High rate of respiratory MDR gram-negative bacteria in H1N1-ARDS treated with ECMO

Dear Editor, During the H1N1 viral respiratory epidemic some patients required admission to the intensive care unit (ICU), and extracorporeal membrane oxygenation (ECMO) was sometimes used after failure of conventional ventilation [1, 2]. Infectious complications of ECMO are ranked second after hemorrhagic complications and are mainly represented by bloodstream infections (BSI) with grampositive cocci [3, 4]. We report the main clinical and microbiological findings of 16 patients (Table 1) with H1N1-ARDS treated with or without ECMO in a regional referral ICU according to the Italian guidelines [2, 5]. The Ethics Committee approved the collection and report of data. All patients were treated with empiric antibiotic treatment and oseltamivir. ECMO was used for 15 ± 14 days (range, 7–46) after a mean of 1.9 days of mechanical ventilation. Bronco-alveolar lavage (BAL) samples were positive during the ICU stay in seven patients: 5 (71.4 %) in the ECMO group [multidrug resistant (MDR) P. aeruginosa, MDR S. maltophilia, S. marcescens, MDR A. baumannii, K. pneumoniae producing carbapenemases (KPC) and Aspergillus fumigatus] compared to two A. baumanni isolates (22.2 %) in the no-ECMO group (p = 0.04). There was only one positive blood culture for S. marcescens in the ECMO group. The mortality was 28.6 and 44.4 % in patients treated with or without ECMO, respectively. A bacterial infection was the probable cause of death in all patients who died, and a possible infection by A. fumigatus was responsible for one death. A selective antibiotic pressure is an important factor for the development of local resistance and also the isolation of MDR strains. Resistant mutants usually may then survive in an environment where several antimicrobials are used, as occurs in the ICU setting where prolonged and combined antibiotic therapy is frequently used. We investigated the possible role of selective antibiotic pressure as a predisposing factor for the isolation of respiratory MDR gram-negative bacteria. The mean daily defined doses (DDDs) at 14 days after the hospital admission were 347 vs. 1,020 (p = 0.04) for meropenem and 316 vs. 632 (p = 0.012) for levofloxacin in the ECMO vs. no-ECMO group, respectively. Conversely, vancomycin and linezolid DDDs were 138 vs. 102 and 561 vs. 122 in the ECMO group and in the no-ECMO group, respectively (not significant). The rate of infections during ECMO varies from 7.5 to 45.5 %, and they are more often caused by grampositive bacteria isolated from the bloodstream [3, 4]. In our ECMO patients only respiratory MDR gramnegative bacteria were isolated, possibly because of the specific setting of H1N1 syndrome, favored by the mechanical ventilation and the comorbidities, but not by a