Ventilator-associated pneumonia is a clear risk factor for acute lung injury which has been poorly described in trauma patients. This prospective study was undertaken to estimate the incidence of such ventilator-associated pneumonia leading to acute lung injury, the risk factors, and the associated morbidity and mortality in a group of multiple trauma patients.
Trauma patients who were mechanically ventilated and survived at least 24 h were included. Ventilator-associated pneumonia was confirmed by a bacterial culture of a blind protected telescoping catheter with at least 10 colony-forming units/ml of at least one pathogen. Episodes of acute lung injury were prospectively recorded.
Ventilator-associated pneumonia was documented in 78 patients of the 175 included (44%) and led to the development of ventilator-associated pneumonia acute lung injury in 18 patients (23%). The sole independent risk factor for ventilator-associated pneumonia leading to acute lung injury was the presence of Haemophilus influenzae (hazard ratio, 8.8; 95% confidence interval, 2.7-28.6). Eleven (61%) of the 18 patients with ventilator-associated pneumonia leading to acute lung injury had development of a ventilator-associated pneumonia recurrence, as compared with 20 (33%) of the 60 patients with ventilator-associated pneumonia alone (P = 0.03). Seven (39%) of the 18 trauma patients with ventilator-associated pneumonia leading to acute lung injury died, as compared with 9 (15%) of the 60 trauma patients with ventilator-associated pneumonia alone (P = 0.04).
Acute lung injury complicated the course of 15% of ventilator-associated pneumonia in trauma patients. H. influenzae seemed to be one of the most frequent bacteria involved and the sole risk factor identified. Occurrence of ventilator-associated pneumonia leading to acute lung injury modified the prognosis of trauma patients.
VENTILATOR-ASSOCIATED pneumonia (VAP) continues to complicate the course of 22–50% of trauma patients receiving mechanical ventilation.1–4A large majority of such VAP cases are early onset.1,4For most studies, VAP occurring in trauma patients did not lead to an increase in attributable mortality.2–5However, VAP is a clear clinical risk factor for the development of acute lung injury (ALI)/acute respiratory distress syndrome (ARDS).6,7In fact, little is known about VAP leading to ALI/ARDS (VAP-ALI) in trauma patients. One of the possible reasons is the difficulty in determining whether VAP developed as a result of ALI/ARDS or whether the syndrome facilitated development of VAP. The only information available is the incidence of VAP-ALI, which ranged between 7 and 24%, depending on the definition of VAP used.8–10Therefore, more information regarding the possible risk factors involved and the related morbidity and mortality could be of value for clinicians in the treatment of such complicated patients.
The aim of this study was to estimate the incidence of such VAP-ALI using sampling of distal airways, the risk factors, and the associated morbidity and mortality in a group of multiple trauma patients.
Materials and Methods
Trauma patients who were mechanically ventilated and survived at least 24 h between January 2002 and January 2004 in the trauma intensive care unit (ICU) were included in the study. Data were collected prospectively for each patient. The severity of illness was evaluated with the first-day Simplified Acute Physiology Score II11and the Organ Dysfunction and/or Infection score,12based on the presence or absence of cardiac, respiratory, renal, hepatic, neurologic, and hematologic dysfunctions and/or infection. Among injury severity scoring systems, the Injury Severity Score,13the Revised Trauma Score,14and the Trauma Injury Severity Score15were determined. The Glasgow Coma Score was reported by the attending physician at arrival on the scene. Various other variables were recorded: age, sex, site of major injury and specific injury, need for transfusion during the first 3 days after ICU admission, duration of mechanical ventilation, duration of stay, and mortality in the ICU. Attention was particularly paid to a history of sleep apnea syndrome, asthma, or chronic obstructive pulmonary disease. Clinical diagnosis criteria for such chronic lung diseases were obtained from general practitioner recordings. Patient clinical status was assessed daily in the ICU, and episodes of ALI/ARDS and VAP were prospectively recorded. Outcome measures included ICU mortality, microbiologically documented pulmonary infection recurrence, and duration of mechanical ventilation.
This study was approved by our hospital ethics committee (Comité de Protection des Personnes se Prêtant à la Recherche Biomédicale Henri Mondor, Créteil, France). Because this study was only observational and did not modify current diagnostic and therapeutic strategies, authorization was given to waive informed consent. All data used in subsequent analyses were anonymous.
We did not use selective digestive decontamination or continuous aspiration of subglottic secretions. All patients were given sucralfate, ranitidine, or omeprazole until enteral nutrition was started. Unless the position was contraindicated, all patients were kept semirecumbent (≥ 30°). Low-molecular-weight heparin was the method of antithrombotic prophylaxis used in trauma at our institution.
Ventilation variables were adjusted to keep the plateau pressure 35 cm H2O or less. After positive end-expiratory pressure titration, the positive end-expiratory pressure value that maintained the best oxygenation with the least hemodynamic effect was accepted.16Inspired oxygen fraction (Fio2) was manipulated to maintain an arterial oxyhemoglobin saturation above 90%.
Amoxicillin–clavulanic acid was prescribed during 48 h for patients with open fractures. Antibiotic therapy for VAP lasted 8 days for every patient except when nonfermenting bacilli were isolated (Pseudomonas , Acinetobacter , and Stenotrophomonas maltophilia ). In such cases, patients received antibiotics for 14 days. Amoxicillin–clavulanic acid was prescribed for early-onset pneumonia17associated during 3 days with gentamicin. Antimicrobial was prescribed during 24 h for perioperative prophylaxis according to local guidelines.
Trauma patients were suspected to have VAP when they had development of a new and persistent lung infiltrate and had purulent tracheal secretions. VAP was confirmed by a bacterial culture of a blind protected telescoping catheter with at least 103colony-forming units/ml of at least one pathogen.18
Protected telescoping catheter was performed as soon as VAP recurrence was suspected. Patients were considered to have microbiologically documented recurrent pulmonary infection when at least one bacterial species grew at a significant concentration from a sample collected during a second blind protected telescoping catheter. If at least one of the initial causative bacterial strains (same antibiotic susceptibility) was grown at a significant concentration from a second distal sample, recurrence was considered a persistent infection (during treatment of the previous episode) or a relapse (occurring after the end of antibiotic therapy for the previous episode). Otherwise, VAP recurrence was considered a superinfection.
Microorganisms isolated from pulmonary source were recorded for microbiologic analysis.
Diagnosis of ALI and ARDS was made according to the American-European Consensus Conference definitions19(acute onset, bilateral infiltrates on frontal chest radiograph, arterial oxygen tension [Pao2]/Fio2less than 300 mmHg for ALI and less than 200 mmHg for ARDS, and the absence of clinical evidence of left atrial hypertension).
Acute lung injury/ARDS was considered a complication of VAP when criteria for ALI/ARDS were fulfilled during the next 48 h after the clinical suspicion of VAP, subsequently confirmed by a positive culture of a protected telescoping catheter, without alternative etiology (pulmonary contusion, transfusion-related ALI, sepsis of other origin than lung, aspiration, fat embolism, pancreatitis). When the patient had previously experienced ALI/ARDS, an improvement of pulmonary gas exchanges (Pao2/Fio2> 300) and a resolution of pulmonary infiltration on chest radiograph or computed tomography scan during at least 48 h were mandated before the new episode of ALI could be attributed to the occurrence of VAP. When VAP occurred in patient who already had ALI/ARDS, VAP was considered to complicate the course of ALI/ARDS.
The final diagnosis was determined by two independent experts (F. S. and C. D.). In cases of disagreement between the two experts, a consensus was reached by a third expert (F. D.).
Characterization of Strains of Haemophilus
Determination of serotype B strains was effected by agglutination of latex particles coated with specific antibodies using the Pastorex Meningitis kit (Bio-Rad Laboratories, Hercules, CA).
Susceptibility to antibiotics was tested using a disk diffusion method on HTM agar (Becton Dickinson France, Le Pont-de-Claix, France) according to the guidelines of the Comité de l'Antibiogramme de la Société Française de Microbiologie.20
Data were computerized and analyzed using the Statview 5.0 statistical package (SAS Institute Inc., Cary, NC). Taking into account the preliminary results during the first 6 months of our study and the data previously reported,8–10we hypothesized an incidence around 20%. Therefore, for an incidence with a precision of 10%, 51 patients with VAP must be included. The agreement between the two experts was assessed by a κ measure of agreement. Normality of the distribution of data were assessed by the Kolmogorov-Smirnov test. Normally distributed continuous variables were compared using the Student t test and expressed as mean ± SD. Nonnormally distributed continuous variables were compared using the Mann–Whitney U test and expressed as the median [25th–75th percentiles]. Chi-square tests or the Fisher exact test were used to compare proportions and rates. Important percentages have been associated with their 95% confidence interval (CI). Time-to-event variables were estimated according to the Kaplan-Meier method and were compared by means of the log-rank test. Multivariate modeling in the form of Cox proportional hazard analysis was used to evaluate potential risk factors for VAP-ALI. Variables associated with VAP-ALI in the univariate analysis with a P value less than 0.10 were entered into the model. Hazard ratios and their 95% CI were calculated for all significant predictors. Statistical significance was defined as a two-tailed P value of 0.05 or less.
During the study period, 564 patients were admitted to the ICU, of whom 317 were trauma patients. Among them, 175 met the inclusion criteria, and all were included (fig. 1). The other 142 patients did not meet the inclusion criteria (fig. 1). Main characteristics and other markers of acute illness are shown in table 1. VAP was documented in 78 patients: 44% [95% CI, 37–52%] for a total of 123 episodes (22 patients had two episodes, and 9 patients had three episodes or more). Overall, 65 patients (37%) had 69 episodes of ARDS/ALI during their ICU stay. VAP led to the development of ARDS/ALI in 18 of 78: 23% [95% CI, 14–32%] patients (15 with ARDS). There was agreement between the two experts in 115 cases of VAP (93%). The remaining 8 cases required a third expert (3 in the final group VAP-ALI). The κ value was 0.92. In the 51 remaining cases, ARDS/ALI was related to pulmonary contusion (n = 31; 45%), transfusion-related ALI (n = 11; 16%), and miscellaneous diseases (n = 9; 13%). Episodes of ALI/ARDS related to VAP occurred at a median of 5 [4–8] days compared with 1 [1–1] for other etiologies (P < 0.001).
Clinical Features and Risk Factors for VAP-ALI
A total of 176 microorganisms were cultured at significant concentrations from protected telescoping catheter specimens during the 123 episodes of VAP. Eight VAP-ALI (44%) episodes were polymicrobial, compared with 34 of the 105 episodes (32%) of VAP alone (P = not significant [NS]). In the 18 episodes of VAP-ALI, 15 were related to early-onset VAP, and 3 were related to VAP recurrence. As indicated in table 2, the organisms most frequently isolated from VAP-ALI patients were Haemophilus influenzae (50%), followed by methicillin-sensitive Staphylococcus aureus (15%) and Escherichia coli (11%).
When the distribution patterns of microorganisms responsible for infection in patients with VAP-ALI and with VAP alone were compared, a higher percentage of H. influenzae was evidenced in the former group. Whereas H. influenzae represented only 11% [95% CI, 6–16%] of the total number of causative pathogens in patients with VAP alone, it represented 50% [30–70%] in patients with VAP-ALI (P < 0.001). Identification of serotype b and antibiotic susceptibilities of H. influenzae isolates are summarized in table 3. No significant differences could be found regarding the frequency of serotype b and antibiotic susceptibilities profiles.
Patients with VAP-ALI or VAP alone seemed similar with regard to sex, trauma injuries, and severity of underlying illness. However, patients with VAP-ALI presented more often chronic lung disease and tended to be older (table 1). Four patients with VAP alone versus 0 VAP-ALI patients presented concomitant positive blood culture (P = NS). Seventy-one percent (14 of 18) of all episodes of VAP-ALI included at least one strain of H. influenzae , versus only 15% of VAP alone (P < 0.001). A multivariate Cox model was built taking into account variables univariately associated with VAP-ALI: age, preexisting chronic lung disease, and presence of H. influenzae . The sole independent risk factor for VAP-ALI was the presence of H. influenzae (hazard ratio, 8.8; 95% CI, 2.7–28.6).
The occurrence of VAP did not seem to markedly influence overall survival in trauma patients: 16 of the 78 (20% [95% CI, 11–29%]) trauma patients who had development of VAP died, as compared with 18 of the 97 (18% [95% CI, 10–26%]) trauma patients who did not have development of VAP (P = NS). However, patients with VAP-ALI had a higher mortality rate (39% [95% CI, 17–61%]) as compared with patients with VAP alone (15% [95% CI, 6–24%]) (P = 0.04; table 1).
Eleven (61%) of the 18 patients with VAP-ALI had VAP recurrence, as compared with 20 (33%) of the 60 patients with VAP alone (P = 0.03). The median number of VAP recurrences was 2 [1–3] in the 11 patients with VAP-ALI and 1 [1–1] in the 20 patients with VAP alone (P = 0.04). Among the 22 recurrences in patients with VAP-ALI, 15 were superinfections, 2 were relapses, and 5 were persistent infections, which is similar to the 23 recurrences in patients with VAP alone (19 were superinfections, 4 were relapses, and 1 was persistent infection) (P = NS). The median time for the first pulmonary infection recurrence was 22 [13–31] days for the patients with VAP-ALI, as compared with 17 [12–20] days for the patients with VAP alone (P = NS).
The total duration of mechanical ventilation was significantly longer for patients with VAP-ALI than for those with VAP alone. The corresponding Kaplan-Meier estimates are shown in figure 2.
Our results showed that 23% of trauma patients with VAP had development of a related ALI/ARDS. Such occurrence of VAP-ALI is followed by an increased number of recurrent VAP and a higher ICU mortality. Among the several risk factors studied, the sole independent was the presence of H. influenzae as a causative pathogen.
During their ICU stay, 16–42% of trauma patients experienced ALI/ARDS,7,9,10,21,22of which 75% occurred during the first 3 days after injury.9,10,21,22Such large variations in incidence could be related to the population of trauma patients studied and the severity of trauma. The majority of such episodes of ALI/ARDS have been related to pulmonary contusion, hemorrhagic shock, and multiple transfusions.8,21VAP accounted for 7–24% of ALI/ARDS, depending of the definition for VAP used.8–10
After trauma, 22–50% of the patients had development of VAP defined by a quantitative culture of distal airways specimen.1–4Such variations could be related to the local antibiotic policy and the severity of trauma patients included, because Injury Severity Score was a risk factor identified in several studies.1,23–25The mortality rate of such VAPs ranged between 17 and 22%.2–5,25For most studies and as in ours, VAP occurring in trauma patients did not lead to an increase in attributable mortality,2–5but has been shown to prolong the duration of mechanical ventilation and ICU stay.2–4
We found that 23% of patients with VAP had development of a related ALI/ARDS, and VAP-ALI represented 15% of all episodes of VAP. Other authors, without precise definition, found an incidence rate between 18 and 42%.1,8–10Therefore, the main difficulty encountered in our study was defining VAP-ALI. Four points must be stressed. First, we used quantitative culture of distal bronchial specimen to confirm a clinically suspected VAP.17,18Second, we mandated a maximum of 48 h elapsed between occurrence of VAP and development of ALI, to ensure that such ALI was a specific complication of VAP. Third, in patients who had previously experienced ALI, a clinical resolution was required. Such requirement was not thoughtless because after pulmonary contusion, miliary patterns or consolidations typically appeared within 4–6 h of injury and then resolved over 3–5 days.26A similar course is reported for transfusion-related ALI.27Last, the final diagnosis of VAP-ALI was determined by two independent experts, with excellent agreement.
Previous studies did not analyze the bacteria involved in the VAP-ALI. One of the most peculiar aspects of VAP-ALI in our study was the recognition of H. influenzae as the only significant risk factor. H. influenzae is among the bacteria normally found in the pharynx. Surveys have indicated that up to 80% of persons are carriers.28Most people are colonized with nontypeable H. influenzae (unencapsulated), but in 3–5% of people, the isolates have capsules—most commonly serotype b.28The increased prevalence of patients with chronic airway disease in the VAP-ALI group may have participated in the high prevalence of H. influenzae infection.28Moreover, pulmonary aspiration is a clearly demonstrated risk factor, which has probably occurred in several patients.4Because 63–82% of VAP cases in trauma patients are early onset,1,4a high rate of H. influenzae was constantly found,2,4and polymicrobial infection is common.3,4,17Specifically, without previous antibiotherapy, H. influenzae was identified in 22–53% of VAP cases after trauma.2,4Interestingly, H. influenzae was isolated more frequently when the severity of illness increased.5,29Despite few reports, H. influenzae has been recognized to be a cause of diffuse pulmonary infiltrates and respiratory failure in both trauma and medical patients.29–33
The reason underlying the high prevalence of H. influenzae –associated VAP-ALI remains speculative. Inasmuch as such an association has not been described with other ARDS causes, a specific host–bacteria relation is probably involved. The most prevalent hypothesis is that bacterial growth occurs when constitutive innate defense mechanisms are impaired, favoring infection with endemic strains. We previously described impairment of neutrophil functions in critically ill patients, including trauma patients.34However, no argument is available, to our knowledge, arguing for a specific impairment against H. influenzae . A recent study35suggests an alternative hypothesis, demonstrating that H. influenzae strains isolated from patients during chronic obstructive pulmonary disease exacerbation induce more inflammation and likely have differences in virulence compared with colonizing strains. Along this line, some authors have shown a specific virulence factor associated with disease-causing strains of H. influenzae .36In our study, we were unable to demonstrate phenotypic differences between H. influenzae strains. We may also hypothesize that blood leakage into airspaces due to trauma favored specifically H. influenzae growth because of its specific requirement of erythrocytes.28Last, a genetically related impairment of the specific response to H. influenzae infection could be involved.37
The finding of a higher rate of VAP recurrences in patients with VAP-ALI than in patients with VAP alone was not unexpected. Besides the impairment of phagocytic function during ARDS, the higher incidence of VAP observed in patients with VAP-ALI is probably mainly the result of their need for a much longer duration of mechanical ventilation than that of other patients, thereby increasing the time during which they are at risk for development of VAP.17In recent studies ALI/ARDS trauma patients had a mortality rate ranging between 16 and 30%,7,9,10,22and Eisner et al. 6even reported a mortality rate as low as 11%. Although trauma ALI/ARDS has a lower mortality rate than does ALI/ARDS in patients with other conditions that lead to ALI/ARDS, when VAP-ALI occurred, a higher mortality rate (39%) was reported. Such high mortality rate was in the range of the 36–42% mortality rate of ARDS secondary to pneumonia.6,7
This study has several limitations. First, it was performed at a single medical center and deserves further investigation at other institutions. Second, there was a relatively small sample size, which provides only limited statistical power and limits the conclusion drawn from the multivariate analysis, particularly regarding potential other factors. Third, the diagnosis/definition of VAP-ALI was not completely resolved. Although the main biologic difference between pneumonia and ALI seems to be a loss of compartmentalization of inflammatory response in the latter setting,38definite clinical criteria for practice are lacking. Thus, American-European Consensus Conference definitions19cannot differentiate between moderate and severe VAP. Therefore, the best way to deal with this methodologic issue was probably to use an expert panel.
In conclusion, H. influenzae seemed to be one of the most frequent bacteria involved in the occurrence of VAP-ALI in trauma patients. Such VAP-ALI worsened the prognosis of these patients.
The authors thank Françoise Zerah-Lancner, M.D. (Assistant Professor, Service of Physiology, Pulmonary Function Tests Unit, Assistance Publique-Hôpitaux de Paris, Hôpital Henri Mondor, Créteil, France), for help in statistical analyses.