Aggressive versus conservative initiation of antibiotics

Sepsis identification in critically ill patients is still a difficult task, and the right time to administer antibiotics after clinical signs have been detected remains an unsolved issue. Nevertheless, in cases of severe sepsis, early use of appropriate antibiotics seems logical, and many retrospective studies have shown a relation between delay in administration of antibiotic therapy and increased risk of death. Tjasa Hranjec and colleagues reported the results of a quasi-experimental study that compared outcomes of treatment with either an aggressive empirical antibiotic strategy or a more conservative strategy for patients admitted to a surgical intensive-care unit (ICU) with sepsis. The investigators reported that the conservative strategy was not associated with higher mortality, and was possibly associated with improved clinical outcomes. The study message is really impressive, but we believe the data and interpretation need some further clarification to put the findings in context. In the methods, the investigators stated that in the conservative protocol period “antimicrobial drugs were withheld until there was microbiological evidence of infection”. Microbiological evidence of infection takes around 48–72 h with typical culture methods, and takes less time only in rare favourable cases or with the use of molecular diagnostic procedures. Therefore, we expected that during the conservative period the median time from clinical signs or blood cultures to antibiotic initiation would be longer than 24 h, but median time from clinical signs to antibiotic initiation was only 24 h (IQR 9–44) and from blood culture samples it was only 22 h (7–58). Therefore, the trigger for initiation of antibiotic therapy in the conservative period is unclear and should be explained by the authors. The higher mortality reported in the aggressive treatment period was attributed by the authors to inadequate empirical therapy and to a longer duration of antibiotic therapy. This hypothesis might be reasonable, but also seriously challenges the authors' conclusions. Indeed, a monodimensional empirical therapy restricted to piperacillin/tazobactam plus vancomycin, irrespective of site of infection and patient risk factors for acquisition of multidrug-resistant microorganisms and organ dysfunction, is overly simplistic and not in accordance with contemporary treatment guidelines and good practice for the use of antibiotic therapy. This risk is particularly true in ICUs with a high rate of multidrug-resistant bacterial infections, as in the study ICU where a more tailored approach should be mandatory. The duration of antibiotic therapy was longer in the aggressive group than in the conservative group (by 5 days on average), but the difference in time to antibiotic administration from initial blood culture averaged only 14 h. Therefore, earlier administration of antibiotics in the aggressively managed cohort seems unlikely to be the sole explanation for prolonged treatment courses. If the authors believe that a longer duration of antibiotic therapy was a key factor in mortality difference between groups, this reasoning should have to be discussed in their conclusion. The investigators also reported that 80 patients had candidaemia or invasive candida episodes, and, of those, 38 were not appropriately managed by empirical therapy. First, the rate of candida infections seems to be very high—80 (7·8%) of 484 patients. In a recent study in 1265 ICUs in 76 countries, the prevalence of candidaemia and invasive candidiasis was 6·9 per 1000 admissions.5 Second, early antifungal therapy is strongly recommended as an empirical therapy in critically ill patients at high risk of candida infection. Again, we believe that the antibiotics used for empirical therapy in the aggressive protocol might have been suboptimal at best, which truly challenges Hranjec and colleagues' conclusions.