and previous and subsequent antibiotic treatment were prospectively recorded. Two criteria for pneumonia were considered: (1) clinically diagnosed pneumonia, according to attending physicians, based on clinical and radiological evolution and the PTC culture results, and (2) microbiologically proven pneumonia (solely based on the result of quantitative PTC culture). Of 91 episodes of suspected HAP in 51 patients, 27 (30%) episodes were clinically diagnosed as pneumonia. When considering clinically diagnosed pneumonia, the sensitivity and specificity of Gram stain examination were similar (respectively, 89% and 62% for EA and 67% and 95% for PTC). When considering microbiologically proven pneumonia, the sensitivity and specificity of Gram stain examination were, respectively, 91% and 64% for EA and 70% and 96% for PTC. The negative predictive value of Gram stain examination of EA and the positive predictive value of Gram stain examination of PTC were high. Our results suggest that the combination of Gram stain examination of paired PTC and EA may contribute to the early diagnosis of HAP in about two-thirds of mechanically ventilated patients, and guide the empiric therapy when needed. In the remaining one-third of patients, the Gram stain examination is not helpful in predicting HAP.

The diagnosis of hospital-acquired pneumonia (HAP) is often based on a combination of clinical and radiographic criteria, and on the results of quantitative cultures of respiratory tract secretions samples. For this purpose, the plugged telescoping catheter (PTC) is considered as sensitive and specific (1), whereas the culture of nonquantitative endotracheal aspirate (EA) may lack specificity (2). Lower respiratory tract samplings with quantitative cultures have been shown to both guide empiric therapy and reduce antibiotic use (3). On the other hand, the prognostic value of an early appropriate treatment of HAP has been recently emphasized (4-6). Indeed, inappropriate or delayed antibiotic therapy of HAP may be associated with poorer outcome of patients affected.

However, the potential value of Gram stain examination of respiratory tract samples has been rarely assessed. The percentage of cells containing intracellular bacteria in bronchoalveolar lavage has been proposed to diagnose ventilator-associated pneumonia early (7). In the first study describing the PTC technique, Gram stain examination of specimens showed poor sensitivity and high specificity to detect culture-positive samples (1). The suboptimal sensitivity of Gram stain examination of PTC samples (1) may cause problems when early empiric antibiotic treatment is needed. Conversely, the sensitivity of nonquantitative EA cultures appears to be high, whereas its specificity is poor (2). In one study of ventilator-associated pneumonia, a negative Gram stain of EA material suggested that pneumonia was not a cause for lung infiltrates and fever (8). Therefore, both techniques, EA and PTC, could be complementary.

The aim of this study was to assess the value of the combination of Gram stain examination of both EA and PTC for early diagnosis of HAP and identification of likely etiologic microorganisms. The results of Gram stain examination of paired EA and PTC could be helpful in the decision-making process of initial empiric therapy in a large percentage of patients suspected of having pneumonia.

Patients

Our hospital is a 410-bed comprehensive cancer center, with a 15-bed medical-surgical intensive care unit (ICU).

Between September 1, 1996 and October 31, 1997, 51 consecutive patients hospitalized for more than 48 h in our ICU and receiving mechanical ventilation were prospectively studied when nosocomial pneumonia was suspected. Respiratory secretions samples were obtained upon request of the attending physician when a first episode of HAP or a superinfection after a first HAP episode was clinically suspected. The clinical suspicion of HAP was based on recent and/or persistent pulmonary infiltrates on the chest radiograph, and at least two of the following clinical criteria: fever (⩾ 38.5° C), leukocytosis (⩾ 104/ mm3), purulent tracheal secretions, and a decrease in the PaO2 /Fi O2 ratio of more than 20%. In these cases, a PTC and an EA were successively obtained. Respiratory tract secretions samplings were performed within 24 h of clinical suspicion of HAP.

The chart of each patient studied was reviewed for the following parameters: demographic data, age and sex, tumoral diagnosis, presence of neutropenia (white blood cells count < 1,000/mm3), reason for admission to the ICU, severity of acute illness on admission to the ICU, assessed by the simplified acute physiologic score (SAPS) II and the number of organ system failures (OSFs) (9, 10), date of intubation, overall duration of MV, and outcome at ICU discharge.

Respiratory secretions samples were obtained because of clinical suspicion of a first episode of HAP (n = 51) or clinical suspicion of superinfection after a first HAP episode (at least 5 d apart; n = 40). Ninety-one samples (EA + PTC) were thus included in this study.

Prior antibiotic treatment (within the last 72 h) was recorded, including whether a new antibiotic treatment class was started within this period. The subsequent treatment was recorded, according to the time of its introduction (after Gram stain examination of PTC and EA, or after culture results). The decision to introduce (or modify) antibiotic therapy after sampling, based on the severity of the septic syndrome and results of Gram staining specimens, was left to the attending physician.

The study was approved by the local Ethics Committee, and informed consent was waived.

Samplings Technique

The following procedures were performed for each patient. An endotracheal aspirate was obtained for Gram staining and culture. A blinded or per-fiberoptic PTC was then performed after careful endotracheal suctioning, as described (1). Briefly, the outer sheath of the PTC unit (Combicath 5829.20; Plastimed Lab, Saint-Leu la Forêt, France) was wedged in the bronchial tree and then retracted 4–5 cm. The inner catheter was extruded 2–3 cm beyond the tip of the outer sheath, dislodging the plug. Three brief aspirations were applied to the inner catheter with a 10-ml syringe, and the catheter was then retracted and removed. Subsequently, 1 ml of saline water (calibrated sample) was flushed through the inner catheter and collected in a sterile vial, and the distal segment of the catheter (4 cm) was aseptically transected and collected in the same vial. When fibroscopy was used, the bronchoscope (Olympus P20 D, Tokyo, Japan) was introduced into the bronchus where samples were to be obtained, as determined by chest radiographic findings and the visual detection of purulent secretions, the PTC catheter was advanced in the suctioning channel of the bronchoscope, and the sampling was performed as described above.

Both specimens were immediately (within 15 min) transferred to the microbiology laboratory for Gram staining and culture, as previously described (1). One milliliter of endotracheal aspirate sample was diluted in 1 ml saline, homogeneized by vortexing; Gram stain examination was performed using 100 μl of fluid, while another 100 μl of fluid was inoculated on blood agar. The quality of endotracheal aspirate specimens was checked by microscopic examination, and none of the samplings showed more than 10 epithelial cells per 10× low-power field. For direct microscopic analysis of PTC samples, one slide was prepared by cytocentrifugation using a Shandon Cytospin 3 (Pittsburg, PA), and Gram stain examination was performed using 100 μl of fluid. The results of cultures of EA and PTC were obtained within 24 to 48 h. PTC cultures were expressed in colony-forming units per milliliter of the original 1 ml sample (cfu/ml).

The microbiologists reading Gram stain slides were blind to the aim of the study during the study period, in order to reproduce the conditions of routine practice.

Definitions and Analysis

A positive culture of the PTC was defined as the recovery of ⩾ 103 cfu/ ml of a potential pathogen (1). Staphylococcus epidermidis or Candida spp. were considered as potential pathogens in immunocompromised patients. Specific microorganisms such as Nocardia sp. were considered as pathogens whatever the bacterial count. Pneumonia due to Aspergillus sp. or other yeasts were not considered in the present study.

Two types of analysis, based on different criteria, were performed. (1) The classification into clinically diagnosed pneumonia was established by the attending physician. (2) The diagnosis of microbiologically proven pneumonia was solely based on the result of quantitative PTC culture, defined by a microbial colony count ⩾ 103 cfu/ml. The clinical classification of the episodes was established by the attending physician. The latter was blind to the study, so as to avoid bias in the interpretation of results and in the therapeutic approach. The diagnosis was based on the results of the quantitative culture of the PTC, and on radiological and clinical evolution. Presence of CT-scan cavitation and culture of blood and/or pleural fluid yielding the same microorganism as found in the bronchial samplings were major arguments for the diagnosis of HAP. In some cases, when an antibiotic treatment active on the microorganisms recovered was introduced within the last 72 h, the diagnosis of HAP was retained despite bacterial counts lower (102 cfu/ml) than the threshold of positivity of the PTC, if the clinical evolution under appropriate treatment was considered compatible by the attending physician.

Populations Studied

Three populations were analyzed: (1) All episodes were included in the pragmatic (interpretative) part of the study based on the clinical diagnosis made by physicians. (2) Only episodes without recent (within the last 72 h) change in antibiotic therapy, which is known to profoundly impact the results of quantitative cultures (11, 12), were analyzed in the population with microbiologically proven pneumonia. (3) Finally, a separate analysis was performed in the subset of patients receiving mechanical ventilation for more than 48 h (defining ventilator-associated pneumonia [4, 6]), using microbiologically confirmed pneumonia in patients without recent antibiotic change as the reference.

Quantitative data are given as mean ± standard deviation (SD). The sensitivity, specificity, and positive (PPV) and negative (NPV) predictive values of Gram stain examination of both EA and PTC were calculated for each group analyzed.

The concordance between Gram stain examination and cultures of both EA and PTC was examined, as were the modifications of antibiotic treatment after Gram stain examination and cultures.

Patients

The 51 patients (15 women and 36 men) included in the study had a mean age (± SD) of 53 (± 13) yr. Their mean SAPS II score on admission was 45 (± 19) and their mean number of OSFs was 1.5 (± 1.3). Fifteen patients had an hematological malignancy and 36 had a solid tumor. The causes of admission were acute respiratory failure (n = 30), severe sepsis or septic shock (n = 8), renal failure or metabolic disorders (n = 2), neurological failure (n = 4), and postoperative monitoring (n = 7). The overall duration of mechanical ventilation was 18 (± 19) d (range: 1–109). Thirty-five patients (69%) died in the ICU.

Samplings

Ninety-one paired samplings (EA + PTC), performed after a mean of 11 (± 15) d of mechanical ventilation (range: 0–98), were analyzed. Sixty-two (58%) samplings were obtained after 48 h of mechanical ventilation. Two-thirds (62 of 91) of PTC samples were obtained blindly and the other one-third was directed under fibroscopy. In 12 episodes of suspected pneumonia, a neutropenia was present at the time of sampling.

Fifty-nine samplings were performed while patients were receiving prior antibiotic treatment, including 21 who had a recent (during the last 3 d) change in antibiotic therapy. Prior antibiotics prescribed were mainly penicillins (n = 4), a combination of a penicillin with a β-lactamase inhibitor (n = 18), third-generation cephalosporins (n = 13), imipenem (n = 14), aminoglycosides (n = 23), glycopeptides (n = 19), cotrimoxazole (n = 7), fluoroquinolones (n = 10), rifampin (n = 3), or antifungals (n = 13).

Clinically Diagnosed Pneumonia

When considering all 91 episodes of suspected pneumonia in the 51 patients, 27 episodes were classified as clinically diagnosed HAP (30%). In 21 of these 27 cases (78%), at least one bacteria was recovered in the PTC sample culture in concentration ⩾ 103 cfu/ml. In 5 episodes with recent (< 72 h) change of antibiotics, the diagnosis of HAP was retained despite a nonsignificant count (102 cfu/ml) of the PTC culture. In addition, one case of pneumonia due to Nocardia sp. (10 cfu/ml) was diagnosed.

In 3 of the 91 episodes, the PTC sample culture yielded significant growth (⩾ 103 cfu/ml), but the diagnosis of pneumonia was not retained by the attending physician, and patients were considered as colonized.

Microbiologically Proven Pneumonia

Seventy episodes without recent change of antibiotic therapy were included in this analysis; a positive PTC culture was found in 23 (33%) episodes, therefore classified as microbiologically proven pneumonia.

Results of Gram Stain Examination of EA

In 48 of all 91 episodes studied, the Gram stain examination of EA was positive; only 24 (50%) of these episodes were clinically diagnosed as pneumonia. Conversely, a clinical diagnosis of HAP was excluded in 40 of 43 episodes with negative Gram stain examinations of the EA (three false negative) (Figure 1). Therefore, Gram stain examination of EA had high sensitivity (89%) and lower specificity (62%) for clinically diagnosed pneumonia.

When considering only the 70 patients without recent changes in antibiotics, Gram stain examination of EA had high sensitivity (91%) and negative predictive value (94%) and lower specificity (64%) and positive predictive value (55%) for the diagnosis of microbiologically proven pneumonia.

Results of Gram Stain Examination of PTC

The Gram stain examination of 21 PTC was positive. Polymorphonuclears were present on direct examination of all cases with subsequent positive culture of the PTC (including in neutropenic patient). Of these 21 episodes, 18 were classified as clinically diagnosed HAP (three false positive). Gram stain examination of PTC had high specificity (95%) but lower sensitivity (67%) for the criterion of clinically diagnosed pneumonia (Figure 1).

When considering only the 70 patients without recent changes in antibiotics, Gram stain examination of PTC had high specificity (96%) and positive predictive value (89%) and slightly lower sensitivity (70%) and negative predictive value (86%) for the diagnosis of microbiologically proven pneumonia.

The results were not modified when neutropenic patients were analyzed separately.

Ventilator-associated Pneumonia

The accuracy of Gram stain examination of both PTC and EA in the subset of 62 patients having received mechanical ventilation for more than 48 h was similar to that in the whole population. The sensitivity, specificity, and negative and positive predictive values of Gram stain examination of EA for clinically diagnosed ventilator-associated pneumonia (VAP) were 95%, 61%, 96%, and 56%, respectively. In the subset of 53 patients without recent changes of antibiotics, the operating values were essentially similar using microbiologically proven pneumonia as the reference (Figure 2).

Microbiological Results

The microorganisms recovered in episodes of HAP are shown in Tables 1 and 2. Seven of nine methicillin-resistant Staphylococcus aureus were recorded after 7 d or more of mechanical ventilation.

The microbiological concordance between Gram staining and cultures was high: among the 21 positive Gram staining of the PTC, 18 totally (n = 16) or partially (only one bacterium seen on Gram stain examination, but two microorganisms recovered from culture; n = 2) concordant cases were recorded, whereas in the 3 other episodes, the culture was negative (n = 1) or showed a microorganism different from the one suspected after Gram staining (n = 2).

In six of the nine clinically diagnosed pneumonia with negative Gram stain examination of the PTC, the microorganism responsible for HAP was found on Gram staining of EA. Overall, among the 27 cases with clinically diagnosed pneumonia, the pathogen responsible would have been correctly identified in 23 instances (85%), by either the Gram stain examination of PTC (n = 17) or EA (n = 6).

Antibiotic Therapy after Samplings

An empiric antibiotic therapy was prescribed within 6 h following the results of Gram stain examination of both EA and PTC in 25 of the 91 episodes of clinically suspected pneumonia.

In 10 episodes, a treatment was introduced on the basis of the positive Gram stain examination of the PTC. After culture results were obtained, the diagnosis of HAP was confirmed in these 10 patients, and the treatment was totally (n = 6) or partially (n = 4) maintained.

The treatment was prescribed on the basis on positive Gram stain examination of EA in only four episodes; only one of these treatments was a posteriori justified, when considering the final clinical diagnosis of HAP.

Finally, an empiric therapy was introduced in 11 episodes despite negative Gram stain examinations of both samples; except for one episode, this change in therapy was not justified when considering the final clinical diagnosis of pneumonia.

A targeted therapy was prescribed after the results of cultures in five cases, within 24 to 48 h following samplings. In four additional cases with clinically diagnosed pneumonia, a prior or recently introduced therapy was considered as appropriate and was not modified.

Three false-positive results of the Gram staining examination of the PTC (when considering the criterion of clinically diagnosed pneumonia) were recorded. One patient was considered as colonized (PTC culture ⩾ 103 cfu/ml without clinically diagnosed pneumonia) and received no treatment. None of the two other false-positive results of the Gram stain examination of the PTC was ensued by an unjustified treatment: in one case, prior antibiotic treatment was maintained because of severe neutropenia; in another patient, a severe bronchitis (without HAP) was diagnosed and the treatment introduced after the result of positive Gram stain examination was continued.

Potential Impact of Direct Examination on Diagnosis and Antibiotic Therapy

If the results of gram stain examination of the two samples were taken together, 58 satisfactory clinical predictions would have been performed (92%) among the 63 cases having positive Gram stain examination of PTC (n = 20) and/or negative Gram stain examination of EA (n = 43). In the remaining 28 cases (31%), a negative Gram stain examination of the PTC together with a positive Gram stain examination of the EA would not allow us to accurately predict the presence of HAP before the culture results; HAP was present in only 6 (21%) of these 28 episodes (Figure 1). These 28 cases were not different from others in terms of date of samplings, prior antibiotics, or recent changes of antibiotic therapy.

We then examined the potential consequences of treatment based only on the results of Gram stain examination of EA or of PTC taken separately in the whole population, including the 27 episodes of clinically diagnosed pneumonia. If all 48 cases with a positive Gram stain examination of EA had been treated empirically, 24 (50%) of these therapeutic decisions would have been unjustified. Conversely, if only the 21 cases with a positive Gram stain examination of PTC had been treated, appropriate treatment would have been delayed in 9 of 27 (33%) episodes of pneumonia.

The role of quantitative “invasive” diagnostic techniques in the evaluation of patients with clinical suspicion of hospital-acquired pneumonia remains controversial (13, 14). However, invasive diagnostic techniques may improve the clinical management and reduce antibiotic use (3), and possibly improve the prognosis of mechanically ventilated patients with a clinical suspicion of HAP (15). Diagnostic testing in this situation mainly serves two purposes: (1) to determine the likelihood of pneumonia as an explanation of new signs and symptoms and (2) to identify the etiologic pathogen when pneumonia is present (2).

The direct examination of lower respiratory tract samplings could be of critical importance for the early diagnosis of HAP and identification of the etiologic microorganism, and for subsequent guidance to appropriate therapy. Several recent studies have emphasized the prognostic importance of an early appropriate therapy: Rello and coworkers found a significant increase in mortality as a result of inappropriate early antibiotic therapy (37 versus 15%) despite guided changes in therapy after the results of cultures (4). The authors highlighted the need for guidelines to improve the initial antibiotic regimen in patients with nosocomial pneumonia. Luna and coworkers found that the mortality rate was reduced when adequate antibiotic therapy was initiated very early, compared with when this therapy was inadequate or no therapy was given (5). Similarly, Kollef and Ward showed that a delay in initiating adequate antibiotic therapy was associated with a higher mortality (6). Conversely, unnecessary empiric treatment in a patient without pneumonia could mask the positivity of samplings from another focus of infection, thus exposing the patient to a potential delay in diagnosis and appropriate treatment.

Despite this evidence, the interest of Gram stain examination of protected respiratory tract samplings for patients suspected of having HAP has rarely been studied, except for bronchoalveolar lavage. Using rapid examination of the bronchoalveolar lavage, the percentage of cells containing intracellular bacteria has been proposed to diagnose ventilator-associated pneumonia early (7). To our knowledge, only two studies have been published focusing on the diagnostic value of direct examination of protected specimen brush samples. The presence of microorganisms on Gram stain examination of protected specimen brush samples has been shown to be highly sensitive and specific of the diagnosis of HAP (16). Marquette and coworkers reported high sensitivity (85%) and specificity (94%) for the Gram stain examination of protected specimen brush (17). However, Pham and coworkers found a poor sensitivity (20%) of Gram stain of the protected specimen brush to detect culture-positive samples, although the specificity was high (95%) (1).

Few data are available regarding the value of direct examination of endotracheal aspirates. Fagon and coworkers used the Gram staining of EA obtained by suction through the bronchoscope to evaluate the clinical judgment in the identification and appropriate treatment of HAP in ventilated patients (18); the final diagnosis of HAP was based on the results of culture of protected specimen brush. The authors concluded that the clinical judgment, based in part on the results of Gram-stained EA, did not permit accurate prediction of the presence of HAP and the best antibiotic regimen. Although the potential usefulness of Gram-stained EA was not emphasized in this study, the Gram stain examination of EA showed a good sensitivity (24 of 27) but lacked specificity (34 false positives out of 57) for the diagnosis of HAP. In another study of ventilator-associated pneumonia, Gram stain examination of EA was also considered a sensitive tool (8). Our results are in accordance with these findings, by showing high negative predictive value and sensitivity, but poor specificity and positive predictive value of EA aspirates. Overall, these results strongly suggest that when the Gram stain examination of EA is negative, the diagnosis of HAP is very unlikely.

The plugged telescoping catheter is an accurate tool for the diagnosis of ventilator-associated pneumonia (1), including cancer patients (19). However, the potential value of the Gram stain examination of the plugged telescoping catheter has not been emphasized. In the first study by Pham and coworkers (1), the sensitivity and specificity of Gram stain examination of PTC samples to predict culture-positive samples were comparable to those found in the present study, showing high specificity and lower sensitivity. These findings suggest that the positivity of Gram stain examination of the PTC is highly suggestive of HAP. The quality of the PTC sampling is, however, of critical importance: the presence of polymorphonuclears in almost all patients with positive PTC is a marker of an adequate sample.

In our study, when the Gram stain of the PTC was positive, the microbiological concordance with the results of cultures was high. Sixteen of the 21 episodes with positive Gram staining of the PTC showed strict concordance with the culture. In the two cases with only partial concordance, the two microorganisms recovered by the culture of PTC were present on the Gram staining of EA, allowing us to guide the choice of antibiotic therapy. However, in two other cases, the false prediction due to the Gram stain examination of the PTC would have been responsible for inappropriate treatment (although this was not the case in the two episodes reported here). When an empiric therapy is needed, the choice of antibiotics may be helped by epidemiological considerations (20), and in some patients by knowledge of prior bronchial colonization, which often precedes infection (21, 22). The results of Gram stain examination of EA could complement the results of Gram stain examination of the PTC. Overall, the Gram stain examination of PTC and/or EA would have correctly identified the pathogen responsible (and possibly allowed adequate selection of antibiotics) in 85% of the 27 clinically diagnosed HAPs.

One limitation of our study is that the 28 cases (31%) with a negative Gram stain examination of the PTC together with a positive Gram stain examination of the EA were not accurately classified (as HAP or not) before the culture results. Although HAP was present in only 6 (21%) of these 28 episodes, these 6 cases with proven pneumonia could have suffered from antibiotic therapy being withheld. Conversely, antibiotic therapy would have been unnecessary in the 22 other cases without pneumonia. Two approaches to therapy are possible in such cases. A liberal attitude would consider treating all patients with such dissociated results empirically: the choice of antibiotics may be guided by the Gram stain examination of EA, surveillance cultures of tracheal aspirates, and local or consensus guidelines (2). Alternatively, a more restrictive approach, considering the potential consequences of unnecessary and inadequate empiric treatment, based on the severity of sepsis, underlying disease, or worsening respiratory condition, may be adopted.

We conducted two types of analyses. The first analysis used the diagnosis of HAP as established by the attending physician, a criterion that reflects clinical practice, but may be debatable. The second analysis took as the only and as a strict criterion the bacterial count of the culture of PTC (using the threshold of 103 cfu/ml), and excluded episodes with recent antibiotic changes, which alter the performances of quantitative cultures (11). Whichever examined population was chosen, the sensitivity and negative predictive values (NPV) of Gram stain examination of EA were high, with poor specificity and positive predictive values (PPV); conversely, the specificity and PPV of Gram stain examination of PTC were high, with lower sensitivity.

We studied only mechanically ventilated patients. However, some of the episodes of pneumonia included in our series were acquired before the onset of mechanical ventilation. When the subset of the patients ventilated after 48 h or more was specifically studied, the operating characteristics of the results of Gram stain examination of respiratory secretions samples were also essentially similar.

The study was performed in a selected population of cancer patients. The value of the PTC has been recently validated in this subset of patients (19). In 12 instances, a neutropenia was present at the time of sampling. The sensitivity and specificity of direct examination of both EA and PTC were similar in the subgroup of neutropenic patients and in the whole population.

PTC is a validated method for the diagnosis of HAP (1, 19), and both EA and PTC are safe and easy to perform. Other techniques, such as protected specimen brush or bronchoalveolar lavage could offer the same results, allowing Gram staining and microscopic identification of intracellular organisms for early diagnosis of nosocomial pneumonia (7). The respective advantages and drawbacks of each of these techniques are not within the scope of the present study. However, it should be mentioned that the latter two techniques usually need bronchoscopy, which is not always available everywhere. Considering the findings by Luna and coworkers, which showed a reduced mortality rate when adequate antibiotic therapy was initiated before performing BAL (5), the feasibility of both EA and PTC at all times allows us to initiate antibiotic therapy without delay in all circumstances, when HAP is suspected.

Although no guidelines for initiating an empiric treatment was given in our unit at the study period, it is noteworthy that most of the patients with a positive Gram stain examination of the PTC were treated early, and that all of these therapies were a posteriori justified. Conversely, almost all treatments initiated on the basis of positive Gram stain examination of the EA only or despite negative results of both sampling proved unnecessary. These findings confirm the need for guidelines to improve therapeutic decisions.

Prior use of antibiotics, present in 59 episodes, could alter the accuracy of pulmonary samplings. A recent change in antibiotic therapy, during the previous 3 d, occurred in 21 instances. Only recent changes alter the performances of quantitative cultures (11), whereas previous antibiotic therapy prescribed to treat an earlier septic episode unrelated to suspected pneumonia does not affect the diagnostic yield of respiratory samplings (23). For this reason, when HAP is suspected, it is important to perform the respiratory samplings before any change in antibiotic therapy.

Our study suggests that the Gram stain examination of paired PTC and EA may help early clinical decision making in two-thirds of mechanically ventilated patients with suspected HAP. In our series, an accurate prediction could be made in 64% of all cases with suspected pneumonia, or more than 90% of episodes presenting with a negative Gram stain examination of the EA or a positive Gram stain examination of the PTC. In the remaining one-third of episodes, a negative Gram stain examination of the PTC with a positive Gram stain examination of the EA does not allow to accurately predict the presence of HAP before the culture results; however, it should be noted that only a minority (21%) of these episodes was secondarily diagnosed as HAP in our patients.

In light of these results, we propose a decisional tree for the early diagnosis and management of suspected HAP.

1. The Gram stain examination of the EA is negative. HAP is very unlikely. No empiric antibiotic treatment for pneumonia is needed pending culture results of samples.

2. The Gram stain examination of the PTC is positive. HAP is very likely, and early empiric treatment should be administered, especially in patients with deteriorating respiratory status or worsening sepsis. The choice of antibiotics can be based on the results of the Gram staining of the PTC and/ or of the EA, as well as on epidemiological data (microbial ecology of the unit, previous known colonization of the patient, duration of mechanical ventilation, and prior antibiotic use, especially broad spectrum drugs [20]). When culture results are obtained, the antibiotic treatment may be maintained, adapted, or stopped (3).

3. The Gram stain examination of PTC is negative and the Gram stain examination of the EA is positive. No satisfactory prediction is allowed before the cultures results. In such cases, the decision to start an empiric treatment could depend on the severity of sepsis and underlying condition. When administered, empiric antibiotic therapy may be stopped after 24–48 h if the quantitative culture is reported as negative.

Using this strategy, only two episodes would have been unduly treated (compared with 24 episodes if therapeutic decisions were based on the results of EA only), and only three patients would have received delayed therapy (compared with nine instances if therapeutic decisions were based on the results of PTC only). Our strategy based on Gram stain examination does not allow us to classify all episodes as either associated or not associated with HAP: 28 episodes remained unclassified after direct examination. The main value of this diagnostic approach is to reduce the percentage of uncertainty to about one-third of the episodes.

The approach we suggest may allow, in a majority of suspected pneumonia episodes, the administration of earlier and more appropriate antibiotic therapy when needed, while avoiding, in part, unnecessary treatments. This approach may lead to improved outcome of patients, a lower risk of emergence of antibiotic resistance, and more appropriate use of resources. However, before extensive adoption, this strategy should be evaluated in a prospective trial, examining antibiotic use and patients' outcomes.

This work is dedicated to Cyrille Tancrède, recently deceased, for his constant, and enthusiastic encouragements and advice. The authors thank the nurses of the intensive care unit of the Institute Gustave-Roussy for their dedicated care and contribution to the conduct of this study: I. Rousseau, N. Torri-Piqueras, M. Courchinoux, C. Bec, D. Mathee, S. Vochelet, and M. Minet (principal nurse); and Mrs A. Morin for skillful secretarial assistance.

1. Pham LH, Brun-Buisson C, Legrand P, Rauss A, Verra F, Brochard L, Lemaire FDiagnosis of nosocomial pneumonia in mechanically ventilated patients: comparison of a plugged telescoping catheter with the protected specimen brush. Am Rev Respir Dis143199110551061
2. American Thoracic Society. Hospital-acquired pneumonia in adults: diagnosis, assessment of severity, initial antimicrobial therapy, and preventive strategies: a consensus statement. Am J Respir Crit Care Med 1995;153:1711–1725.
3. Bonten MJM, Bergmans DCJJ, Stobberingh EE, van der Geest S, De Leeuw PW, van Tiel FH, Gaillard CAImplementation of bronchoscopic techniques in the diagnosis of ventilator-associated pneumonia to reduce antibiotic use. Am J Respir Crit Care Med156199718201824
4. Rello J, Gallego M, Mariscal D, Soñora R, Valles JThe value of routine microbial investigation in ventilator-associated pneumonia. Am J Respir Crit Care Med1561997196200
5. Luna CM, Vujacich P, Niederman MS, Vay C, Gherardi C, Matera J, Jolly ECImpact of BAL data on the therapy and outcome of ventilator-associated pneumonia. Chest1111997676685
6. Kollef MH, Ward SThe influence of mini-BAL cultures on patient outcomes: implications for the antibiotic management of ventilator-associated pneumonia. Chest1131998412420
7. Chastre J, Fagon JY, Soler P, Domart Y, Pierre J, Dombret MC, Gibert C, Hance AJQuantification of BAL cells containing intracellular bacteria rapidly identifies ventilated patients with nosocomial pneumonia. Chest951989190S192S
8. Salata RS, Lederman MM, Shlaes DM, Jacobs MR, Eckstein E, Tweardy D, Toossi Z, Chmielewski R, Marino J, King CH, Graham RC, Ellner JJDiagnosis of nosocomial pneumonia in intubated, intensive care patients. Am Rev Respir Dis351987426432
9. Le Gall JR, Lemeshow S, Saulnier FA new simplified acute physiology score (SAPS II) based on a European/North American multicenter study. JAMA270199329572963
10. Knaus WA, Draper EA, Wagner DP, Zimmerman JEPrognosis in acute organ-system failure. Ann Surg111985685693
11. Souweine B, Veber B, Bedos JP, Gachot B, Dombret MC, Regnier B, Wolff MDiagnostic accuracy of protected specimen brush and bronchoalveolar lavage in nosocomial pneumonia: impact of previous antimicrobial treatments. Crit Care Med261998236244
12. Montravers P, Fagon JY, Chastre J, Lecso M, Dombret MC, Trouillet JL, Gibert CFollow-up protected specimen brushes to assess treatment in nosocomial pneumonia. Am Rev Respir Dis14719933844
13. Chastre J, Fagon JYInvasive diagnostic testing should routinely be used to manage ventilated patients with suspected pneumonia. Am J Respir Crit Care Med1501994570574
14. Niederman MS, Torres A, Summer WInvasive diagnostic testing is not needed routinely to manage suspected ventilator-associated pneumonia. Am J Respir Crit Care Med1501994565569
15. Fagon JY, Chastre J, Wolff M, Gervais C, Parer-Aubas S, Stéphan F, Similowski T, Mercat A, Diehl JL, Sollet JP, Tenaillon Afor the VAP Trial Group. Invasive and non invasive strategies for management of suspected ventilator-associated pneumonia: a randomized trial. Ann Intern Med1322000621630
16. Mertens AH, Nagler JM, Galdermans DI, Slabbynck HR, Weise BS, Coolen DDiagnostic value of direct examination of protected specimen brush samples in nosocomial pneumonia. Eur J Clin Microbiol Infect Dis151996807810
17. Marquette CH, Wallet F, Nevière R, Copin MC, Saulnier F, Drault JN, Mehdaoui H, Mathieu D, Ramon PDiagnostic value of direct examination of the protected specimen brush in ventilator-associated pneumonia. Eur Respir J71994105113
18. Fagon JY, Chastre J, Hance AJ, Domart Y, Trouillet JL, Gibert CEvaluation of clinical judgment in the identification and treatment of nosocomial pneumonia in ventilated patients. Chest1031993547553
19. Casetta M, Blot F, Antoun S, Leclercq B, Tancrède C, Doyon F, Nitenberg GDiagnosis of nosocomial pneumonia in cancer patients undergoing mechanical ventilation: a prospective comparison of the plugged telescoping catheter with the protected specimen brush. Chest115199916411645
20. Trouillet JL, Chastre J, Vuagnat A, Joly-Guillou ML, Combaux D, Dombret MC, Gibert CVentilator-associated pneumonia caused by potentially drug-resistant bacteria. Am J Respir Crit Care Med1571998531539
21. Johanson WG, Pierce AK, Sanford JP, Thomas DGNosocomial respiratory tract infection with gram-negative bacilli: the significance of colonization of the respiratory tract. Ann Intern Med771972701706
22. Delclaux C, Roupie E, Blot F, Brochard L, Lemaire F, Brun-Buisson CLower respiratory tract colonization and infection during severe acute respiratory distress syndrome: incidence and diagnosis. Am J Respir Crit Care Med156199710921098
23. Timsit JF, Misset B, Renaud B, Goldstein FW, Carlet JEffect of previous antimicrobial therapy on the accuracy of the main procedures used to diagnose nosocomial pneumonia in patients who are using ventilation. Chest108199510361040
Correspondence and requests for reprints should be addressed to François Blot, Service de Réanimation, 39 rue Camille Desmoulins, Institut Gustave Roussy, 94805 Villejuif, France. E-mail:

Related

No related items
American Journal of Respiratory and Critical Care Medicine
162
5

Click to see any corrections or updates and to confirm this is the authentic version of record