Ventilator associated pneumonia (VAP) is the most common nosocomial infection acquired in intensive care unit (ICU), occurring in 8 to 28 %1 of patients under invasive mechanical ventilation (IMV) for 48 hours or more. It is also an independent mortality factor, contributing to an estimate attributable mortality rate of 20 to 30 %.2

VAP diagnosis remains a challenge for the clinician, being essential for early and adequate antibiotherapy institution, assumptions that are significantly associated with hospital mortality reduction.3,4

Clinical diagnosis approach based on conjugation of clinical, radiologic and laboratory criteria5 demonstrated sensitivity and specificity values (69 % and 75 %, respectively) lower than desirable,6 often leading to an over estimation of disease incidence, due to difficulties in differentiating colonization from infection and in differentiating non infectious diseases that can mimic VAP findings.

Besides diagnosis difficulty, VAP antibiotherapy selection is conditioned by the possibility of potentially multidrug resistant organisms (MDRO) infection, agents particularly involved in late-onset pneumonia.7,8 Therefore, it is essential to obtain an assertive microbiologic diagnosis allowing an increase of VAP specificity diagnosis and adequacy of initial therapy and, whenever possible, to reduce antibiotics consumption, their side effects and pressure selection.

In this context, different microbiologic approaches have been used, namely invasive like bronchoalveolar lavage (BAL) and quantitative cultures. Although literature data about their value are controversial and despite two decades of research, there is still not a gold standard technique for VAP definitive diagnosis.

Quantitative cultures have been used to help overcome difficulties in distinguishing colonization from infection. Threshold of 104 colony forming unit (CFU)/ml in BAL presents sensitivity and specificity values that range between 42–93 % and 45–100 % respectively, higher than those described for endotracheal aspirate (ETA) 106 threshold, between 36–82 % and 72–85 %.7

Among studies that compare the impact of invasive techniques with quantitative cultures to those non-invasive,9–18 only five are controlled and randomized.12–16 A French group showed statistically significant differences with more antibiotic-free days and reduced mortality at 14th day in patients included in the invasive strategy. 15 Afterwards, a meta-analysis concluded that the use of invasive techniques affected antibiotic prescription but not mortality. 19 More recently, a Canadian study found no differences in antibiotic use and outcomes in patients submitted to any of these technical approaches.16

Variability in studies design, their frequent power reduced sample size and heterogeneity of criteria applied to populations, may partly explain the diversity of literature data.17,19

Despite controversy, scientific evidence demonstrating benefit in invasive techniques use seems to exist and they are recommended in some consensus documents,7,8 namely if procedure is feasible and biological sample is viable.8

The authors performed a preliminary study with the main goal of evaluating the influence of BAL positive results and quantitative cultures in microbiologic diagnosis and treatment of suspected late VAP in patients with prior antibiotherapy. As a secondary objective, the authors analyzed ICU mortality and length of stay.

A retrospective analysis of clinical files of patients admitted in two adults multidisciplinary Intensive Care Units (28 beds) of a university hospital, which had positive results on BAL performed in a one year period (September 2007–2008) was carried out.

All patients had clinical suspicion of VAP, were admitted for more than four days and were previously submitted to antibiotherapy (minimum duration of two days, prophylaxis included).

VAP was suspected in the presence of a new or worsened radiological infiltrate associated with at least two of the following three criteria: purulent respiratory secretions, temperature of over 38.° C or under 36.° C, leukocyte count of over 10,000/mm3 or leukopenia under 4,000/mm3.

BAL was obtained by flexible bronchoscopy (FBC) as the following protocol: 1) adequate patient sedation and/or curarization; 2) fraction of inspired oxygen of 100 %; 3) introduction of bronchoscope through endotracheal tube using a special adaptor; 4) avoidance of suctioning and local anesthetic use; 5) wedge of the bronchoscope into a subsegmental bronchus corresponding to the area of radiologic abnormality; 6) sequential instillation of six aliquots of sterile saline (20ml each) and slow manual aspiration; 7) rejection of the first aliquot; 8) microbiologic assessment of BAL fluid obtained (minimum 40–60ml) in one hour maximum period. BAL was analysed in a quantitative or qualitative way.

ETA was collected by aseptic technique, within 24 hours of BAL achievement, and processed, in all cases, qualitatively. Growth of one or more microorganisms in quantitative or qualitative cultural exams of these biological products was considered as a positive result.

Therapeutic considerations were based on an ICU antibiotic therapy protocol. Concepts of maintenance and change of therapeutic approach were used, the latter including adjustment/correction of inappropriate treatment, de-escalation and beginning of directed antibiotherapy. Inappropriation was defined by identification of a non suspected agent at time of empiric therapy institution or of a microorganism whose antimicrobial susceptibility test showed resistance to at least one, two in case of P. aeruginosa, prescribed antibiotics. De-escalation could have included reduction in the number of antibiotics, narrowing of antibiotic spectrum, reduction of duration or discontinuation of antibiotherapy.

Results were presented as percentage for categorical and median (interquartile range) or mean (± standard deviation) for continuous variables. Statistical analysis was performed using SPSS, version 16.0. Chi square and Fisher tests were used for proportions, and Mann–Whitney U and Kruskal-Wallis H tests for continuous variables. Statistical significance was set to a value of p ≤ .05.

The main finding of this preliminary study was that in this population with suspected late VAP and prior antibiotherapy, positive BAL and quantitative cultures allowed changes in therapeutic approach, which were in concordance with those described in literature, resulting in antibiotic adequacy and consumption reduction. In addition, there were no statistically significant differences between patients submitted to different therapeutic approaches concerning ICU mortality or length of stay.

The BAL microbiological yield obtained was of 29.2 %. A meta-analysis of randomized studies 19 documented a microbiological positivity in patients submitted to invasive techniques that varies between 44.1 and 68.9 %. However, in these studies, besides the mean duration of IMV being lower, between 6 and 11 days compared with 13 days in our study (median value coincided with mean), also the patients percentage with prior antibiotic prescription was considerably lower, between 50.4 and 78.2 % versus 100 % in our patients. These differences are probably due to the fact that attending to resources concerns in our environment, BAL has been mainly used in late PAV in patients with signs of clinic deterioration, non response despite empirical therapy or immunocompromised, which may explain the higher percentage of prior antibiotherapy use or suspension impossibility, and, therefore, the higher probability of false negatives results.7

Agents isolated on BAL were consistent with those predicted in this type of nosocomial pneumonia. 20,21 According to Trouillet, VAP classification in four groups according to IMV duration and the presence or absence of antibiotherapy may be a useful strategy for antibiotic selection.20 Relevance is placed on the need for covering potentially MDRO, such as P. aeruginosa, A. baumannii, S. maltophilia and MRSA, particularly implicated in group 4 pneumonia (> 7 days of IMV and prior antibiotherapy), as was the case of those included in our study, in which these agents were recovery in 56.3 %.

After knowing BAL cultural result, therapeutic approach was changed in 38.0 %, a rate that is consistent with the described in the previously mentioned meta-analysis (27.0 to 41.7 %).19

In addition, and although a statistical difference was not found, BAL quantitative cultures seem to have influenced prescription, which was more frequent when CFU counts were > 104 (48.5 % versus 28.9 %). Of notice that therapeutic changes were also made with bacterial loads below the threshold of 10 4 (15.8 %), based on the understanding that patient's immunosuppression status and prior antibiotherapy should be considered in quantitative cultures interpretation.7

In most cases, empirical therapy was adequate. Inappropriation was responsible for antibiotic changes in 16.9 %. P. aeruginosa (57.1 %) was the most frequent agent responsible for a non suspected drug resistance. Aspergillus and Candida spp. (60.0 %), considered pathogenic in immunosuppression context, were the most often non suspected agents isolated.

De-escalation occurred in 12.7 %, both by reduction in antibiotics number and spectrum narrowing. However, de-escalation can also occur through reduction in the duration or discontinuation of antibiotherapy. With negative cultures and satisfactory evolution, antibiotics discontinuation is of particular importance,22 namely in the presence of non-fermentative Gram negative bacilli, where a negative result is significant, even under antibiotherapy or recent change of antibiotics, as these agents are frequently persistent and hard to eradicate.7 It is important to enhance, however, that as negative cultures were not analysed there may be an underestimation in the true percentage of de-escalation and therapeutic changes in these patients, so this is a limitation to point out.

It is also important to refer literature data documenting that de-escalation occurs more often when quantitative cultures of invasive techniques are used,14,17 as opposed to non-invasive, while others contradict these findings. 23 In contrast, it seems to be consensual that de-escalation occurs more frequently after implementation of antibiotherapy protocols, based on international recommendations and local epidemiology.24 Despite the mentioned, the high percentage of multiple site infections may affect de-escalation possibility, as occurred in patients in which therapy was maintained, in whom de-escalation based on quantitative cultures had only been possible in a small percentage of cases (9.1 %).

Finally, lower ICU overall mortality and length of stay were found in patients in whom therapy was maintained, corroborating that “right at first time“ may be associated with a better prognosis. On the other way, VAP attributable mortality was lower in the group in which therapy was changed after BAL result knowledge and although this group had a longer ICU length of stay, it was clearly influenced by the value obtained in patients in which direct antibiotherapy was started. Moreover, although the limited size of the sample in each group/subgroup imposes serious limitations in regard to the outcomes analysis, the absence of a statistically significant difference between patients submitted to different therapeutic approaches, may be in agreement to some literature data, indicating that invasive techniques could lead to prescription changes but have no impact on patients' outcomes.19

The impact of BAL use as compared with non invasive techniques can only be inferred by another type of study, namely a randomized one.

Authors declare that they don't have any conflict of interest.