Comment on: Changing epidemiology of central venous catheter-related bloodstream infections: increasing prevalence of Gram-negative pathogens

Sir, Marcos et al. recently published an epidemiological study reporting an increasing incidence of Gram-negative bacteria in central venous catheter-related bloodstream infections (CRBSIs) in their setting from 1991 to 2008. The authors concluded that, if similar data are found in other series, current management of CRBSIs should be revised. The latest Infectious Diseases Society of America (IDSA) guidelines for empirical antimicrobial therapy only recommend routine coverage for Gram-positive bacteria, whereas coverage for Gramnegative microorganisms is reserved for certain circumstances (critically ill or neutropenic patients, those with a femoral catheter and patients with a known Gram-negative infection). In this regard, generalization of broad-spectrum empirical therapy in common infections such as CRBSIs could lead to adverse ecological effects and a potential increase in the incidence of nosocomial infections due to multidrug-resistant bacteria. In order to contrast the results reported by Marcos et al., we conducted a descriptive study to investigate the microbiological epidemiology of CRBSIs in our setting, a 1200 bed teaching hospital, from January 2007 to December 2010. Data were collected from a daily updated database of all patients with CRBSIs, which is part of a continuous, laboratorybased, prospective surveillance programme for hospital-acquired bacteraemia in our general hospital. All adults with CRBSIs are diagnosed by conventional criteria according to IDSA guidelines and are evaluated, treated and followed-up by infectious disease specialists. The main results [episodes of CRBSIs, episodes of CRBSIs due to Gram-negative bacteria (GN-CRBSIs), percentage of GN-CRBSIs relative to overall CRBSI cases and incidence rates] are shown in Table 1. A significant increase over time in the number of GN-CRBSIs in relation to the total CRBSI cases occurred exclusively in our critical care units, whereas the percentage has remained stable (even showing a slight decrease in 2010) in the conventional wards. Analysis of the incidence rates yielded a progressive decrease in CRBSIs (from 0.54 to 0.31 episodes/1000 patient-days) that was more evident in critical care units (from 3.17 to 1.12 episodes/ 1000 patient-days) than in the remaining hospital wards. These epidemiological findings, which differ from those reported by Marcos et al., can be explained by implementation of a bundled strategy for CRBSI prevention in our hospital, starting in 2008. This strategy consisted of multifaceted interventional and educational programmes for all involved staff to improve compliance with evidence-based recommended practices regarding the insertion, maintenance and removal of catheters. This approach has resulted in a significant decrease in the overall incidence of CRBSIs, but not of GN-CRBSIs. Similar results have been reported by other authors, who observed an important decrease in the percentages of Gram-positive and fungal CRBSIs with implementation of a multifaceted strategy. In conclusion, knowledge of the epidemiological trends in CRBSIs in our setting has prompted physicians to continue with the current recommendations for empirical treatment. Gramnegative coverage is mainly considered for critically ill patients, which is precisely the patient profile in critical care units, the only wards where the percentage of GN-CRBSIs is increasing. Our experience illustrates the importance of having local data to enable adaptation of the general recommendations proposed by global agencies.