Abstract
Noninvasive and invasive diagnostic techniques have been shown to achieve comparable performances in the evaluation of suspected ventilator-associated pneumonia (VAP). We studied the impact of both approaches on outcome in a prospective, open, and randomized study in three intensive care units (ICUs) of a 1,000-bed tertiary care university hospital. Patients with suspected VAP were randomly assigned to noninvasive (Group 1) versus invasive (Group 2) investigation (tracheobronchial aspirates [TBAS] versus bronchoscopically retrieved protected specimen brush [PSB] and bronchoalveolar lavage [BAL]. Samples were cultured quantitatively, and BAL fluid (BALF) was examined for intracellular organisms (ICO) additionally. Initial empiric antimicrobial treatment was administered following the guidelines of the American Thoracic Society (ATS) and adjusted according to culture results (and ICO counts in Group 2). Outcome variables included length of ICU stay and mechanical ventilation as well as mortality. Overall, 76 patients (39 noninvasive, 37 invasive) were investigated. VAP was microbiologically confirmed in 23 of 39 (59%) and 23 of 37 (62%) (p = 0.78). There were no differences with regard to the frequencies of community-acquired and potentially drug-resistant microorganisms (PDRM). Antimicrobial treatment was changed in seven patients (18%) of Group 1 and 10 patients (27%) of Group 2 because of etiologic findings (including five of 17 with ICO = 2% (p = not significant [NS]). Length of ICU stay and mechanical ventilation were also not significantly different in both groups. Crude 30-d mortality was 31 of 76 (41%), and 18 of 39 (46%) in Group 1 and 14 of 37 (38%) in Group 2 (p = 0.46). Adjusted mortality was 16% versus 11% (p = 0.53), and mortality of microbiologically confirmed pneumonia 10 of 23 (44%) in both groups (p = 1.0). We conclude that the outcome of VAP was not influenced by the techniques used for microbial investigation.
Ventilator-associated pneumonia (VAP) occurs in approximately 20 to 30% of patients mechanically ventilated for more than 48 h (1, 2). Mortality of patients with VAP may exceed 50% (2, 3). Although attributable mortality of VAP remains controversial, most studies indicate that, in contrast to early-onset VAP, late-onset VAP has an attributable mortality of 20 to 30% (4, 5). Rapid initiation of an appropriate antimicrobial treatment is crucial because inappropriate antimicrobial treatment has been identified as an adverse prognostic factor (6). On the other hand, antimicrobial drugs have been shown to represent an independent risk factor for the development of VAP due to high-risk potentially multiresistant microorganisms (7). Thus, the foremost importance to establish adequate diagnostic techniques is evident.
In the last decade, considerable corresponding efforts have been made, including the comparison of noninvasive and invasive bronchoscopic techniques (8, 9) as well as their validation in postmortem studies using histology and tissue microbiology as reference tests (10-15). Despite these extensive investigations, the method of choice for the diagnosis of VAP remained controversial (16, 17). Moreover, even autopsy histology was found to contain inaccuracies (18). It became evident that there is no irrefutable reference to evaluate diagnostic techniques.
Nevertheless, it has been argued that most techniques have operating characteristics comparable to generally accepted procedures such as ventilation perfusion scans and computed tomographic (CT) staging of bronchial carcinoma (19). Together with careful clinical judgement, microbial investigation may therefore contribute to an optimal management of patients with clinical suspicion of VAP (20). If we accept this view, the crucial issue of the impact of different diagnostic approaches to important clinical outcome variables remains to be settled. Clinical decision analysis studies are of limited value because they largely depend on the exact operative characteristics that are assumed for each technique (21, 22). Preliminary clinical studies have suggested that there are no significant differences with regard to morbidity and mortality using noninvasive as compared with invasive diagnostic approaches (23-25). However, in these studies, initial empiric antimicrobial treatment was not standardized. Moreover, the potential impact of significant intracellular organisms (ICO counts) in bronchoalveolar lavage fluid (BALF) on initial empiric antimicrobial treatment decisions was not evaluated.
We therefore carried out a randomized study to compare outcome variables in patients investigated by either diagnostic approach (including ICO counts in BALF) and treated by a standardized algorithm according to the American Thoracic Society (ATS) guidelines for the initial empiric antimicrobial treatment of nosocomial pneumonia (26).
We prospectively studied 76 consecutive patients admitted to a respiratory or surgical intensive care unit (ICU) of a 1,000-bed teaching hospital and requiring mechanical ventilation for more than 48 h with clinical suspicion of VAP during an 18-mo period. Clinical suspicion of VAP was based on the presence of new and/or progressive infiltrates in chest radiograph, and at least two of the following criteria: fever ⩾ 38° C or hypothermia ⩽ 35° C, leukocytosis ⩾ 12 × 109/L or leukopenia < 4 × 109/L, or purulent respiratory secretions. Exclusion criteria were immunosuppression (human immunodeficiency virus [HIV] infection, organ transplantation, solid and hematologic malignancies) and critically ill patients without an estimated life expectancy of their underlying disease of at least 3 mo. The study was approved by the local ethical committee and informed consent was obtained in each case from the next of kin.
Patients were randomly assigned using a computer-generated randomization table into one of two groups, noninvasive investigation (Group 1) or invasive investigation (Group 2) of suspected VAP. The first group underwent quantitative tracheobronchial aspiration (TBAS), the second group fiberoptic bronchoscopy with protected specimen brush (PSB) and bronchoalveolar lavage (BAL).
The following variables were recorded: age, gender, comorbidity, McCabe score, causes of ICU admission, cause of mechanical ventilation, and antimicrobial treatment prior to diagnostic evaluation for pneumonia. Moreover, Acute Physiology and Chronic Health Evaluation (APACHE) II score and ratio of arterial oxygen pressure to fraction of inspired oxygen (PaO2 /Fi O2 ) on admission (with predicted mortality at the same time), body temperature, mean arterial blood pressure, acute lung injury score (27), multiple organ failure (MOF) score (28), duration of mechanical ventilation, duration of ICU stay, and cost for antimicrobial treatment/patient/day were documented.
In all patients, two blood cultures were obtained. TBAS in Group 1 was obtained by sterile means using a Mocstrap suction catheter and collected in a mucus collector (Productes Clinics S.A; La Llagosta, Barcelona, Spain). Bronchoscopy in Group 2 patients was performed without interrupting mechanical ventilation through the endotracheal tube and using a special adaptor. Ventilator settings were adapted during the procedure to ensure proper oxygenation and ventilation. After adequate sedation and curarization, the bronchoscope (BF30; Olympus, New Hyde Park, NY) was introduced without instilling local anestetics and avoiding bronchial suctioning. The sequence of sampling was always PSB followed by BAL. The PSB (Microbiology brush; Mill-Rose Lab., Mentor, OH) sample was retrieved from the area of infiltration in the chest radiograph. Thereafter, the bronchoscope was wedged into a subsegmental bronchus from the same area where PSB was performed. A volume of 150 ml sterile saline was instilled in aliquots of 20 ml, 30 ml, 50 ml, and 50 ml. The first aliquot was discarded. BAL was not performed in patients with PaO2 /Fi O2 < 100 or hemodynamic instability.
TBAS samples were mechanically homogenized using glass beads and were vortexed for 1 min. PSB was aseptically cut into a sterile tube containing 1 ml of Ringer's lactate and vortexed for 1 min. Serial dilutions from TBAS were prepared in sterile normal saline. PSB and BAL were mixed 1:1 with sterile normal saline. Thereafter, serial dilutions were inoculated into the following agar media: 5% sheep blood, chocolate, CDC agar, blood charcoal yeast extract (BCYE-α), and Sabouraud dextrose, BHA, and BHACC agar. All cultures were incubated at 37° C under aerobic and CO2-enriched atmosphere. Cultures were evaluated for growth 24 h and 48 h later and discarded, if negative, 5 d after, except for Sabouraud dextrose, BHA, and BHACC agar which were evaluated for 4 wk and BCYE-α media which were evaluated at 10 d. All microorganisms isolated were identified by standard laboratory methods (29). An aliqout of BALF was centrifuged and samples were stained with Gram and May-Grünwald-Giemsa stain. Overall 200 cells were examined at × 100 magnification and the percentage of cells containing intracellular organisms per field (ICO count) was determined.
No patient received antimicrobial treatment for the current VAP episode prior to the study. After diagnostic sampling, all patients received initial empiric antimicrobial treatment according to ATS guidelines (26). Modifications of antimicrobial treatment were made on the basis of ICO counts in Gram stain of BALF samples available at the day of investigation (only Group 2) and by the results of cultures of TBAS or PSB and/or BAL, available after > 24 h. In patients with persistent clinical suspicion of VAP, antimicrobial treatment was not stopped even in the presence of nonsignificant or negative culture results.
Response to antimicrobial treatment was evaluated within 48 to 72 h after its initiation. Failure of antimicrobial treatment for VAP was considered present if at least one of the following criteria was met: (1) persistent body temperature = 38° C or < 35 ° C and purulent tracheobronchial secretions; (2) radiographic spread of infiltrates = 50%; (3) development of septic shock or multiorgan failure (as defied by MOF score = 6); and (4) death probably or definitely caused by pneumonia. This was assumed on clinical grounds or autopsy in the absence of evidence for other extrapulmonary causes of death.
In case of failure of antimicrobial treatment, patients were completely reevaluated for extrapulmonary causes of infection. In addition, all patients underwent a repeated microbial investigation relying on the same technique to which the patient was initially assigned. Antimicrobial treatment was changed according to the pathogens isolated (regardless of colony counts) or empirically in case of negative cultures. The latter regimen basically consisted of vancomycin, imipenem, and amikacin. Initial empiric antimicrobial treatment was considered inadequate if the isolated microorganism was not adequately covered or resistant (despite possible changes made by ICO observed in BAL in group 2). Monotherapy for Pseudomonas aeruginosa was also considered inadequate.
To determine the cost of both approaches, the following charges were considered: Noninvasive sampling: suction trap, cultures; invasive sampling: bronchoscopy procedure, PSB device, cultures.
Pneumonia was considered microbiologically confirmed in the presence of clinical suspicion of pneumonia and (1) a blood culture revealing a bacterial pathogen in the absence of an extrapulmonary focus or (2) growth in bacterial culture of TBAS 105 colony-forming units (cfu)/ml in Group 1, and PSB 103 cfu/ml or BAL 104 cfu/ml in Group 2. ICO counts were considered significant in the presence of = 2% ICO in 200 cells (20). The onset of VAP was classified according to the ATS guidelines, i.e., early-onset up to 4 d and late-onset after = 5 d after hospitalization (26). Bacterial microorganisms were classified into potentially pathogenic microorganisms (PPMs) or nonpotential pathogenic microorganisms (non-PPMs) as described elsewhere (30). For purposes of analysis, bacterial PPMs were classified as community-acquired (Streptococcus pneumoniae, methicillin-susceptible Staphylococcus aureus [MSSA], Haemophilus influenzae), and potentially drug-resistant microorganisms (PDRMs), including methicillin-resistant S. aureus (MRSA), P. aeruginosa, Stenotrophomonas maltophilia, and Acinetobacter spp).
The study was designed to detect a reduction from 40% (1, 4, 5) in Group 1 to 10% in Group 2 in mortality rate 30 d after diagnostic evaluation with a confidence level of 95% and a power of 1-β of 80%. Accordingly, the calculated sample size was 76 patients distributed randomly into the two study groups. Predicted hospital mortality rates were calculated from APACHE II scores and the diagnostic category score according to Knaus and coworkers (31). Adjusted mortality was defined as the difference between the number of crude observed and the number of predicted hospital deaths.
Results are expressed as mean ± SD. The t test for quantitative variables and the chi-square test (Fisher exact test when needed) to compare proportions were used. Multivariate analysis to predict significant growth in culture was performed by stepwise forward logistic regression. All p values are two-sided and the level of significance was set at 5%.
Overall, 76 patients were randomized (Group 1, noninvasive sampling, n = 39 and Group 2, invasive sampling, n = 37, including eight with PSB alone and 29 with PSB and BAL). The causes for ICU admission and general characteristics of the patient population are listed in Table 1. There were no significant differences between both groups.
| Group 1 (Noninvasive Sampling) | Group 2 (Invasive Sampling) | p Value | ||||
|---|---|---|---|---|---|---|
| Age, yr | 65 ± 11 | 67 ± 13 | 0.65 | |||
| Sex, M/F | 29/10 | 24/13 | 0.37 | |||
| Causes for ICU admission | ||||||
| Medical ICU | ||||||
| Respiratory | 14 | 11 | 0.57 | |||
| Cardiovascular | 7 | 9 | 0.49 | |||
| Neurologic | 3 | 2 | 0.69 | |||
| Surgical ICU | ||||||
| Cardiovascular | 8 | 7 | 0.86 | |||
| Respiratory | 5 | 4 | 0.79 | |||
| Gastrointestinal | 2 | 2 | 0.96 | |||
| Trauma | 0 | 2 | 0.23 | |||
| McCabe's score | 0.77 | |||||
| Nonfatal | 5 | 3 | ||||
| Ultimately fatal | 22 | 23 | ||||
| Rapidly fatal | 12 | 11 | ||||
| APACHE II on admission | 19 ± 6 | 20 ± 6 | 0.89 | |||
| Predicted mortality on admission, % | 30 ± 24 | 27 ± 24 | 0.48 | |||
| MOF score on admission | 3.2 ± 1.8 | 3.5 ± 1.7 | 0.7 | |||
| Length of mechanical ventilation on inclusion, d |
6.2 ± 5 | 6.0 ± 4 | 0.42 | |||
| PaO2 /Fi O2 on inclusion | 203 ± 98 | 235 ± 117 | 0.19 | |||
| Number of patients on antimicrobial treatment prior to VAP, n |
33 | 26 | 0.13 | |||
Bacterial cultures from Group 1 showed significant growth of PPMs above the threshold in 20 of 39 (51%) and from Group 2 in 22 of 37 (60%) patients (p = 0.47). With regard to Group 2, all 22 (100%) had significant growth in PSB and 11 (50%) in BALF. Fifteen patients (Group 1: n= 11, Group 2: n = 4; p = 0.06) had early-onset pneumonia (p = 0.06) and 61 patients (Group 1: 28, Group 2: 33) had late-onset pneumonia (p = 0.06). A total of 58 microorganisms were isolated in significant amounts and are listed in Table 2. Both groups had six patients with polymicrobial growth in culture (15% and 16%, respectively). There were no significant differences with regard to the frequencies of community-acquired pathogens (n = 11) and PDRM (n = 36), including P. aeruginosa (n = 23) in both groups. Another 21 PPMs were found in colony counts below the threshold, including community-acquired pathogens (n = 1) and PDRM (n = 16), including P. aeruginosa (n = 11). Finally, 21 mon-PPMs were isolated, including 15 in significant and six in nonsignificant amounts.
| Group 1 (n = 39) | Group 2 (n = 37) | |||||||
|---|---|---|---|---|---|---|---|---|
| Below the Threshold | Above the Threshold | Below the Thresholds | Above the Thresholds | |||||
| Streptococcus pneumoniae | 0 | 2 | 0 | 1 | ||||
| Staphylococcus aureus (MSSA) | 0 | 2 | 1 | 2 | ||||
| Staphylococcus aureus (MRSA) | 2 | 2 | 2 | 6 | ||||
| Haemophilus influenzae | 0 | 2 | 0 | 2 | ||||
| Enterobacter cloacae | 0 | 1 | 0 | 0 | ||||
| Serratia marcescens | 0 | 0 | 1 | 3 | ||||
| Proteus mirabilis | 0 | 1 | 0 | 1 | ||||
| Proteus vulgaris | 0 | 0 | 0 | 2 | ||||
| Morganella morgagni | 0 | 0 | 0 | 1 | ||||
| Eikenella corrodens | 0 | 1 | 0 | 0 | ||||
| Pseudomonas aeruginosa | 1 | 13 | 10 | 10 | ||||
| Stenotrophomonas maltophilia | 0 | 2 | 3 | 0 | ||||
| Acinetobacter baumanii | 0 | 1 | 1 | 2 | ||||
| Aspergillus fumigatus* | 0 | 1 | 0 | 0 | ||||
| Total | 3 | 28 | 18 | 30 | ||||
Overall, 17 of 29 (59%) patients in whom BAL was feasible had ICO counts equal to 2% in BALF. Of these, five of 17 (29%) had no pathogen in significant amounts in culture. On the other hand, six of 12 (50%) patients with ICO counts less than 2% had growth in culture above the threshold. Thus, using significant growth in culture as reference, the sensitivity of ICO counts was 12 of 17 (71%) and the specificity seven of 12 (58%). Blood cultures were positive in six patients (three in each group). In two cases, blood cultures revealed the same pathogen retrieved in significant amounts in cultures of respiratory secretion samples (S. pneumoniae and Serratia spp), in two further cases, concordance was only qualitative (P. aeruginosa and Acinetobacter spp), and in the remaining two cases, pathogens in blood culture were not present in cultures of respiratory secretion samples (H. influenzae and MSSA). The total number of microbiologically confirmed pneumonia was 23 of 39 (59%) and 23 of 37 (62%), respectively (p = 0.78).
Fifty-nine (78%) patients received antimicrobial treatment before the study for causes other than the current episode of VAP, including 33 (87%) from Group 1, and 26 (70%) from Group 2 (p = 0.08). Thirteen (17%) had recently introduced antimicrobial treatment within the last 72 h (nine in Group 1 and four in Group 2, p = 0.27), and 46 (61%) an antimicrobial treatment = 72 h (24 in Group 1 and 22 in Group 2, p = 0.27).
The reasons for antimicrobial pretreatment were (in Group 1 and 2, respectively): pulmonary indications (n = 21 and 19), including suspicion of aspiration during intubation (n = 2 and 4), purulent tracheobronchitis (n = 9 and 8), acute exacerbation of chronic obstructive pulmonary disease (COPD) (n = 4 and 6), community-acquired pneumonia (n = 5 and 1), empyema (n = 1 in Group 1); and nonpulmonary indications (n = 11 and 8), including postoperative prophylaxis (n = 8 and 2), urinary tract infection (n = 1 in Group 2), endocarditis (n = 1 in each group), meningitis (n = 1 in Group 2), peritonitis (n = 2 in each group), and phlebitis (n = 1 in Group 2).
The following prior antimicrobial drugs were administered as monotherapy or combination therapy: Group 1 aminoglycosides (n = 30), third-generation antipseudomonal cephalosporins (n = 20), vancomycin (n = 12), quinolones (n = 11), carbapenems (n = 6), third-generation cepahlosporins (n = 4), others (n = 7); Group 2: aminoglycosides (n = 22), third-generation antipseudomonal cephalosporins (n = 18), vancomycin (n = 11), quinolones (n = 6), carbapenems (n = 6), third-generation cephalosporins (n = 2), others (n = 8).
Microbial investigation revealed significant growth in culture in 14 of 17 (82%) patients without antimicrobial treatment (6 of 6 in Group 1 and 8 of 11 in Group 2) and in 26 of 46 (57%) with antimicrobial treatment for = 72 h (13 of 24 in group 1 and 13 of 22 in Group 2), but only in 4 of 13 (31%) patients with recently introduced antimicrobial treatment within the last 72 h before microbial investigation (3 of 9 in Group 1 and 1 of 4 in Group 2) (p = 0.02).
In a multivariate analysis including antimicrobial treatment before microbial investigation (yes/no), recently introduced antimicrobial treatment (< 72 h versus = 72 h), early- versus late-onset pneumonia, and diagnostic technique used (noninvasive versus invasive), only antimicrobial treatment before microbial investigation was inversely associated with significant growth in culture (RR 0.22 95% confidence interval [CI] 0.06 to 0.85, p = 0.03).
In patients with significant growth in bacterial cultures, initial antimicrobial treatment was changed in seven patients (18%) of Group 1 and 10 patients (28%) of Group 2 because of etiologic findings (p = not significant [NS]). The following microorganisms were responsible for antimicrobial treatment modifications in Group 1: S. pneumoniae (n = 1), MSSA (n = 1), MRSA (n = 1), P. aeruginosa (n = 2), S. maltophilia (n = 2), and Acinetobacter baumanii (n = 1). In one patient with Aspergillus fumigatus, treatment was not changed. Conversely, in Group 2, organisms implicated in the modification of initial antimicrobial treatment were: MSSA (n = 1), MRSA (n = 4), P. aeruginosa (n = 2), and A. baumanii (n = 1). In one patient with P. aeruginosa and another with A. baumanii, antimicrobial treatment was not changed. Adjustments in antimicrobial treatment were made in five of 17 (29%) patients with ICO counts of 2%.
The total duration of antimicrobial treatment for VAP was 12 ± 4 in Group 1 versus 13 ± 4 d in Group 2 (p = 0.48). Comparisons of the outcome of initial empiric antimicrobial treatment with regard to the diagnostic findings are listed in Table 3. Failure of initial empiric antimicrobial treatment was observed in 20 (51%) patients of Group 1 and 15 (41%) of Group 2 (p = 0.35). It was associated with inappropriate initial antimicrobial treatment in seven of 39 (18%) patients of Group 1 and six of 37 (16%) of Group 2 (p = 0.84).
| Outcome/Diagnostic Results | Group 1 n (%) | Group 2 n (%) | ||
|---|---|---|---|---|
| Cured | 19 (49) | 22 (59) | ||
| With significant growth in culture | 11 (28) | 9 (24) | ||
| With adequate treatment | 10 (26) | 5 (14) | ||
| With inadequate treatment | 1 (3) | 4 (11) | ||
| With nonsignificant growth of PPMs | 1 (3) | 6 (16) | ||
| With growth of non-PPMs | 2 (5) | 3 (8) | ||
| With negative cultures | 5 (13) | 4 (11) | ||
| Antimicrobial treatment failure | 20 (51) | 15 (41) | ||
| With significant growth in culture | 11 (28) | 13 (35) | ||
| With adequate treatment | 4 (10) | 7 (19) | ||
| With inadequate treatment | 7 (18) | 6 (16) | ||
| With nonsignificant growth of PPMs | 1 (3) | 1 (3) | ||
| With growth of non-PPMs | 4 (10) | 1 (3) | ||
| With negative cultures | 4 (10) | 0 |
Of 35 patients with failure of initial antimicrobial treatment, 11 were not reinvestigated because of unstable clinical conditions or death and 24 had repeated microbial investigation (15 in Group 1 and nine in Group 2). Of these latter patients, 17 (71%) had PPMs in culture, including eight (33%) with significant growth (five in Group 1 and three in Group 2). MRSA and P. aeruginosa were the main pathogens involved. Overall nine of 17 (53%) pathogens were already present in the initial investigation (five in Group 1 and four in Group 2). The results are listed in detail in Table 4.
| Diagnostic Result | Group 1 (n = 20) | Group 2 (n = 20) | ||
|---|---|---|---|---|
| No reevaluation | 5 | 6 | ||
| Significant growth in culture | 5 (3)* | 3 (1) | ||
| MRSA | 0 | 2 | ||
| Pseudomonas aeruginosa | 5 (3) | 1 (1) | ||
| Nonsignificant growth in culture of PPMs | 5 (2) | 4 (3) | ||
| MSSA | 1 | 0 | ||
| MRSA | 2 (1) | 1 (1) | ||
| Pseudomonas aeruginosa | 2 (1) | 3 (2) | ||
| Non-PPMs | 2 | 1 | ||
| Negative cultures | 3 | 1 |
The total length of ICU stay was 21 ± 18 d in Group 1 versus 21 ± 15 d in Group 2 (p = 0.9). It was 22 ± 23 d versus 19 ± 12 d (p = 0.52) and 23 ± 18 d versus 18 ± 7 d (p = 0.24) in survivors and nonsurvivors of Group 1 and 2, respectively. The total length of mechanical ventilation was 20 ± 24 d versus 19 ± 15 d (p = 0.84). The observed crude mortality rate within 30 d after diagnostic evaluation in the whole population was 32 of 76 (42%). It was 18 of 39 (46%) in Group 1 versus 14 of 37 (38%) in Group 2 (p = 0.46). There were also no significant differences with regard to death attributable to pneumonia and adjusted mortality (Table 5).
| Mortality | Group 1 | Group 2 | p Value | |||
|---|---|---|---|---|---|---|
| Crude | 18/39 (46%) | 14/37 (38%) | 0.46 | |||
| Attributable to pneumonia | 10/18 (56%) | 10/14 (71%) | 0.36 | |||
| Adjusted | 16% | 11% | 0.53 | |||
| According to APACHE II | ||||||
| < 20 | 6/19 (32%) | 5/18 (28%) | 0.80 | |||
| = 20 | 12/20 (60%) | 8/19 (42%) | 0.26 | |||
| Early-onset pneumonia | 3/11 (27%) | 0/4 (0%) | 0.52 | |||
| Late-onset pneumonia | 15/28 (54%) | 14/33 (42%) | 0.38 | |||
| Microbiologically confirmed pneumonia | ||||||
| (in respiratory secretion samples | 10/23 (44%) | 10/23 (44%) | 1.0 | |||
| and/or blood cultures) | ||||||
| Respiratory secretion samples | ||||||
| Growth of PPMs | ||||||
| Significant growth | 8/20 (40%) | 10/22 (46%) | 0.72 | |||
| Nonsignificant growth | 1/2 (50%) | 1/8 (13%) | 0.38 | |||
| Regardless of colony counts | 9/22 (41%) | 11/30 (37%) | 0.76 | |||
| Community-acquired pathogens | ||||||
| Significant growth in culture | 1/4 (25%) | 1/4 (25%) | 1.0 | |||
| Nonsignificant growth in culture | — | 0/1 (0%) | — | |||
| Regardless of colony counts | 1/4 (25%) | 1/5 (20%) | 1.0 | |||
| PDRM | ||||||
| Significant growth in culture | 6/16 (38%) | 7/16 (44%) | 0.72 | |||
| Nonsignificant growth in culture | 1/2 (50%) | 2/8 (25%) | 1.0 | |||
| Regardless of colony counts | 7/18 (39%) | 9/24 (38%) | 0.93 | |||
| Only growth in culture of non-PMMs | 9/16 (56%) | 3/7 (43%) | 0.67 | |||
| Negative culture | 4/9 (44%) | 1/3 (33%) | 1.0 | |||
| Initial empiric antimicrobial treatment for VAP | ||||||
| Adequate | 5/15 (33%) | 6/13 (46%) | 0.49 | |||
| Inadequate | 5/8 (63%) | 4/10 (40%) | 0.63 |
Mortality in the whole population was higher in patients with an APACHE II score above the median (cutoff = 20) (11 [30%] versus 20 [51%] nonsurvivors, p = 0.06). A similar trend was true in Group 1 (6 [32%] versus 12 [60%] nonsurvivors, p = 0.08), but not for Group 2 (5 [28%] versus 8 [42%] nonsurvivors, p = 0.34). Mortality according to APACHE II scores was not significantly different when both groups were compared (Table 5).
Mortality was also not significantly different when early-onset and late-onset pneumonia were compared in both groups (Table 5). With regard to patients with microbiologically confirmed pneumonia, there also were no differences with regard to outcome variables. The total length of ICU stay was 24 ± 24 d versus 21 ± 16 d (p = 0.65) and of mechanical ventilation 23 ± 32 d versus 18 ± 16 d (p = 0.55). Mortality rates were 10 of 23 (44%) in both groups (p = 1.0). Likewise, both groups had comparable mortality rates with regard to community-acquired pathogens and PDRM (Table 5). There were also no significant differences when comparing mortality in the subgroup of patients with P. aeruginosa (four of 13 [31%] in group 1 versus five of 10 [50%] in Group 2, p = 0.42). The same was true for both groups with exclusively non-PPMs in culture and negative cultures.
Finally, when analyzing mortality according to the adequacy of initial empiric antimicrobial treatment, there were no differences in mortality between patients with adequate as compared with inadequate treatment (11 of 28 [39%] versus nine of 18 [50%] nonsurvivors, p = 0.47). This was also true when comparing the effect of adequate and inadequate antimicrobial treatment in both groups (Table 5).
The costs for each procedure were as follows: microbiological processing US $22, susceptibility testing US $10, fiberoptic bronchoscopy US $217, PSB and BAL US $65 each. Thus, the total cost for noninvasive investigation was US $28.9 ± 6.9 for noninvasive as compared with US $367.9 ± 26 for invasive investigation (p < 0.0001).
This study had the following three main results: First, the diagnostic yield on noninvasive and invasive techniques for VAP was similar (Group 1: 51%; Group 2: 60%). Antimicrobial treatment before microbial investigation was the only independent factor inversely associated with significant growth in culture. Second, important clinical outcome variables such as length of ICU stay and mechanical ventilation, as well as crude and adjusted 30-d mortality were not significantly different in both groups, also when mortality was adjusted for APACHE II, microbiologically confirmed pneumonia, early- versus late-onset pneumonia, the presence of grouped causative microorganisms, and adequate versus inadequate initial antimicrobial treatment. Third, cost for microbial investigation was significantly higher using invasive techniques.
Although the performance of noninvasive and invasive techniques have varied considerably in different studies, no technique could consistently be shown to achieve a superior diagnostic yield as compared with another (8, 9, 11, 15). Accordingly, in our present study, quantitative tracheobronchial sampling and bronchoscopy including PSB and BAL achieved a very similar yield with regard to significant growth in culture. The number of cultures with significant growth was clearly dependent on the presence of prior antimicrobial treatment, with the highest numbers in patients without antimicrobial treatment and the lowest in patients with a recently introduced antimicrobial treatment within the last 72 h before microbial investigation. The dependence of significant growth in culture from prior antimicrobial treatment was also evident when early- and late-onset pneumonia were considered separately. Accordingly, only prior antimicrobial treatment was inversely associated with significant growth in culture. These findings corroborate those of others who noted that prior antimicrobial treatment most significantly influences the diagnostic yield when recently introduced within 72 h or 24 h (13, 32). Thus, the diagnostic yield for the evaluation of pneumonia was independent of the diagnostic technique used.
Mortality rates were not different in either group. This was also true for mortality rates in both groups with microbiologically confirmed pneumonia, community-acquired pathogens, PDRMs, P. aeruginosa, nonsignificant amounts of exclusively PPMs, exclusively non-PPMs in any amount, negative culture results, and late-onset pneumonia. The only slight imbalance in mortality was obvious for early-onset pneumonia, which obviously is due to the limited sample size of these subgroups (n = 15). Likewise, length of ICU stay and mechanical ventilation were also not influenced by diagnostic testing. Consistent with our previous pilot study which also did not show significant differences with regard to mortality, length of ICU stay, and mechanical ventilation (25), our data confirm that diagnostic techniques had no influence on outcome of VAP. There is now further evidence in support of the use of noninvasive techniques for the initial management of VAP. Because noninvasive methods are clearly less expensive, have less side effects, and do not show differences in outcome compared with invasive techniques, there is no further rationale not to support noninvasive techniques in routine practice.
This is also true in face of a preliminary communication of a French multicenter randomized study reporting an improved outcome in the group investigated by an invasive diagnostic approach (33). In this study, quantitative processing was restricted to invasive respiratory tract secretion samples, and positive qualitative tracheobronchial aspirate samples were handled as if they would represent positive quantitative results. Thus, the study design does not allow a comparison of optimally applied invasive and noninvasive diagnostic methods. Rather, it compares a modern (invasive and quantitative) with a conventional (noninvasive and qualitative) approach. Finally, the interpretation of positive qualitative tracheobronchial culture results as applied in this study probably does not reflect routine practice in most ICU settings, rendering even this comparison of questionable value. Overall, the value of qualitative tracheobronchial aspirate sampling as compared with quantitative sampling is not settled.
In addition, to our previous study (25), important additional conclusions can be drawn. The present study is sound evidence that the use of invasive diagnostic techniques does not reduce the mortality associated with VAP from an estimated 40% to a desired 10%. A reduction of mortality of less than 75% would have required considerably larger population sizes, which probably can only be recruited in mulitcenter designs. Multicenter studies, however, are potentially liable to considerable biases owing to differences in settings and diagnostic procedures. Thus, unicenter studies form an important part of research addressing the impact of diagnostic techniques on outcome even when statistical power is lower than in future multicenter studies. The amount of community- acquired pathogens and PDRMs (including P. aeruginosa) was well balanced in both groups. Therefore, the results of outcome analysis are also valid with regard to potentially causative microorganisms of VAP. Patients received initial empirical antimicrobial treatment for VAP according to a standardized algorithm as suggested by the ATS guidelines. Thus, our study avoided potential bias of outcome measures due to nonstandardized initial antimicrobial treatment approaches. Applying this treatment approach, we no longer found differences in failure of initial antimicrobial treatment and the proportions of changes in antimicrobial treatment with regard to the diagnostic technique used.
Changes in antimicrobial treatment were also evaluated with regard to results from ICO counts in BALF. Modification of antimicrobial treatment was directed by ICO counts in only five patients and ICO counts were not found to have an impact on outcome variables. The study included a standardized reevaluation and antimicrobial treatment also in patients with failure to initial antimicrobial treatment. These patients underwent repeated microbial investigation with the same technique to which they were initially assigned. Thus, outcome measures were also controlled for potential bias due to follow-up investigations and nonstandardized secondary antimicrobial treatment approaches. We included a comparison of cost for the evaluation of suspected pneumonia, which demonstrated significantly higher costs using bronchoscopic techniques.
On the other hand, two issues remained unchanged as compared with our pilot study. First, we did not follow the suggestion to withhold antimicrobial treatment in patients with negative or nonsignificant culture results (19). It is well known that false-negative culture results may result from sampling errors of prior antimicrobial treatment. As we could show in a recent study, an algorithm guiding antimicrobial treatment exclusively by microbiological results does not increase the overall diagnostic accuracy but involves the risk of undertreatment, and, as a consequence, of death from pneumonia (34). Therefore, in our view, it would be unethical to guide antimicrobial treatment decisions exclusively on quantitative culture results above the threshold. Instead, antimicrobial treatment decisions were made according to a balanced judgement including clinical and microbiological criteria by the attending physician. Second, the majority of patients had prior antimicrobial treatment for other reasons than VAP. Although it would be interesting to perform a study exclusively on patients without antimicrobial pretreatment, such a population clearly would represent a minority also in ICU routine practice, and, therefore, the results would not be applicable to the general population.
Overall crude mortality within 30 d after microbial investigation for suspected pneumonia was 41% and fit well with pretest assumptions. On the other hand, adjusted mortality from VAP in our study was 16% in Group 1 and 11% in Group 2, which is in the lower range of the expected. These numbers maybe explained by a strict adherence to the treatment guidelines of the ATS for nosocomial pneumonia, thus reducing the proportion of inappropriate antimicrobial treatment, which in turn had been described as a major adverse prognostic factor of VAP. In fact, in one large multicenter study confirming the adverse prognostic potential of prior inadequate antimicrobial treatment, 67% of patients met this criterion (6). A similar number of 68% was found by others (35). The corresponding numbers in our study were 21% in Group 1 and 27% in Group 2, which is well comparable to the 28% found in our previous study (25) and to the 24% found by other investigators (36). Because inadequate initial antimicrobial treatment is the key target where diagnostic approaches may have a potential of improving the outcome of VAP, its reduction by appropriate policies of initial antimicrobial treatment may also considerably reduce attributable mortality. On the other hand, the appropriateness of these policies strongly depends on the local epidemiology. Accordingly, in a recent cohort study in which initial inadequate antimicrobial treatment and other confounders were controlled, there was no attributable mortality of VAP (37). We only found a nonsignificant trend for excess mortality in patients treated with inadequate antimicrobial treatment prior to the results of microbial investigation (39% versus 50%). Consistently, in our study, mortality in patients with a corresponding inadequate antimicrobial treatment was not significantly different. Thus, the potential impact of any diagnostic technique on outcome variables strongly depends on the appropriateness of initial empiric antimicrobial treatment policies. This should be kept in mind when comparing results of studies designed to compare outcome variables with regard to different diagnostic techniques.
In conclusion, the main inverse predictor of significant growth in culture was prior antimicrobial treatment. Length of ICU stay and mechanical ventilation as well as mortality were not significantly influenced by diagnostic techniques, and the present data make it improbable that even small differences in outcome can be detected, including considerably higher patient numbers. Thus, current evidence supports noninvasive microbial investigation as the principal approach to microbial investigation in suspected VAP.
Supported by Commisionat per a Universitats i Recerca de la Generalitat de Catalunya 1997 SGR 00086, and IDIBAPS Hospital Clinic Barcelona FIS 98/0138 and SEPAR 503.
| 1. | Torres A., Aznar R., Gatell J. M., Jimenez P., Gonzalez J., Ferrer A., Celis R., Rodriquez-Roisin R.Incidence, risk, and prognostic factors of nosocomial pneumonia in mechanically ventilated patients. Am. Rev. Respir. Dis.1421990523528 |
| 2. | Rello J., Quintana E., Ausina V., Castella J., Luquin M., Net A., Prats G.Incidence, etiology, and outcome of nosocomial pneumonia in mechanically ventilated patients. Chest1001991439444 |
| 3. | Fagon J. Y., Chastre J., Vuagnat A., Trouillet J. L., Novara A., Gibert C.Nosocomial pneumonia and mortality among patients in intensive care units. J.A.M.A.2751996866869 |
| 4. | Fagon J. Y., Chastre J., Hance A. J., Montravers P. H., Novara A., Gibert C.Nosocomial pneumonia in ventilated patients: a cohort study evaluating attributable mortality and hospital stay. Am. J. Med.941993281288 |
| 5. | Kollef M., Silver P., Murphy D. M., Trovillion E.The effect of late-onset pneumonia in determining patient mortality. Chest108199516551662 |
| 6. | Alvarez-Lerma F.the ICU-acquired Pneumonia Study GroupModification of empiric antibiotic treatment in patients with pneumonia acquired in the intensive care unit. Intensive Care Med.221996387394 |
| 7. | Rello J., Ausina V., Ricart M., Castella J., Prats G.Impact of previous antimicrobial therapy on the etiology and outcome of ventilator-associated pneumonia. Chest104199312301235 |
| 8. | El-Ebiary M., Torres A., González J., Puig De La J., Bellacasa, Garcia C., Jimenez De Anta M. T., Ferrer M., Rodriguez-Roisin R.Quantitative cultures of endotracheal aspirates for the diagnosis of ventilator-associated pneumonia. Am. Rev. Respir. Dis.148199315521557 |
| 9. | Marquette C. H., Georges H., Wallet F., Ramon P., Saulnier F., Neviere R., Mathieu D., Rime A., Tonnel A. B.Diagnostic efficiency of endotracheal aspirates with quantitative bacterial cultures in intubated patients with suspected pneumonia. Am. Rev. Respir. Dis.1481993138144 |
| 10. | Rouby J. J., Martin De Lassale E., Poete P., Nicolas M. H., Bodin L., Jarlier V., Le Charpentier Y., Grosset J., Viars P.Nosocomial bronchopneumonia in the critically ill. Am. Rev. Respir. Dis.146199210591066 |
| 11. | Torres A., El-Ebiary M., Padró L., Gonzalez J., Puig De La J., Bellacasa, Ramirez J., Xaubet A., Ferrer M., Rodriguez-Roisin R.Validation of different techniques for the diagnosis of ventilator-associated pneumonia. Am. J. Respir. Crit. Care Med.1491994324331 |
| 12. | Marquette C. H., Copin M. C., Wallet F., Nevierte R., Saulnier F., Mathieu D., Durocher A., Ramonn P., Tonnel A. B.Diagnostic tests for pneumonia in ventilated patients: prospective evaluation of diagnostic accuracy using histology as a diagnostic gold standard. Am. J. Respir. Crit. Care Med.151199518781888 |
| 13. | Chastre J., Fagon J. Y., Bornet-Lesco M., Calvat S., Dombret M. C., Ah R., Khani, Basset F., Gibert C.Evaluation of bronchoscopic techniques for the diagnosis of nosocomial pneumonia. Am. J. Respir. Crit. Care Med.1521995231240 |
| 14. | Papazian L., Thomas P., Garbe L., Guignon I., Thirion X., Charrel J., Bollet C., Fuentes P., Gouin F.Bronchoscopic or blind sampling techniques for the diagnosis of ventilator-associated pneumonia. Am. J. Respir. Crit. Care Med.152199519821991 |
| 15. | Kirtland S. H., Corley D. E., Winterbauer R. H., Springmeyer S. C., Casey K. R., Hampson N. B., Dreis D. F.The diagnosis of ventilator-associated pneumonia: a comparison of histologic, microbiologic, and clinical criteria. Chest1121997445457 |
| 16. | Niederman M. S., Torres A., Summer W.Invasive diagnostic testing is not needed routinely to manage suspected ventilator-associated pneumonia. Am. J. Respir. Crit. Care Med.1501994565569 |
| 17. | Chastre J., Fagon J. Y.Invasive diagnostic testing should be routinely used to manage ventilated patients with suspected pneumonia. Am. J. Crit. Care Med.1501994570574 |
| 18. | Corley D. E., Kirtland S. H., Winterbauer R., Hammar S. P., Dail D. H., Bauermeister D. E., Bolen J. W.Reproducibility of the histologic diagnosis of pneumonia among a panel of for pathologists: analysis of a gold standard. Chest1121997458456 |
| 19. | Wunderink R.Mortality and the diagnosis of ventilator-associated pneumonia: a new direction. Am. J. Respir. Crit. Care Med.1571998349350 |
| 20. | Torres A., El-Ebiary M.Diagnostic approaches and hospital-acquired pneumonia. Sem. Respir. Crit. Care Med.181997149161 |
| 21. | Baker A. M., Bowton D. L., Haponik E. F.Decision making in nosocomial pneumonia: an analytic approach to the interpretation of quantitative bronchoscopic cultures. Chest10719958595 |
| 22. | Sterling T. R., Ho E. J., Brehm W. T., Kirkpatrick M. B.Diagnosis and treatment of ventilator-associated pneumonia–impact on survival. Chest110199610251034 |
| 23. | Timsit J. F., Chevret S., Valcke J., Misset B., Renaud B., Goldstein F. W., Vaury P., Carlet J.Mortality of nosocomial pneumonia in ventilated patients: influence of diagnostic tools. Am. J. Respir. Crit. Care Med.1541996116123 |
| 24. | Bregeon R., Papazian L., Visconti A., Gregoire R., Thirion X., Gouin F.Relationship of microbiologic diagnostic criteria to morbidity and mortality in patients with ventilator-associated pneumonia. J.A.M.A.2771997655662 |
| 25. | Sanchez-Nieto J. M., Torres A., Garcia-Cordoba F., El-Ebiary M., Carillo A., Ruiz M., Nunez M. L., Niederman M. S.Impact of invasive and noninvasive quantitative culture sampling on outcome of ventilator-associated pneumonia: a pilot study. Am. J. Respir. Crit. Care Med.1571998371376 |
| 26. | American Thoracic SocietyHospital-acquired pneumonia in adults: diagnosis, assessment of severity, initial antimicrobial therapy, and preventive strategies. Am. J. Respir. Crit. Care Med.153199517111725 |
| 27. | Murray J. F., Mattay M.A., Luce J. M., Flick M. R.An expanded definition of the adult respiratory distress syndrome. Am. Rev. Respir. Dis.1381988720723 |
| 28. | Marshall J. C., Christou N. V., Horn R., Meakins J. L.The microbiology of multiple organ failure. Arch. Surg.1231988309314 |
| 29. | Balows, A., W. J. Hausler, Jr., K. L. Herrmann, H. D. Isenberg, and H. S. Shadomy, editors. 1991. Manual of Clinical Microbiology, 5th ed., section III. American Society for Microbiology, Washington, DC. |
| 30. | Cabello H., Torres A., Celis R., El-Ebiary M., Puig de la Bellacasa J., Xaubet A., Gonzalez J., Agusti C., Soler N.Distal airway bacterial colonisation in healthy subjects and chronic lung diseases: a bronchoscopic study. Eur. Respir. J.10199711371144 |
| 31. | Knaus W. A., Draper E. A., Wagner D. P., Zimmerman J. E.APPACHE II: a severity of disease classification system. Crit. Care Med.131985818829 |
| 32. | Souweine B., Veber B., Bedos J. P., Gachot B., Dombret M. C., Regnier B., Wolff M.Diagnostic accuracy of protected specimen brush and bronchoalveolar lavage in nosocomial pneumonia: impact of previous antimicrobial treatments. Crit. Care Med.261998236244 |
| 33. | Fagon Y. I., Chastre J.on behalf of the VAP Trial GroupComparison of two diagnostic strategies in patients with a clinical suspicion of ventilator-associated pneumonia (VAP) (abstract). Am. J. Respir. Crit. Care Med.1591999A518 |
| 34. | Fabregas, N., S. Ewig, A. Torres, M. El-Ebiary, J. Ramirez, J. Puig de la Bellacasa, T. Bauer, and H. Cabello. 1999. Clinical diagnosis of ventilator-associated pneumonia revisited: comparative evaluation using immediate postmortem biopsies. Thorax 54:867–873, |
| 35. | Luna C., Vujachich P., Niederman M. S., Gherardi C., Matera J., Jolly E. C.Impact of BAL data on the therapy and outcome of ventilator-associated pneumonia. Chest1111997676687 |
| 36. | Rello J., Gallego M., Mariscal D., Sonora R., Valles J.the value of routine microbial investigation in ventilator-associated pneumonia. Am. J. Respir. Crit. Care Med.1561997196200 |
| 37. | Papazian L., Bregeon F., Thirion X., Gregoire R., Saux P., Denis J. P., Perin G., Charrel J., Dumon J., Affay J. P., Gouin F.Effect of ventilator-associated pneumonia on mortality and morbidity. Am. J. Respir. Crit. Care Med.15419969197 |
