Bronchoscopy in Ventilator-associated Pneumonia: Agreement of Calibrated Loop and Serial Dilution

The reported incidence of ventilator-associated pneumonia (VAP) is between 8 and 28%, with an associated mortality rate between 24 and 76% (1). Microbiological evaluation is necessary for the diagnosis of VAP and to define pathogen-specific antimicrobial treatment. Qualitative culture of endotracheal aspirates from endotracheally intubated patients is associated with a high percentage of false-positive results (2).

On the basis of findings from a randomized multicenter clinical trial (3), national and international consensus conferences have recommended that quantitative culture of bronchoalveolar (BAL) fluid or protected specimen brush be part of the standard of care in patients with suspected VAP (2, 4-7). The serial dilution technique is considered to be the gold standard for quantitative culture. Bacterial growth of 10^sup 4^ cfu/ml or greater in BAL fluid using the serial dilution technique is needed to support a diagnosis of VAP (8). However, this technique is cumbersome and labor intensive. To circumvent this problem, some laboratories use semiquantitative cultures obtained by the calibrated loop technique. The average processing time for serial dilution is approximately 28 min as compared with 8 min for the calibrated loop procedure (9). There is a paucity of data, however, that address the agreement between the calibrated loop and serial dilution techniques. In a study limited to seven bronchoscopic protected specimen brush samples, Middleton and colleagues (9) showed an acceptable agreement between the two techniques. The purpose of this prospective work is to determine the agreement between the calibrated loop and serial dilution techniques in measuring quantitative bacterial growth in the BAL fluid of patients with suspected VAP. To our knowledge, there are no previous publications evaluating the agreement between these two techniques in the diagnosis of VAP using BAL fluid. Some of the results of this study were reported in the form of an abstract (10).

METHODS

See the online supplement for the detailed version of METHODS, including the bronchoscopy procedure and microbiological assessment.

Adults who underwent BAL for suspected VAP were eligible for the study and prospectively enrolled. Patients younger than 18 yr and those who did not authorize their medical records to be reviewed were excluded. The decision to perform BAL was made by the primary care providers. We collected demographic, intensive care unit- and mechanical ventilator-related, antibiotic, severity, and outcome data.

The 1992 International Consensus Conference guidelines for the standardization of bronchoscopy for the diagnosis of VAP were followed (8). Gram stain was performed on the BAL fluid. Five milliliters of the fluid were vortexed for 30 to 60 s but not centrifuged. Quantitative cultures were performed by both serial dilution and calibrated loop methods as described by Baselski and colleagues (11). Specific colonies were counted for each dilution and mean colony counts were determined from the media inoculated for each dilution. Growth was reported as colony-forming units per milliliter. Nonpathogens, such as viridans group streptococci, coagulase-negative staphylococci, Neisseria spp., and Lactobacillus spp., were reported as "usual flora" without quantitation. Because fungi rarely cause nosocomial pneumonia in immunocompetent patients and Candida and Aspergillus species often colonize the airways of patients receiving invasive mechanical ventilation, these organisms were not taken into account.

Data were summarized as mean (SD), median (interquartile range), and proportions. We used a ?^sup 2^ test for categoric variables for comparison between groups. We used the dilution with the highest mean number of colony-forming units per milliliter measured by each of the two techniques in our analysis. We applied tests of bias and precision to determine the agreement between the calibrated loop and serial dilution techniques, as described by Altman (12) and Bland (13). We performed tests of bias and precision after logarithmic (log10) transformation of the data. We added 1 to all quantitative values before logarithmic transformation to eliminate the error arising from logarithmic transformation of 0. We also calculated the sensitivity, specificity, negative predictive value, and positive predictive value, with their 95% confidence interval (95% CI), of the calibrated loop technique in the diagnosis of VAP using the serial dilution technique as the gold standard. Organisms that were too numerous to count were designated as more than 4 log10 cfu/ml. Subgroup analyses were performed for patients with and without antibiotics at the time of BAL. A quantitative threshold of 4 log10 (10,000) cfu/ml was used for the diagnosis of VAP.

The hypothesis to be tested was that the mean difference within pairs of quantitative culture values obtained by the calibrated loop and serial dilution techniques is 0. On the basis of the study by Middleton and coworkers, we calculated a sample size of 18 patients, who were not receiving antibiotics at the time of BAL, to provide a power of 80% (two-tailed) at the significance level of 0.05 (9). Because our pilot study had shown that almost 80% of the patients with suspected VAP were receiving antibiotics at the time of BAL, we determined that we would need 90 patients.

RESULTS

The study included 104 patients who had 121 BAL samples obtained from July 30, 2003, to October 23, 2005. Over 90% of the patients were white, with a mean age of 61.5 yr (Table 1). At the time of BAL, the patients had been in the intensive care unit and had been receiving mechanical ventilation for a median of 8 d and had been hospitalized for a median of 10.5 d (Table 1). The most common reasons for initiation of mechanical ventilation were pneumonia (26 patients), acute respiratory distress syndrome (21 patients), and sepsis (11 patients). (See Table E1 of the online supplement for the complete indications for mechanical ventilation.) The hospital mortality rate was 47.1% (49 of 104 patients).

Patients were receiving antibiotics when 90 of the 121 (74.4%) BAL samples were obtained for the following suspected infections: community-acquired pneumonia (14 patients), previous episode of VAP (10 patients), hospital-acquired pneumonia (9 patients), health care-associated pneumonia (9 patients), intraabdominal infection (10 patients), urinary tract infection (6 patients), vascular catheter-associated infection (4 patients), others (12 patients), and undetermined nosocomial source (16 patients). Fifty-seven of the 121 BAL samples (47.1%) yielded one or more organisms: in 42 of the 90 samples (46.7%) in patients with antibiotics compared with 15 of the 31 samples (48.4%) in patients without antibiotics (p = 0.8685). Antibiotic changes had been made within the preceding 3 d when 73 of the 121 BALs (60.3%) were performed. Thirty-two of the 73 BALs in patients with antibiotic changes within 3 d (44%) yielded one or more organisms compared with 25 of the 48 without changes (52%; p = 0.3739). Thirty-seven BAL samples yielded only usual flora and 27 had no bacterial growth. Ninety organisms, 63 of them bacterial, were isolated from 57 BAL samples. The most frequently isolated organisms were Staphylococcus aureus (18 isolates; methicillin-resistant, 8), Pseudomonas aeruginosa (17 isolates), and Stenotrophomonas maltophilia (10 isolates). (See Table E2 for the complete list.) Most of the potentially antibiotic-resistant organisms were isolated in patients who were receiving antibiotics: 11 of the 18 S. aureus (5 of the 8 isolates that were methicillin resistant), 12 of the 17 P. aeruginosa and 8 of the 10 S. maltophilia were isolated in patients who were receiving antibiotics at the time of or within 3 d before the BAL. The colony counts of 13 bacterial organisms were too numerous to count by the calibrated loop technique, and the counts ranged from 4.70 to 6.74 log10 cfu/ml by the serial dilution technique. The calibration loop technique identified 18 patients with VAP by 23 bacterial organisms. The serial dilution technique identified 17 patients with VAP (all of whom were also identified by the calibration loop) by 22 bacterial organisms.

Paired quantitative culture values measured by each of the two techniques were available for 50 bacterial organisms: 35 from patients with antibiotics and 15 from patients without antibiotics at the time of BAL. The mean (SD) colony counts of these 50 bacterial organisms were 2.855 (1.551) log10 cfu/ml by the calibrated loop method compared with 2.475 (1.657) log10 cfu/ml by the serial dilution method (p = 0.0100). The biases, precisions, and 95% limits of agreement are listed in Table 2. The biases show statistically significant differences in the mean colony counts of bacterial organisms isolated in patients who were receiving antibiotics at the time of BAL. The Bland-Altman graph for the 50 bacterial organisms is shown in Figure 1. Three values, two from patients without antibiotics and one from a patient receiving antibiotic at the time of BAL, were outside the 95% limits of agreement. Using 4 log10 as the cut-off point, there was only one discordance between the two techniques (Figure 2); this discordance was for P. aeruginosa, which was isolated at 4.02 log10 cfu/ml by the calibrated loop technique and 3.48 log10 cfu/ml by the serial dilution technique.

The sensitivity (95% CI), specificity (95% CI), positive predictive value (95% CI), and negative predictive value (95% CI) of the calibrated loop technique for the diagnosis of VAP were 100% (85-100%), 98% (87-100%), 96% (79-99%), and 100% (91-100%), respectively.

The times spent to inoculate the plates were available for nine BAL samples. It took an average (SD) of 2.8 (2.2) min for the calibration loop compared with 11.0 (3.7) min for the serial dilution (p = 0.0003).

DISCUSSION

In this prospective observational study, we determined the difference between the calibrated loop and serial dilution techniques in measuring quantitative bacterial growth in BAL fluid of patients with suspected VAP. The majority of patients were receiving antibiotics at the time of bronchoscopy. There was overall good agreement between the calibrated loop and serial dilution techniques in the diagnosis of VAP in all but one patient. Colony counts that were too numerous to count by the calibrated loop technique correlated with counts of 4.70 log10 cfu/ml or higher by the serial dilution technique.

Our study shows that the calibrated loop technique can be used instead of the serial dilution technique to measure the quantitative growth of microorganisms in BAL fluid of patients with suspected VAP. Using 4 log10 cfu/ml as the cut-off point, there was only one discrepancy. When the concentrations of microorganisms in BAL fluids were very high, their exact counts could not be measured by the calibrated loop method. In this situation, the serial dilution technique showed counts higher than 4 log10 cfu/ml in the current study, suggesting that the use of the calibrated loop technique will not compromise the care of patients with suspected VAP.

Recent or current antibiotic use may influence the number of intracellular bacteria within cells recovered by BAL (14). However, antibiotics being administered to treat previous infections unrelated to and at the time of the suspected VAP do not appear to affect the diagnostic yield of BAL (15). In the current study, there was agreement between the calibrated loop and serial dilution techniques in both patients with and without antibiotics at the time of BAL.

Our study has limitations. It was performed in a single medical center and the majority of patients were receiving antibiotics at the time of BAL. Most of the patients had underlying acute pulmonary diseases, including pneumonia and acute respiratory distress syndrome. Our findings may not apply to other patient populations because of these weaknesses. The reporting of some low pathogenic bacteria as "usual flora" without specification and quantitation ignores the fact that these organisms may cause infection in some patients (4, 6). Despite these limitations, our work demonstrates that there is clinically important agreement between the serial dilution and calibrated loop techniques for the diagnosis of VAP using BAL fluid. The two techniques can be used interchangeably for the diagnosis of VAP by quantitative culture of BAL fluid in patients receiving invasive mechanical ventilation. Because the calibrated loop technique is less time-consuming, it can be used in the diagnosis of VAP instead of the serial dilution technique.

Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript.