Ventilator-associated tracheobronchitis (VAT) is common among mechanically ventilated critically ill patients 123. Previous studies found VAT to be associated with increased duration of mechanical ventilation and intensive care unit (ICU) stay 145. VAT is probably an intermediate process between lower respiratory tract colonization and ventilator-associated pneumonia (VAP). Postmortem studies showed a continuum between bronchitis and pneumonia in mechanically ventilated ICU patients 6.

Few studies have evaluated the impact of antibiotic treatment on the outcome of critically ill patients with VAT 145. In a prospective observational study 1, our group investigated the impact of antibiotic treatment on the outcome of patients with VAT. Among the 201 patients with VAT, 136 received antibiotics. The mortality rate was significantly lower in VAT patients who received antibiotics than in those who did not receive antibiotics. However, after exclusion of VAT patients who developed subsequent VAP, no significant difference was found in mortality rate. A beneficial effect of antimicrobial treatment on the duration of mechanical ventilation was also suggested by an observational case-control study performed in chronic obstructive pulmonary disease (COPD) patients with VAT 4. However, another case-control study performed in VAT patients without chronic respiratory failure found no impact of antimicrobial treatment on the duration of mechanical ventilation 5. Furthermore, it has been shown that systemic antibiotics have no effect on the transition from VAT to VAP 1.

Although no firm evidence on the beneficial effects of antibiotic treatment in patients with VAT exists, ICU physicians frequently treat these patients with antibiotics 78910. However, excessive usage of antibiotics in the ICU is associated with the subsequent emergence of multidrug-resistant (MDR) bacteria and worse outcome 11121314. In their recent guidelines 15, the American Thoracic Society (ATS) and the Infectious Disease Society of America recommended the performance of randomized studies to determine whether patients with VAT should be treated with antibiotics. Therefore, we conducted this prospective, randomized, controlled, multicenter study to determine the impact of antimicrobial treatment on outcome in VAT patients.

The study was conducted in 12 ICUs in the north of France from June 2005 to June 2007. The study protocol was approved by the institutional review board on human research of the Lille university hospital. All patients or their next of kin gave written informed consent before enrolment in the study.

The eligibility criteria for the study were age older than 18 years and the presence of a first episode of VAT diagnosed more than 48 hours after starting mechanical ventilation. Before random assignment, patients were excluded if they (a) were pregnant, (b) had a history of severe immunosuppression, (c) had a tracheostomy at ICU admission (however, patients were eligible if they had a tracheostomy performed after ICU admission), (d) had a VAP before VAT, (e) had already participated in this study, (f) were already included in another trial, or (g) had little chance of survival as defined by a Simplified Acute Physiology Score (SAPS II) of greater than 65 points.

Sixty-five patients were eligible for this study. Seven patients refused to participate. Fifty-eight patients were randomly assigned, including 22 patients in the antibiotic group and 36 patients in the no antibiotic group. Fourteen patients were excluded from the modified ITT analysis (4 of 22 [18%] versus 10 of 36 [27%], P = 0.533, in the antibiotic and no antibiotic groups, respectively). Among the 14 excluded patients, 8 patients were excluded for do-not-resuscitate orders (4 of 22 [18%] versus 4 of 36 [11%], P = 0.462) and 6 patients were excluded because they were randomly assigned to the no antibiotic group but received antibiotics for infections other than VAT or subsequent VAP (5 bacteremia and 1 severe sepsis) during the 8 days following random assignment. No patient was excluded for impossible weaning from mechanical ventilation (Figure 1).

The planned interim analysis was performed 2 years after starting the study because the number of included patients was less than 175. The study was stopped by the local institutional review board and safety committee because the interim analysis found a significant difference in ICU mortality.

Patient characteristics were similar at ICU admission and at the day of random assignment (Tables 1 and 2). Patients with community-acquired pneumonia at ICU admission had all completed antibiotic treatment for community-acquired pneumonia before inclusion in the study.

The main results of our study are the following: (a) In patients with VAT, antibiotic treatment was associated with significantly lower ICU mortality and subsequent VAP rates and more mechanical ventilation-free days. (b) No significant difference was found in the rate of infection or colonization related to MDR bacteria diagnosed after random assignment between the two groups. (c) No significant difference was found in the total duration of mechanical ventilation or ICU stay between the antibiotic and no antibiotic groups.

To our knowledge, this is the first randomized study aiming at evaluating the impact of antibiotic treatment on the outcome of patients with VAT. The beneficial effect of antibiotics found in this study on the number of days free of mechanical ventilation could be explained by the reduction of secretion volume and tracheobronchial inflammation. Palmer and colleagues 2324 investigated the impact of aerosolized antibiotics on secretion volume in chronically mechanically ventilated patients with VAT. In those studies, aerosolized antibiotics eradicated respiratory pathogens, decreased inflammatory cells and the volume of secretions, and were not associated with increased resistance. Increased secretion volume is a well-known risk factor for difficult weaning from mechanical ventilation 25. However, these factors were not evaluated in our study. The absence of a significant difference in the total duration of mechanical ventilation is probably related to the small number of patients included in the study as compared with the number of patients required to demonstrate a significant difference. However, the number of days free of mechanical ventilation was significantly higher in the antibiotic group. This could be explained by the fact that the mortality rate was significantly higher in patients in the no antibiotic group and by the longer duration of mechanical ventilation before random assignment in the antibiotic group.

Lower rates of VAP and ICU mortality were found in VAT patients who received antimicrobial treatment. Similar results were found in a recent randomized study conducted in COPD patients mechanically ventilated for severe acute exacerbation 26. However, in that study, all included patients had community-acquired bronchitis. In addition, no bacteria could be found in 38% of included patients. Although the severity of illness and predicted mortality were similar in the two groups, mortality rate was significantly higher in the control group. This result is probably related not to VAT but to the higher rate of VAP in control patients. In addition, all VAP episodes were late-onset and the rate of P. aeruginosa VAP was high. Previous studies demonstrated that VAP was associated with increased mortality rate 2728. A recent study found higher mortality rates in patients with late-onset VAP as compared with patients with early-onset VAP 28. P. aeruginosa VAP was also found to be associated with high mortality rates 29. However, other studies suggested that VAP was not associated with an increased mortality rate 3031. Another potential explanation for the higher mortality rate in untreated patients is the possible presence of pneumonia in these patients. VAT may be difficult to differentiate from VAP because of the low sensitivity of chest portable radiographs in ICU patients 3233. Though not statistically significant, the duration of mechanical ventilation from random assignment to VAP occurrence was shorter in the no antibiotic group than in the antibiotic group. This result suggests that VAP might have been present at the time of random assignment despite the absence of new infiltrate on the chest radiograph. In a prospective, observational, multicenter, cohort study performed on 2,706 patients, outcomes of patients with suspected pneumonia and normal chest radiographs (33%) have been prospectively investigated 34. Similar rates of positive sputum cultures, positive blood cultures, and mortality were found in patients without radiographic pneumonia as compared with patients with radiographic pneumonia. In a recent study 35, accuracy of chest radiography was compared with high-resolution computed tomography (HRCT) in 47 patients with suspected community-acquired pneumonia. HRCT identified all 18 community-acquired pneumonia cases (38%) apparent on radiographs as well as 8 additional cases (17%). The performance of HRCT could be suggested to better diagnose VAP in critically ill patients. However, recent guidelines require the presence of new infiltrate on a chest radiograph as a criterion for VAP diagnosis 15. Therefore, a baseline examination should be available for all patients to diagnose a new infiltrate on HRCT. Such a strategy would be expensive and difficult to apply in critically ill patients. The absence of new infiltrate on a chest radiograph could be more difficult to diagnose in patients with an abnormal chest radiograph at ICU admission. In our study, 38% of study patients had an abnormal chest radiograph at ICU admission. However, patients admitted to the ICU frequently have an abnormal chest radiograph 36.

The rate of COPD (44%) was high. However, no significant difference was found in COPD rate between the two groups. A previous observational study identified COPD as a risk factor for VAT 1. The rate of patients with multiple organ failure was significantly higher in the control group than in the antibiotic group. This result could be explained by the higher rate of VAP in these patients. Previous studies found VAP to be associated with multiple organ failure 3738.

VAT could also be difficult to differentiate from lower respiratory tract colonization. Several factors support the presence of infection rather than colonization in our patients: (a) Quantitative endotracheal aspirate was used with a high threshold (10⁶ cfu/mL) to diagnose VAT, (b) only new bacteria were taken into account, (c) all patients had fever, and (d) leucocyte, C-reactive protein, and procalcitonin levels were high in study patients. Although fever and high leucocyte and C-reactive protein levels may simply reflect the presence of systemic inflammatory response, procalcitonin is useful in differentiating bacterial sepsis from systemic inflammatory response in critically ill patients 394041. However, the exclusion of pathogens present at the time of intubation could be a matter of debate since these pathogens could be responsible for VAT. In addition, microorganisms cultured at a lower concentration (<10⁶ cfu/mL) might be associated with VAT 2. On the other hand, one could argue that patients with a high microorganism count on tracheal aspirate cultures and no radiographic infiltrates should be treated with antibiotics. However, stable patients receiving prolonged mechanical ventilation without clinical pneumonia have a high alveolar burden of bacteria 42. Therefore, the presence of purulent tracheal aspirate and fever is important to determine patients who would benefit from antimicrobial treatment.

This study has some limitations. First, the trial stopped early after the planned interim analysis showed a significant reduction of ICU mortality rate in the antibiotic group. Therefore, several random assignment blocks could not be ended, resulting in an imbalance in the numbers of patients randomly assigned to the antibiotic or control group. One could argue that no significant difference was found in the primary endpoint. However, the significant difference in ICU mortality, subsequent VAP, and mechanical ventilation-free days represents overwhelming evidence of benefit to justify stopping the trial early. In addition, this endpoint was no longer relevant given the difference in ICU mortality. Second, the study was not blinded and antibiotic treatment was not standardized in all treated patients. However, blinding was not possible using a targeted antibiotic strategy based on results of previous endotracheal aspirate culture. The aim of such a strategy was to reduce the usage of broad-spectrum antibiotics and to provide a higher rate of appropriate initial antibiotic treatment. A recent study found routine surveillance endotracheal aspirate useful to prescribe appropriate antibiotic therapy in patients with VAP 43. Furthermore, the results of our study were not evaluated by an independent committee to account for the absence of blinding. However, ICU mortality was significantly different between the two groups. This outcome does not need to be assessed by a committee blinded to patient assignment. Third, the number of patients screened for this study could not be provided and the number of included patients was lower than initially expected. Potential reasons for this slow recruitment include strict inclusion and exclusion criteria and difficulties in differentiating VAT from VAP. Fourth, 21 of 36 (58%) patients randomly assigned to the no antibiotic group had received antibiotics during the 8 days following random assignment, including 15 patients (41%) for subsequent VAP and 6 patients (16%) for infections other than VAT or subsequent VAP. However, subsequent VAP was a secondary endpoint. In addition, to adjust for this confounding factor, we have performed a modified ITT analysis excluding those patients randomly assigned to the no antibiotic group but who received antibiotics for infections other than VAT or subsequent VAP. Fifth, a computed tomography scan was not systematically performed to search for nosocomial sinusitis. In addition, no information could be provided on hospital mortality, oral care, and type of nutrition. Finally, because of the small sample size, a type I error could not be excluded. However, the significant difference found in ICU mortality is probably related to the significant difference in subsequent VAP rates between the two groups.

We conclude that, in patients with VAT, antimicrobial treatment is associated with a greater number of days free of mechanical ventilation and lower rates of VAP and ICU mortality. However, antibiotic treatment has no significant impact on the total duration of mechanical ventilation or ICU stay.

• In patients with ventilator-associated tracheobronchitis, antibiotic treatment is associated with significantly lower intensive care unit (ICU) mortality and subsequent ventilator-associated pneumonia rates and more mechanical ventilation-free days.

• In these patients, antibiotic treatment has no significant impact on the total duration of mechanical ventilation or ICU stay.

ATS = American Thoracic Society; cfu = colony-forming units; COPD = chronic obstructive pulmonary disease; HRCT = high-resolution computed tomography; ICU = intensive care unit; ITT = intention-to-treat; MDR = multidrug-resistant; VAP = ventilator-associated pneumonia; VAT = ventilator-associated tracheobronchitis.

The authors declare that they have no competing interests.

SN helped to design the study and to collect data, had full access to all data in the study, wrote the manuscript, and had final responsibility for the decision to submit it for publication. DM and AD helped to design the study and to collect data. RF, EJ, F Decamps, F Dewavrin, and GB helped to collect data. CDP performed statistical analyses. All authors participated in critical revision of the manuscript. All authors read and approved the final manuscript.