Setting
We conducted a one-day point prevalence screening (PPS) and one-month surveillance of clinical cultures for CRAB in the MICU of Tan Tock Seng Hospital, a teaching hospital in Singapore. The hospital has 1500 beds and houses four intensive care units, namely the MICU, Surgical Intensive Care Unit, Neurological Intensive Care Unit and the Coronary Care Unit. The MICU contains a total of 6 high-dependency care beds and 12 single-room intensive care beds, of which 3 are negative- pressure rooms.
Study procedures
The study comprised two components: the first was a one-day PPS of patients who were in the MICU on the day of screening (23 July 2015); while the second component was a one-month surveillance of clinical cultures of CRAB isolated from patients within the MICU in the month of July 2015 (Fig. 1).
During the PPS, we collected patients’ endotracheal tube (ETT) aspirates and swabs from their nares, axillae, groin, wounds and exit sites of drains, if any. Swabs were not taken from patients who had been in the MICU for less than 24 h or who were known CRAB carriers.
Environmental swabs were taken from all the patient rooms in the MICU, regardless of whether they were occupied by patients at the time. The following surfaces were swabbed: physiological monitor terminals, bedside rails, automatic door buttons, infusion pumps, resuscitation bags, ventilator panels, ventilator humidifiers, suction bottles, stethoscopes, patients’ lockers, all-purpose tables, sink faucet openings, and sink drainage sites. Swabs were also taken from the MICU environment outside the patients’ rooms. This included the computer keyboards and ‘mice’, cleaned ventilators, equipment carts and emergency carts.
We followed the PPS with a one-month surveillance for clinical cultures to detect onward clonal transmission. The Department of Laboratory Medicine cultured, identified and stored all CRAB isolates from existing and newly admitted MICU patients.
Clinical data collection
We investigated genetic linkage by collecting data from electronic medical records on ward movements of involved patients, overlap of healthcare workers caring for the patients during their stay in the MICU and sharing of any medical equipment.
Specimen collection and culture
All samples were collected by two microbiology doctors and eight infection control nurses who were trained and supervised by the microbiology doctors. Sterile cotton swabs were moistened with sterile water and were rolled over the surface five times. Each swab was collected in Amie’s transport media. The samples were inoculated onto MacConkey agar and incubated at 37 °C for 48 h. A meropenem disc (10 μg, BD Diagnostics, USA) was placed on each plate to select for carbapenem-resistant organisms. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS, Bruker Daltonics, GmbH, Germany) was performed on colonies growing next to the meropenem disc after overnight incubation to identify CRAB. Susceptibility testing was performed on all isolates of A. baumannii using the Kirby –Bauer method and those with a meropenem zone ≤14 mm were interpreted as resistant based on the Clinical Laboratory and Standards Institute criteria (M100- S22) [9].
Whole genome sequencing, species identification and strain typing
Sequencing libraries for each isolate were prepared according to the manufacturer’s recommendation using the Illumina Nextera XT kit (Illumina Inc., USA). The Illumina FastQ files are in the process of being deposited into the GenBank database. De novo assembly of the Illumina reads was performed using SPAdes Genome Assembler [10]. Bacterial species were identified using Kraken [11]. Multilocus STs were identified using SRST2 [12]. Isolates with species discordance comparing phenotypic with genotypic speciation were excluded from further analysis.
Bacterial core genome analysis and determination of transmission clusters
The program Parsnp was used to generate core genome alignments (i.e. conserved orthologous regions present in all included genomes) for all isolates [13]. As previously described, Parsnp screens and excludes sequences based on a threshold MUMi distance. Input for this alignment was the de novo assemblies from the bacterial isolates, as well as a reference sequence available from GenBank (Accession Number: NC_009085.1). The core nucleotide alignments were used as input for Gubbins to exclude putative single-nucleotide polymorphisms (SNPs) arising from recombination [14]. Sequences were then aligned to be concatenated into an artificial ‘genome assembly’, which was used to create a Maximum Likelihood phylogenetic tree with 100 replicates for bootstrap value calculation.
To determine transmission clusters of highest confidence, we defined transmission clusters as isolates that had pair-wise SNP distance less than the pair-wise SNP threshold [15]. The pair-wise SNP threshold was defined as the maximum intra-patient SNP count comparing all repeated isolates with identical species and strain type from the same patient [16]. The pair-wise SNP threshold implemented for this study based on the maximum pair-wise SNP count between CRAB isolates isolated from the same patient (excluding environmental samples) was equal or less than 11 SNP.