Ventilator associated pneumonia (VAP) is the most frequent nosocomial infection in the intensive care unit (ICU) 12, is a major determinant for increases in ICU length of stay and ventilator days 3 and is associated with a two-fold increase in the mortality rate 4. In contrast to the importance of starting therapy promptly, as any delay increases mortality, morbidity and cost 156, the initial choice of antibiotic is still a matter of debate. In particular, the use of a combination therapy versus a monotherapy, especially against particular strains such as Pseudomonas aeruginosa, remains controversial. In a recent meta-analysis, no benefit was found for combination therapy over monotherapy in terms of mortality or prevention of resistant germs 7.

Combination therapy is actually proposed not only to broaden the antibacterial spectrum to each potential pathogen, but also to raise the bactericidal activity of the treatment. If it is difficult to demonstrate an effect in terms of mortality 7, other outcome variables, including the evolution of inflammatory parameters and ventilator dependence, could be used to assess the effectiveness of different regimens. Temperature, leukocytosis and PaO₂/FiO₂ (arterial oxygen tension/inspiratory oxygen fraction) are in fact part of the clinical pulmonary infection score (CPIS) used to define the presence of VAP and to follow the response to treatment 89. In addition to these parameters, C-reactive protein (CRP), a sensitive marker of inflammation produced by the liver in response to cytokines released by activated mononuclear phagocytes cells 10, is also used for the diagnosis and monitoring of different acute inflammatory and infectious processes 1112. In community acquired pneumonia (CAP), it was confirmed to be a useful marker for the differentiation of bacterial and viral infections 131415. It also allows an evaluation of the adequacy of antibacterial treatment as the CRP plasma levels decreased by 50% within 3.3 days in patients receiving adequate treatment 16. One can wonder whether this advantage holds true for VAP.

The aim of this study, therefore, was to evaluate the clinical evolution of patients with VAP who were treated empirically with one or two drugs, comparing as the main outcome the time course of clinical inflammatory parameters and duration of ventilatory support, and mortality as a secondary outcome. The a priori hypothesis is that synergistic action of combination therapy will accelerate the resolution of inflammation.

This prospective and randomised study was performed in the 26-bed general ICU of the University Hospital of Liege Sart-Tilman, Belgium, over a total period of 21 months, from April 2002 to December 2003. The study received the approval of the Ethics Committee of the hospital. Patients or their next of kin had to give their written informed consent.

Seventy-four patients fulfilling the clinical VAP criteria were randomised into three groups. Of these, 24 patients received cefepime only (group C), 26 received cefepime with amikacin (group C-A) and 24 received cefepime with levofloxacin (group C-L).

Pneumonia was not microbiologically confirmed in 15 of these patients: 4 in the C group, 7 in the C-A group and 4 in the C-L group. Quantitative cultures were obtained in 34 patients (12, 12 and 10 for the C, C-A and C-L groups, respectively). The mean CPIS of these 34 patients (10.1 ± 1.7) was not significantly different from those without quantitative culture (9.5 ± 1.9). The mean CPIS of the 15 patients in whom VAP was not confirmed was significantly lower (6.3 ± 2.1).

The characteristics of the 59 patients in whom VAP was confirmed are given in Table 2. There were no significant differences between the three groups in terms of age, sex, underlying diseases, APACHE II and SOFA scores, CPIS, length of stay in the hospital before infection and duration of mechanical ventilation before VAP.

Eighty two bacteria strains were isolated from tracheal aspirate or bronchoalveolar lavage samples; 2 bacteria strains were isolated from 23 patients and only 1 from the remaining 36 patients. There was no statistically significant difference in the distribution of bacteria strains between the three groups (Table 3). Three patients did not receive an adequate treatment (one in the C group and two in the C-L group), all due to the presence of methicillin resistant Staphylococcus aureus. In addition, in the C-L group, one P. aeruginosa was resistant to quinolones.

There were no significant differences in the evolution of PaO₂/FiO₂, temperature, leukocytosis and CRP level between the three groups. The corresponding values at day 1 and day 8 are given in Table 4 and the CRP evolution is shown in Figure 1. When the 15 excluded patients were added to the analysis, there was still no difference between groups and the evolution was the same.

The aim of this study was to compare monotherapy versus combination therapy for the treatment of VAP and to look for differences in the clinical and inflammatory parameters, including CRP level. This is the reason why we optimised the doses of antibiotics as recently recommended 25, and took a highly selected subgroup of ICU patients, represented mainly by post-trauma, postoperative or post-intracranial haemorrhage patients without infection at their entry into the ICU. This choice was made in order to obtain as far as possible a group of patients who did not have any life-threatening conditions other than VAP.

The diagnosis of VAP remains controversial. Some experts recommend obtaining quantitative culture of a protected specimen of pulmonary secretions 1819 while others accept the use of the CPIS score based on clinical data 92021. However, this score is far from sensitive and also has a low specificity, as recently published 2223. In particular, the recent retrospective multicenter French study of Luyt and colleagues 24 comparing CPIS and quantitative culture in patients with VAP found a specificity as low as 47%. Due to the fact that quantitative cultures are not routinely available in our hospital, we used a modified CPIS to confirm the presence of VAP, in which we incorporated the rapid recent increase of CRP level, as it is one of our current diagnostic tools, and enrolled patients before obtaining the culture results. As such, all 59 patients with confirmed VAP had a CPIS above 6 with at least a positive non-quantitative culture. It is interesting to note that in this study, patients fulfilling the clinical criteria of VAP but with an absence of pathogens in their lung secretions had a CPIS significantly lower than the others and, in contrast, all patients with quantitative cultures had a CPIS of more than 6. It would be worth validating this 'new' CPIS in another prospective study.

The association of two drugs in the empirical treatment of VAP is recommended in order to increase the spectrum of activity, to decrease the emergence of resistance during treatment, and to improve the bactericidal activity of therapy 25. If the first reason is obvious because of the presence of potential nosocomial resistant pathogens and cannot be questioned, the two others remain a matter of debate. Cefepime, a fourth generation cephalosporin, possesses bactericidal activity against the great majority of nosocomial pathogens in our hospital and covers almost 100% of non-inducible Enterobacteriaceae, more than 90% of inducible Enterobacteriaceae and nearly 90% of P. aeruginosa. We decided, therefore, to use it as a first line therapy for VAP, in association or not with either amikacin or levofloxacin, two other antibiotics with excellent activity against nosocomial pathogens. The combination of an anti-Pseudomonas beta-lactam with either an aminoglycoside or a quinolone such as levofloxacin or ciprofloxacin is indeed part of the published guidelines for the empirical treatment of VAP 25.

Among inflammatory parameters, CRP was chosen because of its routine use and its well known time course evolution. Interleukin 6 or procalcitonin, two other inflammatory parameters, well described as markers of severity, have not been validated as markers of response to antibiotic therapy 2627. CRP, however, which is used as a non-specific marker of inflammatory events and of bacterial infection, has been used to monitor response to antibiotic treatment 11. Recently, the evolution of CRP levels has been described after CAP. Hansson and colleagues 16 showed that the mean time to a 50% decrease in CRP level was 3.3 days after CAP. Our results recorded in VAP patients, however, show that the decrease in CRP level is much slower (six days to obtain a 50% reduction). Of course, critically ill patients often have causes other than just a lung infection that increase their CRP levels. The vast majority of patients included in this study were either post-trauma or postoperative and their infections were a new insult but not the first they had encountered; however, the time delay of more than nine days between the hospital entry and the occurrence of infection (Table 2) should have been long enough to allow the CRP level, which usually peaks at 48 hours after trauma or surgery, to decrease. Figure 1 shows that the CRP levels peaked the second day after inclusion for VAP and confirms the new insult caused by the infection. The plasma half-life of CRP is about 19 hours and seems to be constant whatever the disease 28; this means that the sole determinant of CRP level is the synthesis rate, which can be sustained after infections complicating trauma or surgery. The circulating CRP concentration decreases if the infectious stimulus ceases, however, and this should fall more rapidly with a better treatment than with a weaker one. This is what we were expecting to find in this study, but this was not the case: CRP levels failed to show any differences between monotherapy and combination therapy. The CRP levels decreased in the same manner in all three groups.

With regard to temperature, leukocytosis and PaO₂/FiO₂, Dennesen and colleagues 29 already described the resolution of infectious parameters after initiation of appropriate therapy for VAP in 27 patients. They showed that improvements in temperature, leucocytosis and PaO₂/FiO₂ were slow and most evident during the first six days 29.

There was no difference in terms of mortality, length of stay and VFDs between the three groups. Moreover, persistence of bacteria at the end of treatment was not reduced by combination therapy and recurrence of infection was as frequent in the three groups. The persistence of bacteria in tracheal aspirates confirms the observation of Dennesen and colleagues 29, who observed 50% persistence for Enterobacteriaceae and 100% persistence for P. aeruginosa in spite of appropriate therapy.

Weaknesses of this study include a lack of blinding of administration to therapy and the relatively small number of patients. We enrolled only 74 patients within 21 months, although we had to treat more than 300 ICU acquired pulmonary infections during this period. This was mainly due to the exclusion criteria, which did not admit patients receiving antibiotic treatment for another infection or infection already treated before randomisation. These criteria excluded all patients ventilated for CAP or for abdominal sepsis and all immunocompromised patients receiving antibiotics, among others. If this study failed to demonstrate any advantage of combination therapy over monotherapy, the use of empirical combination therapy remains probably relevant in patients with potentially multiresistant strains, as recommended by Chastre 30. Importantly, the size of the samples in this study does not allow us to draw definitive conclusions regarding this issue.

We failed to demonstrate any advantage of combination therapy over monotherapy. None of the recorded parameters, including time course of clinical and inflammatory parameters, duration of mechanical ventilation, ICU length of stay, bacteriological response and mortality, were influenced by the type of treatment. The mean time to obtain a 50% decrease in CRP levels after VAP was about six days, twice that seen after CAP, reflecting in part the complexity of the different insults ICU patients often may suffer from.

• The beneficial effects of an empirical combination antibiotic therapy for VAP were not confirmed in terms of the evolution of clinical and biological parameters.

• CRP plasma levels decreased by 50% within six days after antibiotic treatment of VAP.

CAP = community-acquired pneumonia; CPIS = clinical pulmonary infection score; CRP = C-reactive protein; ICU = intensive care unit; SOFA = sequential organ failure assessment; VAP = ventilator-associated pneumonia; VFD = ventilatory free days.

The authors declare that they have no competing interests.

PD and MM originated the study and contributed to the analysis of data. PD, JCP and CG prepared the paper. CG and MN collected the data, and JLC, DL and JCP enrolled the patients. All authors read and approved the final manuscript.