Setting, Design and Population
We retrospectively studied electronic microbiology and patient data gathered in the period from November 2011 until July 2019 on the ICU of the Amsterdam University Medical Centre, location VU medical centre (VUmc), a 28 bed ICU in a 730-bed tertiary care centre in the Netherlands. Data were derived from an automated database combining laboratory data with pseudo-anonymous patient data from the Electronic patient dossier (EPD). This database was constructed for antimicrobial stewardship and infection prevention purposes and consists of microbiological data, admission and discharge data, Medicine Administration Records, Surgical interventions and a small amount of patient data: sex, date of birth, date of death. Data visualization was performed using TIBCO® Spotfire®. The non-absorbable antibiotics in the SDD regimen consist of application of paste (colistin 2%, tobramycin 2%, amphoterin B 2%) in the oral cavity and of suspension (colistin 100 mg, tobramycin 80 mg, amphotericin 500 mg) via the nasogastric tube. This regimen was applied q.i.d. until 26-05-2017, thereafter the same regimen was applied t.i.d. Patients admitted to the ICU also receive 4 days of intravenous cefotaxime q.i.d. 1 g, this practice did not change over the course of time.
All adult patients with an ICU admission of at least 72h and with at least 2 surveillance cultures drawn on two separate days were included in the analysis.
Surveillance cultures were taken on admission to the ICU and thereafter once a week on Mondays for pharynx and anus, and on Mondays and Thursdays for sputum. All surveillance cultures were included in the analysis. Surveillance cultures within 72h of admission represented the flora at admission on the ICU and are further called baseline surveillance cultures.
Based on previous literature the following aerobic GNB were defined as PPMs: Klebsiella, Enterobacter, Citrobacter, Proteus, Morganella, Serratia, Acinetobacter and Pseudomonas species. By assessing the prevalence of GNB in the blood cultures of the patients in our cohort we found that Stenotrophomonas maltophilia was also a frequent cultured pathogen in the study period and was therefore added as an PPM. Because we also want to assess the reduction of carriage of endogenous “normal” but potentially pathogenic flora we added Escherichia Coli to the list.
In the Amsterdam UMC medical microbiology laboratories, antimicrobial susceptibility was tested using automated systems, gradient tests and/or using the disk diffusion method.
Decontamination was defined as the reduction of Gram-negative bacterial load to a level at which surveillance cultures are negative (rectal or faeces, pharyngeal and sputum). The number of days in which decontamination should occur to be considered successful, i.e. adequate to reduce infectious complications and mortality, has not been previously defined. In the study of de Smet et al -- in which the relationship between SDD and reduction of mortality was confirmed -- the frequency of GNB isolation from rectal swabs among patients receiving SDD was reduced from 56% at day 3 to 15% at day 14. The SDD regimen used in the study of de Smet et al was identical to the q.i.d. regimen used in this study. Therefore we chose to define successful decontamination as a surveillance culture result negative for GNB within 14 days without positive follow-up surveillance culture for GNB during ICU-admission, i.e. follow-up lasted until the moment of discharge from the ICU.
In case of new fever two blood cultures were drawn. ICU-acquired bacteremia with GNB was defined as bacteraemia occurring at least 48 hours after ICU admission with growth of either Enterobacterales or glucose-nonfermenting Gram-negative rods, without documented bacteraemia with the same species in the first 48 hours of ICU admission. Polymicrobial bacteraemia was defined when one or more microorganisms were isolated from one or more blood cultures, and clinical evidence suggested they had arisen from a common source and were part of the same episode. If the source was unknown, all positive blood cultures occurring within 48 hours of each other are considered as a single bacteraemia.
28-day all-cause mortality is defined as death for any cause within 28 days after the date of admission to the ICU.
Due to the before-after study design, we anticipated that time dependent factors such as antimicrobial resistance could introduce bias. We described the combined prevalence of susceptibility for the components of SDD (tobramycin and colistin) in surveillance cultures at baseline.
Two groups were formed on the basis of the dosing frequency of SDD, q.i.d. versus t.i.d. Continuous variables are presented as median and interquartile range. Categorical variables are presented as percentages. For the difference in proportion of patients with successful decontamination of PPMs in both groups and for comparison of susceptibility of Gram-negative bacteria at baseline a Chi-square test was used. For the difference in time to decontamination of PPMs from surveillance cultures a Kaplan Meier curve was used. For equivalence testing the two one-sided test (TOST) procedure was used. The largest clinically acceptable effect for which equivalence can be declared was a mean difference of 10%. The equivalence limit was set to 0.1 (dE = 0.1). All data available was used, no formal sample size was calculated. Odds ratio was calculated to compare 28-day all-cause mortality between the two groups. Data analysis was performed using R Statistical Software (version 3.6.1; R Foundation for Statistical Computing, Vienna, Austria).