Protected Bronchoalveolar Lavage 1- 4 A New Bronchoscopic Technique to Retrieve Uncontaminated Distal Airway Secretions

G. UMBERTO MEDURI, DAVID H. BEALS, AMADO G. MAIJUB, and VICKIE BASELSKI

Introduction SUMMARY We tested the effectiveness of protected bronchoalveolar lavage (PBAL), performed

Bronchoalveolar lavage(BAL) of a lung through a protected transbronchoscopic balloon-tipped (PST) catheter, in collecting distal airway subsegment, obtained through a fiberopsecretions with a minimal degree of contamination. The PBAL had" 1% squamous epithelial cells tic bronchoscope wedged in an airway, in 91% of specimens and an absence of bacterial growth in 59% of patients without pneumonia. samples a large area of the alveolar surUsing a threshold of 104 cfu/ml we had one false positive result in 33 patients without pneumonia face and is a sensitive tool in diagnosing and one false negative In 13 patients with pneumonia. Quantitative bacterial cultures of the PBAL specimens had a diagnostic sensitivity of 97% and a specificity of 92%, with a positive predictive lower respiratory tract infections (table value of 97% and a negative predictive value of 92%. The diagnostic efficiency was 96%. The pres1). To reach the bronchial tree, however, ence of intracellular organisms In ~ 2% of the recovered alveolar cells (Glemsa stain) was seen the bronchoscope must traverse the in all but two patients with pneumonia (on corticosteroids) and in none of the patients without pneuoropharynx or the endotracheal tube monia. Gram stains of the PBAL specimens were positive In all but one patient with pneumonia where resident bacteria are likely to be and negative in all but one patient without infection (patient with endobronchial narrowing secondintroduced into the suction channel of ary to neoplasm with false positive cultures). Either the Giemsa or the Gram stain was positive the instrument. Oropharyngeal and train all patients with pneumonia, allowing early and accurate diagnosis of lower respiratory tract incheobronchial contaminants (frequentfection before the results of cultures were available. The time off antibiotic therapy before bronchosly present in high concentration) are copy did not affect the result of PBAL cultures, contrary to what we observed for the protected found in 89070 of BAL specimens from AM REV RESPIR DIS 1991; 143:855-864 brush specimen. patients without infection (table 2). For this reason bronchoalveolar lavage has had limited use in diagnosing bacterial pneumonia. A technique that could efMethods pneumonia. Twopatients, one in each group, fectively decrease contamination of the weretransplant recipients on immunosuppresPatient Selection respiratory tract secretions retrieved with sive therapy. BAL should, in theory, improve the spec- Control group. The control group consisted of 18 patients, none of whom had antibiotic Premedication and Anesthesia ificity of this procedure while maintain- therapy for at least 7 days before bronchosGroup C 1 received 120mg nebulized lidocaine ing a high degree of sensitivity. copy. All were inpatients at the Memphis VA A protected transbronchoscopic bal- Medical Center and were admitted an aver- via intermittent positive-pressure breathing loon-tipped (PHI) catheter was developed age of 6 days before the procedure. All un- (IPPB) followed by 3 to 4 ml of 2OJolidocaine for BAL to avoid exposing the instilled derwent bronchoscopy for reasons other than injected through the bronchoscope and perand aspirated BAL solution to the con- pneumonia: IS patients had lung cancer, 2 colated on the vocal cords to achieve complete anesthesia before entering the trachea. taminants present in the suction lumen had hemoptysis, and 1 had sarcoidosis. None had fever, leukocytosis, purulent sputum, or of the bronchoscope. Our prospective clinical study was designed to evaluate a pneumonic infiltrate on chest radiograph. (Received in original form May 29, 1990 and in this technique and had five objectives: These 18 control patients were divided into revised form October 15, 1990) three subgroups (C.. C2, and C3)of 6 patients (1) to determine to what extent protected each, the difference between them being the bronchoalveolar lavage (PBAL) using a route of administration of topical anesthesia From the Departments of Medicine, PatholoPBT catheter can effectively collect low- and the sequence of sampling of PBAL and gy, and Clinical Microbiology, University of TennesseeHealth SciencesCenter, and the Department er respiratory tract secretions with a min- protected specimen brushing (PSB). of Veterans Affairs Medical Center, Memphis, imal degree of contamination; (2) to esStudy group. The study population con- Tennessee. tablish the value of PBAL specimens in sisted of three groups (S.. S2' and S3) of paPresented in part at the National Meeting of diagnosing bacterial pneumonia; (3) to tients suspected of having pneumonia. All the American College of Chest Physicians, Bosassess the effect of topical anesthesia ad- were off antibiotic therapy for at least 48 h. ton, Massachusetts, October 1989,and at the World ministration and the sequence of sam- Group S1 (n = 25) patients were intubated Conference on Lung Health, Boston, Massachusetts, May 1990. pling procedures on culture results; (4) and mechanically ventilated and had clinical 3 Supported in part by the Richard P. Ettinger, signs suggestive of pneumonia. All had fever to assess the effect of previous antibiotic (~ 38.30 C), a new radiographic density, and Sr., Fellowship in Cancer Research and by a grant therapy on the results of cultures; and either a macroscopically purulent tracheal from Merck, Sharp, and Dohme Company. 4 Correspondence and requests for reprints (5)to determine whether quantitative cul- aspirate or leukocytosis (> 10,000cells/mm"). be addressed to G. Umberto Meduri, M.D., tures of PBAL specimens at one dilution Groups S2and S3consisted of three nonven- should Division of Pulmonary and Critical Care, Univercan accurately predict the presence of tilated patients each with suspected hospital- sity of TennesseeHealth SciencesCenter, 956Court acquired (S2) or community-acquired (S3) Avenue, Room H314, Memphis, TN 38163. pneumonia. 1

2

855

856

MEDURI, BEALS, MAIJUB, AND BASELSKI

TABLE 1 RESULTS OF QUANTITATIVE CULTURES IN PATIENTS WITH BACTERIAL PNEUMONIA*

Reference

BAL Amount

Significant Growth

No. of Patients

Type

(m~

(cfu/m~

13

immuno

150

105 (13/13) 104 (15/15) 104 (8/8) 105 (3/5) 103 (17/25) 58/66 88

BAL Kahn and Jones (20) Thorpe et a/. (21)

240

15

Chastre et a/. (22) Torres t et a/. (6)

8

immuno

240

5

vent

100

25

vent

150

66 100

Total

% PBAL

9

vent

150

4

immuno CAP

150

< 10

3

~

103

~

104

~

105

19

2

17

17

2

15

8

8

15

4

5

2

4

52

22

12

14

4

111 100

26 23

17

15

28 25

40 36

11

104 (9/9) 104 (3/4) 12/13 92

13 100

Total

%

Colony-forming Units/ml

No. of Isolates

11

5 16 100

1 6

2

2

2 13

13 81

Definition of abbreviations: immuno = immunosuppressed; vent = mechanically ventilated patient; CAP = community-acquired pneumonia. • Those organisms that grew at or above the selected diagnostic threshold are indicated by boldface type. t This investigator used quantitative bacterial culture at one dilution only.

TABLE 2 RESULTS OF QUANTITATIVE CULTURES IN PATIENTS WITHOUT BACTERIAL PNEUMONIA *

Reference BAL Kahn and Jones (20) Chastre et a/. (22) Kirkpatrick and Bass (26) Total % PBAL Total %

No. of Patients

Type

18 44 13 8 83 100

Co I St i St v Co n

18 15 33 100

Col St v, I

Colony Forming Units/ml

No Growth

No. of Isolates

< 103

2 5 1 1 9 11

26 55 22 16 119 100

9 10 15 14 48 40

6 12 3 2 23 19

9

14 8 22 100

10 4 14 64

3 4 7 31

9

18 55

~

103

~

104

~

105

5 19 2

6 14 2

26 22

22 19

0

It

0 0

1 5t

Definition of abbreviations: Co = control group; St = study group without pneumonia; i = immunusuppressed patients; I patients with lung disease other than pneumonia; n = normal volunteers; v = ventilated patients without pneumonia. • Those organisms that grew at or above the selected diagnostic threshold are indicated by boldface type. t Control patient with significant endobronchial narrowing due to primary carcinoma.

Groups C2 and C3 received 400 mg nebulized lidocaine via IPPB and did not have injection of additional lidocaine through the bronchoscope before entering the trachea. Groups C 1 and C2 had PSB done before PBAL, and Group C3 had PBAL done before PSB. Nonintubated patients received0.6mg atropine sulfate and 60 mg codeine intramuscularlyapproximately 15min before bronchoscopy. Intravenously administered midazolam was used intraoperatively as needed to achieve additional sedation. Intubated patients received midazolam and morphine sulfate intravenously as preoperative medication before receiving a paralytic agent (vecuronium bromide).

=

Protected Transbronchoscopic Balloon-Tipped Catheter The radiopaque PBT catheter is made of polyurethane, has an outer diameter of 2.3 mm, and is designed to be introduced through a large-channel (2.6 mm) fiberoptic bronchoscope. It has two lumina: (1) a large, openended lumen (1 mm) that runs the length of the catheter for irrigation and for aspiration of liquid and gases, and (2) a small lumen that communicates with a recessed,high-compliance, low-pressure latexballoon at the distal end of the catheter (figure 1). The balloon is 12 mm long, reaches an outer diameter of 10 to 12 mm when inflated with 1.5 to 2 ml

of air, and allows optimal occlusion at the levelof the third-generation bronchi. To facilitate the use of this device a black stripe was placed around the catheter at the proximal end ofthe balloon, at 72 cm from the tip (the length of the fiberoptic bronchoscope channel) and at 2-cm intervals (x 3) thereafter. A thin polyethylene glycol diaphragm was applied at the tip of the catheter to prevent contaminants from entering the system. The diaphragm was easily expelled by flushing 2 ml sterilesaline through the irrigation lumen with a 3-ml syringe. It then dissolved at body temperature.

Specimen Collection and Processing Monitoring. During bronchoscopy each patient had continuous finger pulse oximetry (Novametrix Medical SystemCo., Wallingford, CT) and electrocardiographic monitoring. Bronchoscopic equipment. A portable bronchoscopic tray created for the procedure included a large-channel fiberoptic bronchoscope (IT-20D; Olympus, New Hyde Park, NY), a PBT catheter (Clinical Instruments Corp., Dudley, MA), a sterile microbiology specimen brush catheter (Microvasive, Watertown, MA), 70070 isopropyl alcohol, two sterile scissors, one sterile wire clipper, one sterilebronchoscope adaptor, two 5-mlLuer Lok® syringes (Becton and Dickenson Co., Cockeysville, MD), one 5OO-ml bottle with sterile nonbacteriostatic saline, five30-ml Luer Lok® syringes, one Falcon tube with 1 ml sterile nonbacteriostatic saline, and two Falcon tubes for collecting PBAL effluent. Bronchoscopic technique. The fiberoptic bronchoscope was inserted via a nostril in nonintubated patients and through an endotracheal tube via a sterile connector in ven-

PROTECTED BRONCHOALVEOLAR LAVAGE

857

of sterile saline, gently aspirating after each instillation. After discarding the first aliquot the subsequent aliquots were equally divided for bacteriologic and cytologic analysis, and 2 ml was placed into a transport medium (Porr-ACul'" vial; Becton and Dickenson Co.) for anaerobic bacterial cultures. The specimens were delivered immediately to the microbiology laboratory for analysis.

Fig. 2. The tip of the flexible fiberoptic bronchoscope is positioned next to the Iingular orifice. The PBT catheter is introduced into the superior subsegment of the lingula with the black stripe, marking the proximal portion of the balloon, visible at the orifice. The balloon is then inflated and wedged. After ejecting the distal diaphragm PBAL is performed through an uncontaminated catheter.

Fig. 1. The transbronchoscopic balloon-tipped catheter and the flexible fiberoptic bronchoscope.

tilated patients. The oral route was used in a fewpatients who could not tolerate the bronchoscope nasally. Avoiding suction and the injection of lidocaine through the bronchoscopic channel (with the exception of control patients in group C1), the bronchoscope tip was then positioned adjacent to the orifice of the bronchial subsegment to be sampled.

Sampling Procedure In Groups C 1 and C2 PSB was performed before PBAL. In Group C 3 p,BAL was performed before PBS. In patients without lung radiographic densities the right middle lobe or the lingula was chosen as the sampling area. No suction was applied until the end of sampling (PSB and PBAL), and the bronchial aspirate obtained at that point was submitted for Gram stain and culture. In patients with lung radiographic densities the sampling area selected was based on the location of the new or progressive density (chest radiograph) or the segment (visualized during bronchoscopy) having purulent secretions in patients with more than one new radiographic density. Except as noted earlier, the sequence of sampling was alwaysPSB followed by PBAL. These procedures were always performed in the same bronchial subsegment with the exception of patients in con-

Laboratory Analysis of Respiratory Secretions Bacterial analysis. The PBAL effluent was vortexed for 30 s and inoculated onto agar plates: blood, chocolate, and MacConkey agars for aerobic cultures and Schaedler blood and selective Schaedler agars for anaerobic cultures. The PBAL effluent also wascultured quantitatively by serial loo-fold dilutions (l0-1, 10-3 , and 10-5 ) of the original specimen. Chocolate agar was used for aerobic quantitative cultures, and Schaedler blood agar was used for anaerobic quantitative cultures. The PSB was vortexed in the 1 ml saline and processed in an identical fashion. Two0.5-ml samples of PBAL fluid wereused for cytospin preparations for Gram and Giemsa stains (figures 5 and 6). Gram stain of the PSB sample was prepared by placing one drop of the saline suspension onto a slide.The PBALfluid was also cultured on media for isolation of fungi and mycobacteria. Giemsa stain methodology. PBAL fluid, 0.5 ml, was used to prepare cytospin slides, which were stained with a modified MayGninwald-Giemsa stain obtained commercially (Baxter Scientific Products Division, Stone Mountain, GA). Air-dried cytospin slides were immersed in May-Grunwald working solution for 15 min in a Coplin jar and rinsed in tap water. The slides were then immersed in a solution of Giemsa working stain for 15 min and rinsed in tap water. After air drying the slides, 300 cells per slide werecounted. The cell differential and percentage of cellswith intracytoplasmic organisms were reported. Open lung biopsy. Open lung biopsieswere obtained directlyfrom the operating room and placed in a sterile Petri dish. Tissue (5 g) was weighed on a Mettler balance, and a tissue homogenate was prepared using a Corning Glassware glass homogenizer (Baxter Scientific Products Division). Microbiologic and mycologicstudies wereperformed in the same manner as described for PBAL specimens. Tissue sections wereprepared and stained with hematoxylin and eosin and evaluated for the presence or absence of a fibrinopurulent exudate within alveolar spaces (figure 7). When exudate was present Giemsa and Gram stains wereprepared and evaluated for the presence or absence of intracytoplasmic organisms within phagocytic cells.

trol Group C3 , who had PBAL followed by PSB in the adjacent subsegment of the same bronchus. Protected specimen brushing. The PSB catheter was advanced 3 em out of the bronchoscope to avoid collectingpooled secretions on the catheter's tip, and the polyethylene glycol plug was ejected into the large airway by protruding the inner cannula. The catheter was then advanced into the desired subsegment and the brush exposed to reach a proximal (nonventilated patients) or peripheral (ventilated patients) position or to enter a visible pool of purulent secretions. The brush was gently rotated and pulled back and forth several times to maximize exposure to secretions. After sampling, the brush was retracted into the inner cannula, the inner cannula retracted into the outer cannula, and the catheter was then removed. The distal portions of the outer and inner cannulas were separately and sequentially wiped clean with 70% isopropyl alcohol, cut with sterile scissors, and discarded. The brush was then advanced, inspected for the presence of secretions, severed with a sterile wire clipper, and placed into a sterile vial with 1 ml sterile saline to avoid drying and rapid loss of bacteria. Protected bronchoalveolar lavage. The PBT catheter was introduced into the suction channel of the bronchoscope and advanced into the desired subsegment. The distal black stripe marking the proximal portion of the balloon was then positioned at the subsegmental orifice (figure 2). After proper placement the balloon was inflated with 1.5 to 2 ml air to occlude the subsegmental orifice. The catheter was gently retracted to verify a tight seal. The distal diaphragm was then exData Analysis pelled by flushing 2 ml sterile saline (with a 3-ml syringe) through the irrigation lumen. Before statistically analyzing the data, culPBAL was performed with five30-mlaliquots ture results were divided into positive and

858

MEDURI, BEALS, MAIJUB, AND BASELSKI

Histologic 1. Foci of consolidation with intense polymorphonuclear leukocyte accumulation in the bronchioles and alveolar spaces Clinical 2. Pleural fluid: organisms on gram stain or culture identical to that recovered in significant concentration in the PSB or PBAl culture 3. Blood culture: positive, with organism identical to that recovered in significant concentration in the PSB or PBAl culture 4. Radiographic: rapid cavitation in the absence of lung carcinoma 5-7 Significant growth on (5) PSB (~ 103 cfu/ml) or (6) PBAl (~ 1()4 cfu/ml), or (7) both PSB and PBAl with "appropriate" response to specific narrow spectrum antibiotic therapy (radiographic clearance and resolution of fever)

Fig. 5. Giemsa stain from protected bronchoalveolarlavage in a patient with pneumonia. Note the presence of intracytoplasmic organismswithin a polymorphonuclear leukocyte (oil immersion, x 1,000).

Fig. 3. Criteria to define the presence of pneumonia.

negative and the patients were divided by the presence or absence of pneumonia. Patients with an undetermined diagnosis were so reported but were not considered for statistical analysis. The diagnostic threshold defining a quantitative bacterial culture as positive, or significant, was a growth of a single bacterial species (not a summation or index) in PBAL ~ 104 cfu/ml and on PSB ~ loa cfu/ml. Growth below these points was defined as insignificant. The criteria used to define the presence or absence of pneumonia are shown in figures 3 and 4, respectively. Pneumonia was definitively excluded for patients without a significant growth on PBAL or PSB cultures when their fever and radiographic density resolved without instituting antibiotic therapy or when a definitive alternative diagnosis was established. A diagnostic workup for every ventilated patient with negative bronchoscopic findings (for example, no infection or malignancy noted) was performed to identify the alternative cause of the fever and radiographic infiltrate.

Cytospin of the PBAL effluent was subjected to Gram and Giemsa stains. The Gram stain was defined as negative if no microorganisms were seen and positive if microorganisms were identified and correlated with what grew in significant concentration in the PBAL culture (figure 6). The Giemsa stain was used to define the alveolar cell popula-

tion differential and the presence of intracellular organisms (figure 5). Squamous epithelial cellsand neutrophils with intracellular organisms were reported as a percentage of the total cellsrecovered. The time without antibiotic therapy before bronchoscopy and the type of treatment receivedafter the procedure were recorded for every patient studied. Postmor-

Histologic lack of consolidation with intense polymorpho1. nuclear leukocyte accumulation in the bronchioles and alveolar spaces 2. Criterion 1 plus a definitive alternative etiology Cytologic 3. Identification of a process other than pneumonia (e.g., malignancy) by PBAl without significant bacterial growth on both PSB and PBAl Clinical: lack of significant growth on quantitative cultures of the PSB « 103 cfu/ml) or PBAl « 104 cfu/ml), with one of the following: Resolution, without antibiotic therapy, of one of the following: 4. Fever 5. Radiographic infiltrate 6. Fever and radiographic infiltrate 7. Radiographic infiltrate and a definitive alternative diagnosis, or 8. Persistent fever or radiographic infiltrate with a definitive alternative diagnosis Fig. 4. Criteria to define the absence of pneumonia.

Fig. 6. Gram stain from the same patient in figure 5 demonstrating intracytoplasmic gram-positive cocci (oil immersion, x1,OOO).

PROTECTED BRONCHOALVEOLAR LAVAGE

Fig. 7. Hematoxylin and eosin (H&E) staining of an open lung biopsy from same patient as in figures 5 and 6. Note the large amount of fibrinopurulent exudate within the alveolar spaces (x40).

859

was diagnosed clinically, three were confirmed by open lung biopsy, and one was confirmed by autopsy. Histologically the proliferative phase of diffuse alveolar damage was noted with no evidence of infection. Tissue cultures had no growth. (3) Two patients had lung cancer: one was documented cytologically and one histologically. (4) One patient had lymphangitic metastasis of breast carcinoma. (5) One patient had line sepsis with congestive heart failure. (6) Bronchiolitis obliterans with organizing pneumonia was diagnosed in one patient via open lung biopsy with histologic and bacteriologic absence of infection.

Results of Bacterial Cultures

tern data were analyzed only if autopsy was performed within 1 wk of bronchoscopy. Statistical analysis consisted of estimating sensitivity, specificity, positive and negative predictive values, and efficiency for these two techniques. Diagnostic results were compared using the chi-square and McNamara's tests. For patients with pneumonia the odds of a patient having a true positive result werecomputed using logistic regression.

Results

A total of 51 patients met the criteria for the study, and 49 completed the protocol. Two patients, both receiving mechanical ventilation, had no fluid retrieved with protected BAL. In both of them the bronchoscope was wedged after removing the PBr catheter, and return of BAL effluent was obtained in one. The average percentage of PBAL fluid retrieved was 36% in controls, 42070 in the nonventilated study patients, and 22070 in the ventilated study patients. This discrepancy may be related, in part, to the reduction in caliber of the irrigation lumen of the PBT catheter that resulted from transmission of high intrathoracic pressures around the inflated balloon in ventilated patients. This portion of the irrigation lumen was designed to accommodate a recessed balloon and was structurally weaker and more compliant. The percentage of

The results of quantitative bacterial cultures of the PBAL and PSB are shown in table 3. In the control group absence of growth in the PBAL specimen was seen in 33070 of patients who received lidocaine injection over the vocal cords (Group C t ) compared with 58070 in the group of patients who had no injection through the bronchoscope (Groups C 2 and C 3 ) . In the control group bronchial aspirates obtained after PSB and PBAL sampling had bacterial isolates in all but one specimen, with moderate or large growth in six. A false positive PBAL culreturned fluid, however, did not affect ture was from a patient with marked narthe diagnostic yield of PBAL. A definirowing of the left upper lobe orifice tive diagnosis was established in 46 pa(PBAL performed in the right middle tients; 13 had pneumonia. lobe bronchus) due to a primary lung carcinoma that had significant polymicrobiPatients with Absence al growth in both PBAL and PSB culof Pneumonia tures. Insignificant growth was seen in In 33 patients there was no pneumonia: 14 PBAL specimens with a total of 19 18 controls, 14 mechanically ventilated bacterial isolates: 15 isolates grew < 1()3 patients, and 1 with suspected commu- cfu/ml and 4 grew ~ 4 x 1()3 cfu/ml. nity-acquired pneumonia. In the 15study In 80070 of the PBAL specimens with bacpatients pneumonia was excluded by lung terial growth isolates were present in onhistology in 8 (5 open lung biopsies and ly one of the three dilutions (10-1, 10-3 , 3 autopsies) and by cytology of the PBAL and 10-5 ) of the aerobic culture plates. effluent in 1 patient. Three patients had The sequence of sampling affected the clinical resolution without antibiotic result of cultures.' The rate of false positherapy, and 3 had a definitive alterna- tive results for the PSB specimens was tive diagnosis established. Only 3 of the 17070 (2 of 12) when it was performed be15 patients without pneumonia received fore PBAL and 50% (3 of 6) when it was antibiotic treatment, all for reasons oth- performed after. Sterile cultures of the er than pneumonia. Two had line sepsis PSB specimens were seen in 58070 when and 1 had meningitis. performed before PBAL and in 50070 Several alternative diagnoses were not- when done after it. Sterile cultures of the ed in study patients without pneumonia. PBAL effluent were seen in 50% of both (1) The diagnosis was atelectasis in five groups (C. + C 2 and C 3 ) . patients who had resolution of the radioResults of Gram and Giemsa Stains logic density within 24 h postprocedure without antibiotic therapy. Most of them Gram stains of the PBAL effluents were received nebulized bronchodilator ther- negative for all but one patient who had apy followed by chest percussion. (2) Five an endobronchial lesion and a significant patients had the diagnosis of adult re- bacterial growth (table 4). Gram stains spiratory distress syndrome (ARDS): one of the PSB specimens were negative in

860

MEDURI, BEALS, MAIJUB, AND BASELSKI

TABLE 3 RESULTS OF QUANTITATIVE BACTERIAL CULTURES No Pneumonia

Pneumonia

Growth

PBAl

PSB

PBAl

Absent

0/13

PSB

18/33

21/33

Insignificant PSB < 1,000 cfu/ml PBAl < 10,000 cfu/ml

14/33 Mean 765 (10-4,000)

7/33 Mean 165 (10-300)

1/13 4 x 103

7/13 Mean 59 (10-150)

1/13

Significant PSB ~ 1,000 cfu/rnl PBAL ~ 10,000 cfu/ml

1/33* 2 x 105

5/33 Mean 22 x 103 (1-100 x 103)

12/13 Mean 14 x 105 (1-53 X 105)

5/13 Mean 5.2 x 105 (1.8-1,200 x 103 )

" False-positive result in a patient with endobronchial narrowing from primary lung carcinoma.

TABLE 4 RESULTS OF GRAM STAINS OF THE PROTECTED BAL

Type Pneumonia, V Pneumonia, NV No pneumonia, V Control Total %

Number 9 4 14 18 45

True Positive

True Negative

False Positive

False Negative

9 3

12/13 92

14 17 31/32 97

1" 1/32 3

1/13 8

Definition of abbreviations: V = ventilated patients; NV = nonventilated patients. " False-positive result in a control patient with endobronchial obstruction.

all patients. Giemsa stains of the PBAL effluents from 25 patients showed no intracellular organisms (lCO) in 24 patients. One patient had 1070 of the total retrieved alveolar cells containing ICO.

Patients with a Definitive Diagnosis of Pneumonia A definitive diagnosis of pneumonia was established in 13patients: 9 were mechanically ventilated (Group S1), 2 had hospital-acquired (Group S2), and 2 had community-acquired pneumonia (Group S3). Diagnosis was made by histology in 3 (2 autopsies and 1 open lung biopsy), by blood cultures in 2, and by clinical and radiographic responses to narrow spectrum antibiotic therapy directed against the organism(s) recovered by culture in significant growth in 8.

Results of Bacterial Cultures Significant growth was seen in 12(92%) PBAL and in 5 (38070) PSB specimens

(p < 0.01; table 5). The mean growth in the positive PBAL specimens was 14 x 105 cfu/ml. Only 1 patient had a growth < 105 cfu/ml. The mean growth in the positive PSB specimens was 5.2 x 105 cfu/ml (table 3). A correlation was made between the results of PSB specimen cultures and the number of days patients had been off antibiotic therapy before having bronchoscopy (table 6). In the four patients who

had been off antibiotics for only 2 days, the PSB specimens were positive in one and negative in three (25% sensitivity). In the two patients who had been off antibiotics for 4 days, one culture was positive and one was negative (50070 sensitivity). In the seven patients who had been off antibiotics for more than 10 days, three PSB specimens were positive and four were negative (43% sensitivity). When patients were diagnosed with PSB the odds in favor of a true positive culture increased an estimated 47070 (p < 0.06) for each additional day off antibiotics. The same organism grew in both PSB and PBAL specimens in 10 of 11 patients. Two PBAL specimens grew an additional organism in significant concentration. Polymicrobial growth was seen in three patients (23070). One patient had been off antibiotics 4 days, and two had been off antibiotics greater than 10 days. The only false negative PBAL culture was from a patient off antibiotics for 48 h who had a recently developed pulmonary infarction (thromboembolic) with a nidus of infection. Bacterial isolates were grown from the PBAL specimens in each one of the three (10-1, 10-3, and 10-5 ) dilutions of the aerobic culture plates from 11 patients and in two dilutions from the other 2.

Results of Gram and Giemsa Stains Gram stains of the PBAL cytospin spec-

imens were positive in all but one patient (92%), and the stains correctly identified the organisms that grew in significant concentrations (table 4). Of the 3 patients with polymicrobial growth 2 had both organisms identified by Gram stain. Gram stains of the PSB specimens were positive in only 3 patients, including 2 with positive cultures. Giemsa stains of the PBAL effluents from 10 patients (table 5) showed that 5 % or more (mean 17070) of retrieved alveolar cells contained intracellular organisms in 7 patients and only 2% ICO in 1. The two patients with negative results were both transplant recipients on immunosuppressive therapy including corticosteroids.

Complications The procedures were generally well tolerated. One patient with ARDS (mechanically ventilated) had transient oxygen desaturation with associated bradycar.. dia but suffered no lasting ill effects. One patient developed a postprocedure pneumothorax. Seven patients developed bronchial hemorrhage: three after PSB (two moderate and one minimal), and four after PBAL (all minimal). None of these bleeding episodes were hemodynamically significant, and no lasting ill effects were noted. Discussion

Historical Review The management of pulmonary infections is simplified and more effective when the etiologic agent causing the pneumonia is identified. To establish a bacteriologic diagnosis of lower respiratory tract infection physicians have traditionally relied upon the findings of sputum Gram stain and culture. Sputum analysis, however, yields reliable results in fewer than 50070 of patients with community-acquired pneumonia (1) and has a limited value for patients with nosocomiallung infection. Hospitalized patients frequently have the upper airways (2, 3) or the endotracheal tube (4-7) colonized by a high concentration of potentially pathogenic organisms. Secretions that are expectorated or suctioned from the lower respiratory tract mix with the flora of the oropharynx or the endotracheal tube and make it difficult or impossible to differentiate infecting from colonizing bacteria. Fiberoptic bronchoscopy (FOB), developed by Ikeda and coworkers in 1968 (8), provides direct visual access to the lower airways for sampling bronchial and parenchymal tissues. To reach the bronchial tree, however, the bronchoscope

PROTECTED BRONCHOALVEOLAR LAVAGE

861 TABLE 5

RESULTS OF QUANTITATIVE BACTERIAL CULTURES IN PATIENTS WITH PNEUMONIA Cultures (cfu/mQ

Mode of Diagnosis'

Patient 1 2 3 4 5 6

1 3, 7

7

7

6 6

PSB

PBAl

10 Staphylococcus aureus 80 S. aureus No growth 30 Streptococcus pneumoniae 1 x 105 Pseudomonas aeruginosa 10 X 105 S. aureus 12 x 105 Enterobacter Sp 3 x 105 Pseudomonas cepacia

6

6

Stains

8 9 10 11 12

1 6 7 1 7

20 S. aureus 80 S. aureus 2.8 x 103 Branhamella catarrhalis 50 S. aureus 1.8 x 103 Klebsiella oxytoca

13

3

140 a strept

1 1 2 25 5 10 10 53 15 13 1 3.5 4 30 2.5 2

105 105 105 105 105 105 105 105 105 105 105 104 103 105 104 105

x x x x x x x x X

x x x x x x x

S. aureus S. aureus Haemophilus influenzae S. pneumoniae P. aeruginosa S. aureus Enterobacter Sp P. cepacia S. aureus S. aureus S. aureus B. catarrhalis S. aureus K. oxytoca S. aureus Staphylococcus epidermidis

Days Off Antibiotics

Gram

Giernsaf

TP TP TP TP TP TP

TP/10 TP/15 TP/5 TP/2 TP/30

> 10

TP

TP/53

> 10

TP

TP/7.5

2 2 > 10 2 4

TP TP TP FN

FN:J:IO TP/9 FN:J:IO

> 10 4

> 10 > 10 2

> 10

Definition of abbreviations: TP = true positive; FN = false negative. • Refer to figure 3 for definition of pneumonia. t Number indicates percentage of cells with intracellular organism. :j: Both false-negative (FN) results were in patients on corticosteroids.

TABLE 6 RESULTS OF BACTERIAL CULTURES AND PREVIOUS USE OF ANTIBIOTIC THERAPY Days Off Antibiotics

> 10 TP

FN

TP

PBAl PSB

7 3

4

2 1

Mean growth, cfu/ml PBAl PSB

16.9 x 105 3.5 x 105

Results

Definition of abbreviations: TP

2

4

= true-positive

FN

15.5 X 105 940 result; FN

must traverse the oropharynx, nasopharynx, or the endotracheal tube, wherecontamination is likelyto occur. In addition, splinting of the vocal cords by the instrument favorsthe passage of oral secretions into the lungs during the procedure and further increases the risk 01 contamination. In a study of 16 patients without lung infection who underwent flexible FOB, Bartlett and colleagues (9) found all bronchoscopic aspirates to be contaminated by oropharyngeal bacteria. Spraying a methylene blue marker onto the posterior pharynx before bronchoscopy, they demonstrated that passage of the instrument introduced oropharyngeal contaminants into the suction channel. PSB and BAL, two bronchoscopic techniques to collect lower airway secretions, have received widespread acceptance in diagnosing pneumonia. The specimen is obtained from the lung segment with a new-infiltrate based on chest radiograph or the segment that has, upon inspection, purulent secretions.

=

TP

FN

3 1

1 3

4.7

X

105

287

false-negative result.

PSB wasintroduced by Wimberleyand coworkers in 1979(10), who perfected a technique previously described with rigid bronchoscopy. The brush, however, samples only a small portion of the peripheral airways, and the amount of secretion retrieved varies from 0.01 to 0.001 mI. The bronchoscopist must therefore follow a precise methodology in obtaining the specimen to minimize contamination of the tracheobronchial tree (no injection of topical lidocaine) and the bronchoscope (avoid suctioning). Quantitative bacterial cultures of lower airway secretions in the absence of antimicrobial chemotherapy are used to differentiate contaminants (low colony count) from bacteria causing infection (high colony count). Repeated studies have shown that bacterial infections of the lung, as well as infections at other anatomic sites, contain 105 or more bacteria/ml of exudate (cfu/ml) (11). A growth at 10-3 dilution of 0.01 or 0.001 ml of PSB secretions represents 105 to 106

bacteria/ml and has been proved by independent investigators to be the diagnostic threshold for pneumonia (12-14). BAL refers to the sequential instillation and aspiration of a physiologic solution (usually 100 to 200 ml) into a lung subsegment through a fiberoptic bronchoscope wedged in an airway. This technique samples a larger portion of lung tissue (approximately 1 million alveoli) because the alveolar surface area distal to the tip of the wedged bronchoscope is estimated to be 100 times greater than that ofthe peripheral airways. Although BAL provides a better reflection of lung microbiology and alveolar content, it is subject to the same risk for contamination previously described for bronchoscopic aspirates. The diagnostic accuracy of BAL has been well established for pneumonia caused by organisms that do not colonize the upper airways and for which contamination of the retrieved secretions does not represent a diagnostic dilemma. BAL has replaced open lung biopsy in diagnosing opportunistic infections in the immunosuppressed host (15) and is useful in detecting pathogens that cause pneumonia by the inhalational route (instead of aspiration), such as tuberculosis (16), legionella (17), and mycoplasma (18, 19). Quantitative cultures of BAL effluents werefirst compared with quantitative cultures of the PSB specimens in six baboons with respiratory failure, not on antibiotics, that had moderate to severe pneumonia based on histology and a polymicrobial growth on tissue culture (5). BAL recovered 74070 of all species

MEDURI, BEALS, MAIJUB, AND BASELSKI

862

present in the lung tissue compared with 410,10 recovered by PSB and 560,10 by transthoracic needle aspirate. Bacterial growth in the BAL was linearly related to tissue values, and the investigators concluded that, quantitatively and qualitatively, BAL specimens provided the most accurate window on lung microbiology. The results of human studies investigating the diagnostic value of BAL bacterial cultures in patients with and without pneumonia are shown in tables 1and 2, respectively, and are compared with our results with PBAL. In a group of 62 patients without pneumonia and undergoing bronchoscopy, Kahn and Jones (20) found that 17 of the 62 patients (27%) had more than 1070 squamous epithelial cells (SEC) in their BAL fluid, an indication of heavy contamination by oropharyngeal flora. Bacterial isolates grew above the cutoff point of lOS cfulml in 11 of these 17 patients (65%) and in only 4 of 57 (70,10) with ~ 1070 SEC. In 13immunosuppressed patients with bacterial pneumonia an uncontaminated BAL (~ 1% SEC) grew at least one microorganism above the 105 cfu/ml threshold. Unprotected BAL, however, could not discriminate pneumonia from bronchitis: 3 of 5 patients with acute bacterial bronchitis had significant growth (> 105 cfu/ml). Thorpe and coworkers (21) analyzed BAL fluid with quantitative bacterial cultures at one dilution by plating 0.01 ml of BAL effluent directly onto culture media and found a growth above lQ4 cfu/ml in all 23 of their patients with pneumonia (8 were immunosuppressed) and in only 4 of 58 (7%) without pneumonia. Gram stain of the cytocentrifuged BAL fluid was positive (one or more organisms seen per 1,000 x oil-immersion field) only in those with bacterial pneumonia (sensitivity of 73%). The role of BAL in diagnosing nosocomial bacterial pneumonia in ventilated patients was addressed by Torres and colleagues (6), who compared the yield of PSB and BAL in 25 patients with ventilator-associated pneumonia recently placed (~ 12h) on antibiotic therapy. The microbiologic processing of the BAL fluid was different from other previously described methods (5, 20-22), and a threshold of l()3 cfulml, similar to that for PSB, was used. A significant growth was found in 720,10 of patients with each technique and in two of seven BAL effluents in a control group (71 % specificity). Agreement between BAL and PSB was 75% with respect to the type of organisms re-

covered; however, it dropped to 56% when the concentration of organisms also was considered. In a smaller study Chastre and coworkers (22) observed a significant degree of overlap in culture results. Using a threshold of 105 cfu/ml, 40% of patients with pneumonia and 150,10 ofpatients without infection would have been misdiagnosed. Microscopic analysis of BAL was shown to be very important for the early recognition of pneumonia before the results of quantitative bacterial cultures were available. In a group of 61 patients intracellular organisms wereseen in more than 7% of the retrieved alveolar cells in 12 of 15 patients with pneumonia and in only 2 of 47 without pneumonia (960,10 specificity) (23). BAL samples a larger and more representative volume of respiratory secretions than PSB, and the former's value in rapidly identifying patients with bacterial pneumonia and in directing antimicrobial therapy has been recognized. Different diagnostic thresholds (103, 104 , and 105 cfu/ml) reported by various investigators depend in part upon the amount of BAL solution injected (table 1), the concentration of organisms in the lung, and the degree of contamination of the BAL fluid in the bronchoscopic channel. Contaminants in the BAL effluent are found in 89% of samples, frequently at a high concentration (table 2). Several methods have been adopted to minimize contamination ofthe BAL fluid, including avoidance of suctioning through the bronchoscope before lavage, performing the procedure in the supine or slight Trendelenburg's position, or discarding the first aliquot of BAL fluid retrieved, which has the highest degree of bronchial contamination. Piperno and colleagues (24) reported on a technique to perform blind nonbronchoscopic BAL (NB-BAL) in intubated patients using a cuffed No.7 French right heart catheter. In 13patients NB-BAL and open lung biopsy of the same lobe were performed immediately after death while mechanical ventilation was continued. Among the 10 positive BAL cultures lung biopsy showed histologic pneumonia in 9 cases and NB-BAL correctly identified 13 of the 14 organisms isolated in lung cultures.

Protected Bronchoalveolar Lavage Our study was designed to investigate the effectiveness of a novel technique, PBAL, in collecting lower airway secretions with a low degree of contamination. The following discussion is centered

around the five objectives of this investigation. I. Contamination of the PBAL specimen. Thirty-three patients had no pneumonia. The PBAL had significant growth for only one patient who had structural endobronchial narrowing, a condition known to frequently cause lower airway colonization (13). The PSB had significant growth in five specimens. If the control group that had PSB performed after PBAL is excluded (C3 ) , only 2 of 27 specimens had a false positive result. Excluding this group from analysis of the PSB results, the specificity ofthe test for PSB was 92070 and for PBAL was 970,10. Both techniques wereeffective in collecting uncontaminated distal airway secretions, with 550,10 of the PBAL and 640,10 of PSB specimens having no growth. If the group that received lidocaine injection (Group Ci) is excluded, however, sterile cultures were obtained in 590,10 of PBAL and in 67% of PSB specimens, excellent results considering that the procedure was performed in hospitalized patients with lung disease. Squamous epithelial cells were absent in 710,10 of PBAL specimens and below the previously established contamination threshold of 2070 in 85% (table 7). Five patients had 20,10 or more SEC. Two of the five belonged to the control group who received lidocaine injection through the FOB. If this group is excluded 91070 of patients had ~ 1% SEC. II. Diagnostic value ofthe PBAL specimen. Quantitative bacterial cultures of the PBAL specimen had one false positive (1 of 33) and one false negative (l of 13)result. Using a diagnostic threshold of 104 cfulml, the sensitivity and specificity of the test were 92 and 970,10, respectively. The positive predictive and negative predictive values were 97 and 920,10, respectively. Both specificity and sensitivity appear superior with PBAL compared with routine BAL, as reported in the literature. Giemsa stains of the PBAL effluents identified intracellular organisms, in ~ 2% of retrieved alveolar cells, in all but two patients with pneumonia, both transplant recipients on corticosteroids. The sensitivity and specificity of this test were 80 and 100%, respectively. Gram stains of the PBAL effluents were positive in all but one patient with pneumonia and negative in all but one patient without infection (table 4). One or both of the two stains were positive in all patients with pneumonia (100% sensitive) and in only one patient without pneumonia (97% specificity), allow-

863

PROTECTED BRONCHOALVEOLAR LAVAGE

TABLE 7 PERCENTAGE OF SQUAMOUS EPITHELIAL CELLS IN THE PBAL EFFLUENT SEC (%) Control Study nonventilated Study ventilated Total

%

0%

1%

2%t

16 3 16

12 1 12

2 1 2

1:1: 1 1

35 100

25 71

5 14

3

1

9

3

Number*

5%

7% 1:1:

1 3

Definition of abbreviations: SEC = squamous epithelial cells. • Number of patients in whom Giemsa stains of PBAL were obtained. t Lack of heavy oropharyngeal contamination < 2%. :j: Both patients in the control group with ~ 2% SEC had lidocaine injection through the bronchoscope (Group C,) before PSB and PBAL sampling.

ing early and accurate diagnosis of pneumonia before the results of cultures are available (24 to 48 h). In our institution the microbiology laboratory processing time for these two stains is usually 2 h. Early availability of these data from stained smears has led to improvement in patient care, particularly of the immunosuppressed and ventilated population.

III Effects of topical anesthesia administration and sequence of sampling procedures. The control g~oup was divided into 3 subgroups of 6 patients each to assess whether the mode of delivery of topical anesthesia and the sequence of sampling affect the results of cultures. Injection of lidocaine through the bronchoscope and onto the vocal cords (Group C 1) resulted in a higher incidence of contamination for both PSB (50 versus 420/0)and PBAL specimens (67 versus 420/0). The sequence of sampling affected the culture results ofthe PSB sample but not the PBAL effluent. False positive results (~ 1,000 cfu/ml) were seen in 170/0 ofPSB specimens when PSB was performed before PBAL (Groups C 1 and C 2 ) and in 50% of specimens when PSB was performed after PBAL (Group C 3 ) . In the study group we therefore used high-dose, nebulized lidocaine for topical anesthesia and avoided injection or suction through the bronchoscope before sampling. PSB was performed before PBAL. The safety and efficacy of using high-dose nebulized lidocaine to obtain endobronchial cultures have been reported (25). I~

Effect ofprevious antibiotic use.

Thirteen patients had pneumonia. Most of them were hospitalized, and four were on antibiotic therapy that was discontinued at least 48 h before bronchoscopy. Discontinuing antimicrobial treatment before bronchoscopy was not associated with clinical worsening, nor did it affect outcome to the best of our knowledge. In two candidates for the

study (not entered), discontinuation was followed by rapid resolution of fever. Another goal of our investigation was to assess the effect of previous antibiotic administration on culture results. In 11 of 12 patients the PSB specimen grew an organism identical to the one recovered with PBAL; however, in most cases the concentration was below the diagnostic threshold of 1,000 cfu/ml. The time off antibiotics before bronchoscopy affected the amount of bacterial growth in both PSB and PBAL cultures and the positivity of the test (table 6). The PSB specimen, as previously discussed, retrieves a small amount of undiluted respiratory secretions and may therefore be more susceptible to the effects of previous antibiotic administration. This may explain the low rate of true positive results seen in the group of patients off antibiotic treatment for 48 h compared with the group off antibiotics 4 or more days (25 versus 440/0). The only false negative result with PBAL cultures was in a patient off antibiotics for 2 days. Intracellular organisms were seen on the Giemsa stain, and an autopsy 3 days after bronchoscopy provided histologic evidence of pneumonia and a recent pulmonary infarction. ~ Predicted accuracyofquantitative cultures at one dilution in diagnosing pneumonia. Two findings lead us to be-

lieve that similar bacteriologic results may be obtained with quantitative bacterial cultures at one dilution instead of three. First, we have found a wide margin between the bacterial growth in the PBAL specimens of patients with and without pneumonia. Second, bacterial isolates grew in only the first of the three dilutions (with one exception) on culture plates in 80% of patients without pneumonia and in all three dilutions in 85% of patients with pneumonia. Therefore, quantitative bacterial cultures at one dilution (10- 3 ) appear to be adequate. This is obtained by plating 0.001 ml of BAL

fluid directly onto the culture media, similar to the processing of a urine culture, a simple and inexpensive method available in all microbiology laboratories.

Conclusions PBAL with a PBT catheter was effective in collecting distal respiratory tract secretions with a minimal degree of contamination. Giemsa and Gram stains of the PBAL effluents were useful in identifying patients with pneumonia before the results of cultures were available. Using a growth threshold of lQ4 cfu/ml to diagnose pneumonia, quantitative bacterial cultures had a sensitivity of 920/0 and a specificity of 970/0. Being off previous antibiotic treatment for at least 48 h before bronchoscopy did not affect the result of PBAL cultures, contrary to what we observed with PSB. These findings support the use of PBAL as a bronchoscopic technique to more rapidly and accurately diagnose bacterial pneumonia. Acknowledgment The writersthank Carol B. Jones, R.N., RS.N. for her expert contributions in data collection; Dr. Elizabeth A. Tolley for statistical analysis; Dr. G. Glen Mayhall for critical reviewof the manuscript; Dr. David Armbruster for editorial assistance; and Ms. VickyFranke for her secretarial expertise. This study could not have been completed without the enthusiastic work of the microbiology and cytology laboratory personnel at the VAMC, MED, and UTMC hospitals and the collaboration of all the fellows and attendings of the pulmonary division. Finally wewish to thank our bronchoscopy assistants, Donald Woolfolk, Cheryl Grisham, and Catherine Hollie, for their unselfish dedication to this project and their professional contribution. Special recognition goes to those investigators who contributed significantly in the early phases of this study: Craig Conoscenti, David Ostrowski, and Bob Gandi. References 1. Levy M, Dromer F, Brion N, Leturdu F, Carbon C. Community-acquired pneumonia. Importance of initial noninvasive bacteriologic and not radiographic investigations. Chest 1988; 92:43-8. 2. Bryan CS, Reynolds KL. Bacteremic nosocomial pneumonia: analysis of 172 episodes from a single metropolitan area. Am Rev Respir Dis 1984; 129:668-71. 3. Bartlett JG, O'Keefe P, Tally FP, Louie TJ, Gorbach SL. Bacteriology of hospital-acquired pneumonia. Arch Intern Med 1986; 146:868-71. 4. Higuchi JH, Coalson 11, Johanson WG Jr. Bacterial diagnosis of nosocomial pneumonia in primates: usefulness of the protected specimen brush. Am Rev Respir Dis 1982; 125:53-7. 5. Johanson WG Jr, Seidenfeld 11, Gomez P, De Los Santos R, Coalson J J. Bacteriologic diagnosis of nosocomial pneumonia following prolonged mechanical ventilation. Am Rev Respir Dis 1988;

MEDURI, BEALS, MAIJUB, AND BASELSKI

864 137:259-64. 6. Torres A, De La Bellacasa JP, Xaubet A, et al. Diagnostic value of quantitative cultures of bronchoalveolar lavage and telescoping plugged catheters in mechanically ventilated patients with bacterial pneumonia. Am Rev Respir Dis 1989; 140:306-10. 7. Hill JD, Ratliff JL, Parrott JCW, et al. Pulmonary pathology in acute respiratory insufficiency: lung biopsy as a diagnostic tool. J Thorac Cardiovasc Surg 1976; 71:64-71. 8. Ikeda S, Yanai N, Ishikaua S. Flexible bronchofiberscope. Keio J Med 1968; 17:1-16. 9. Bartlett JG, Alexander J, Mayhew J, SullivanSigler N, Gorbach SL. Should fiberoptic bronchoscopy aspiratesbe cultured? Am Rev RespirDis 1976; 114:73-8. 10. Wimberley N, Faling LJ, Bartlett JG. A fiberoptic bronchoscopy technique to obtain uncontaminated lower airway secretions for bacterial culture. Am Rev Respir Dis 1979; 119:337-43. 11. Bartlett JG. Invasive diagnostic techniques in pulmonary infections. In: Pennington JE, ed. Respiratory infections: diagnosis and management. 2nd ed. New York: Raven Press, 1989; 52-68. 12. Chastre J, Viau F, Brun P, et al. Prospective evaluation of the protected specimen brush for the diagnosis of pulmonary infections in ventilated patients. Am Rev Respir Dis 1984; 130:924-9.

13. Pollock M II, Hawkins EL, Bonner JR, Sparkman T, Bass JB Jr. Diagnosis of bacterial pulmonary infections with quantitative protected catheter cultures obtained during bronchoscopy. J Clin Microbiol 1983; 17:255-9. 14. Fagon JY, Chastre J, Hance AJ, et al. Detection of nosocomial lung infection in ventilated patients: use of a protected specimen brush and quantitative culture techniques in 147patients. Am Rev Respir Dis 1988; 138:110-6. 15. Stover DE, Zaman MB, Hajdu SI, Lange M, Gold J, Armstrong D. Bronchoalveolar lavage in the diagnosis of diffuse pulmonary infiltrates in the immunosuppressed host. Ann Intern Med 1984; 101:1-7. 16. De Gracia J, Curull V, Vidal R, et al. Diagnostic value of bronchoalveolar lavage in suspected pulmonary tuberculosis. Chest 1988;93:329-32. 17. Kohorst WR, Schonfeld SA, Macklin JE, Whitcomb ME. Rapid diagnosis of Legionnaires' disease by bronchoalveolar lavage. Chest 1983; 84:186-90. 18. Parides GC, Bloom JW, Amepel NM, RayCG. Mycoplasma and ureaplasma in bronchoalveolar lavage fluids from immunocompromised hosts. Diagn Microbiol Infect Dis 1988; 9:55-7. 19. Lehtomaki K, Kleemola M, Tukianen P, Kantanen ML, Laitinen LA. Isolation of Mycoplasma pneumoniae from bronchoalveolar lavage fluid. J

Infect Dis 1987; 155:1339-41. 20. Kahn FW, Jones JM. Diagnosing bacterial respiratory infection by bronchoalveolar lavage. J Infect Dis 1987; 155:862-9. 21. Thorpe JE, Baughman RP, Frame PT, WesseIer TA, Staneck JL. Bronchoalveolar lavage for diagnosing acute bacterial pneumonia. J Infect Dis 1987; 155:855-61. 22. Chastre J, Fagon JY, Soler P, et al. Diagnosis of nosocomial bacterial pneumonia in intubated patients undergoing ventilation: comparison of the usefulness of bronchoalveolar lavage and the protected specimenbrush. Am J Moo 1988; 85:499-506. 23. Chastre J, Fagon JY, Soler P, et al. Quantification of BAL cells containing intracellular bacteria rapidly identifies ventilated patients with nosocomial pneumonia. Chest 1989; 95:190S-2S. 24. Piperno D, Gaussorgues P, Bachmann P, Jaboulay JM, Robert D. Diagnostic value of nonbronchoscopic bronchoalveolar lavage during mechanical ventilation. Chest 1988; 93:223. 25. Berger R, McConnell JW, Phillips B, Overman TL. Safety and efficacy of using high-dose topical and nebulized anesthesia to obtain endobronchial cultures. Chest 1989; 95:299-303. 26. Kirkpatrick MB, Bass JB Jr. Quantitative bacterial cultures of bronchoalveolar lavage fluids and protected brush catheter specimens from normal subjects. Am Rev Respir Dis 1989; 139:546-8.