This retrospective study was performed at the Antwerp University Hospital, the 573 bed tertiary referral hospital of the University of Antwerp. Pharmacy records were reviewed to identify patients who were admitted to the 39 bed ICU, and were receiving colistin (Colistineb®, colistin sulphomethate sodium, Forest Laboratories, Kent, UK) between June 2007 and June 2009 for therapy of nosocomial pneumonia associated with MDR P. aeruginosa. The primary objective was to compare the treatment outcome of the pneumonia (clinical success; mortality) between different colistin treatment groups (parenteral treatment, inhalation treatment, combination of parenteral and inhalation treatment). The study was approved by the ethical committee (ref. 10/22/154) of the hospital which judged that an informed consent was not required. The following data were recorded for each patient: age, sex, predicted mortality rate based on the Simplified Acute Physiology 3 Score (SAPS3) - score, Sequential Organ Failure Assessment (SOFA) - score at admission to the ICU, and at the onset of the colistin therapy, concomitant other infections, length and dosage of colistin therapy, length of stay, time between admission and development of the pneumonia, simultaneous use of other antibiotics, clinical and microbiological response, and mortality during the episode of the infection [3, 4]. An automated Patient Data Management system (PDMS from iMD-soft) was used to collect most of the data.
Diagnosis of nosocomial pneumonia was defined as pneumonia that occured ≥ 48 hours after hospital admission. Pneumonia was diagnosed by the finding of a new pulmonary infiltrate associated with at least two of the following criteria: fever > 38°C, leukocytosis > 11,000 cells/mm3, or purulent respiratory secretions [5]. Ventilator associated pneumonia (VAP) was defined as a nosocomial pneumonia in a patient with at least 48 hours of mechanical ventilatory support [5]. The etiology of the pneumonia was established by isolation of the organism from blood cultures, or by culturing the organism from endotracheal aspirates or mini-BALs irrespective of the bacterial count.
The antibiotic regimen was determined by the primary treating ICU physician on the basis of a standardised dose regimen, the clinical follow up and response. Generally, ICU physicians decided to treat patients arbitrarily with parenterally administered colistin, together or without inhalation, when they estimated that the infection was "more severe" or if another infection focus with P. aeruginosa was present. Patients were excluded for analysis if they received ≤ 2 days of colistin therapy. A favorable clinical response was defined as a resolution of presenting symptoms and signs at the end of the treatment, and an unfavorable clinical response was defined as persistence or worsening of presenting symptoms and signs [6]. Favorable microbial response was defined as eradication of the strain in a respiratory sample collected in the period of 1 day before until 7 days after discontinuation of the colistin therapy.
Renal function was monitored by serial measurement of the serum creatinine level. Creatinine at the onset of the colistin therapy, maximal creatinine during the colistin therapy, and prior renal replacement therapy (RRT) were recorded. Renal failure was defined following the RIFLE criteria as described by Hartzell et al [7]. Neurotoxicity rates were not determined because of the lack of objective criteria in our patients.
Bacterial isolates were identified by biochemical manual testing methods. Susceptibility testing, including susceptibility for colistin, was performed using disk diffusion testing (Sensitabs, Rosco diagnostics, Denmark) using Clinical Laboratory and Standards Institute (CLSI) breakpoints [8]. Multidrug-resistance was defined as non-susceptibility to at least six of the following antibiotics: meropenem, amikacin, piperacillin-tazobactam, ceftazidime, cefepime, aztreonam, and ciprofloxacin [9]. Quality control was performed weekly in the study period since 2007 using ATCC 27853, and was always situated between reference values.
Colistin was administered intravenously at a dosage of 62500 IU/kg/day in 3 to 4 divided doses, and adjusted for creatinine clearance if necessary. If colistin was administered by inhalation, a standardised regimen of 2 mIU/aerosol t.i.d. was administered. In mechanically ventilated patients, the efficient administration of the inhaled/aerosolized colistin was guaranteed by the use of the standard nebulizers for the ventilators used: Servo Ultra Nebulizer 145/345 for the Servo-I (Maquet, Solna, Sweden) and the 84 12 935 nebulizer for the Dräger Evita 4 Ventilator (Dräger, Lennik, Belgium). During spontaneous breathing, optimal aerosol particle deposition was executed by standard delivery system (Cirrus drug nebulizer, Intersurgical, Berkshire, UK).
All data were entered into a database and analyzed using SAS 9.2. (SAS Institute Inc., North Carolina, USA). For the continuous data, the non-parametric Kruskall Wallis test was used. For the categorical data, we used the non-parametric Fisher's Exact test.