Background The Hospital Infections and Antibiotic Policy Committee guidelines recommend antibiotics to cover coagulase-negative staphylococcus and Gram-negative bacilli with vancomycin + aminoglycoside or aztreonam if Catheter-Related Bacteraemia (CRB) is suspected. Fungal coverage has to be evaluated.

Purpose To assess compliance with the antibiotic treatment protocol in the CRB in the Intensive Care Unit (ICU).

Materials and Methods Observational prospective 6-month study in a 32-bed ICU in a tertiary hospital in patients hospitalised ≥48 hours carrying a Central Venous Catheter (CVC).

Demographic and antibiotic treatment were recorded and compared with the empirical treatment recommended.

Results From 8 September 2011 to 8 March 2012, 596 patients were admitted to ICU; 571 patients used CVC; 390 (68.3%) males, mean age 61.0 ± 15.6 years; the number of CVC used was 844, equivalent to 5578 CVC days.

During this period 114 CVCs were removed in patients with fever and 11 cases of CRB were confirmed (10 patients); incidence 1.97 CRB/1000 CVC days.

Microbiology: 1 Morganella morganii (treatment levofloxacin + piperacillin/tazobactam); 2 methicillin-sensitive Staphylococcus aureus (one treated with meropenem, another levofloxacin + teicoplanin); 3 Staphylococcus epidermidis (one treated with linezolid, the second with piperacillin/tazobactam + teicoplanin, and the last with linezolid + meropenem + caspofungin); 1 Escherichia coli (treatment piperacillin/tazobactam); 1 Pseudomonas aeruginosa (treatment piperacillin/tazobactam); 2 carbapenemase-positive Klebsiella pneumoniae (treated with piperacillin/tazobactam + voriconazole) and 1 Candida glabrata (patient received fluconazole + levofloxacin).

Empiric antibiotic treatment wasn’t correct in 8 cases of CRB, lacking empirical Gram-positive coverage in 7 cases and Gram-negative in 1 case. However, according to microbiological results, bacteraemia coverage was correct in 90%.

Conclusions Protocol compliance is low in the ICU for empirical treatment of CRB. A large number of CVCs were removed for fever with no clear correlation with CRB. Patients with fever of unknown origin receive broad-spectrum antibiotic treatment including antibiotic coverage of a wider spectrum than is strictly necessary for CVC infection. Yet 72.72% of patients would not receive appropriate empirical treatment if CRB was suspected.

No conflict of interest.