Study design
A prospective observational study was performed at the Dr. Cipto Mangunkusumo General Hospital, a 1000-bed teaching hospital in Jakarta, Indonesia, from April–October 2013 and from April–August 2014. We conducted this study in two ICUs: the 12-bedded adult ICU and the 8-bedded Emergency Room (ER)-ICU with an average of 1010 and 415 admissions per year, respectively. The adult ICU is an open ward with mechanical ventilation facilities, admitting patients with mixed medical and surgical indications, and one designated nurse per patient during first shifts (7 am-3 pm) and a 1:1.5 nurse/patient ratio during other shifts. The ER-ICU has the same design, and the nurse-to-patient ratio in the first shifts is 1:1 and during the other shifts 1:2. The populations served by these two ICUs were identical, and there was also no difference in the service provided.
The study was performed in the framework of a larger study that focused on carbapenem-nonsusceptible Klebsiella pneumoniae, Pseudomonas aeruginosa, and A. baumannii-calcoaceticus complex.
All adult patients (≥18 years old) admitted to one of the two ICUs and hospitalized for more than 48 h were eligible for enrollment in this study. The first screening cultures were taken on the day of admission, and if a patient was discharged before 48 h, he or she was excluded. Informed consent was obtained from the patient or their relatives as applicable. Demographic and clinical characteristics such as age, gender, medical or surgical indication, underlying diseases, hospitalization history, and previous use of antibiotics were recorded on admission.
Systemic inflammatory response syndrome (SIRS) criteria on admission were used as a screening tool to assess (severity of) septic illness. The SIRS criteria were calculated and included in the study, as this was practice at the time of the study [6].
The quick Sequential Organ Failure Assessment (qSOFA) score is a new bedside prompt that may identify patients with suspected infection and helps to determine sepsis in all healthcare environments. The qSOFA score assigns one point for each of the following conditions: systolic blood pressure ≤ 100 mmHg, respiratory rate ≥ 22 breaths per minute, and altered mentation (Glasgow coma scale <15). A qSOFA score ≥ 2 at the onset of infection is associated with a greater risk of death and prolonged ICU stay [6].
The primary outcome measure was acquisition of a carbapenem-nonsusceptible A. baumannii-calcoaceticus complex. Acquisition is defined as a screening culture or clinical culture with first detection of A. baumannii-calcoaceticus complex, with reduced susceptibility to a carbapenem, that was not present within the first 48 h of admission. Secondary outcome measures were length of stay in the ICU, and mortality during ICU stay.
Environmental samples (Additional file 1: Table S1), were taken twice (in October 2013 and December 2014), simultaneously in both ICUs. Screening of healthcare workers (HCWs) was performed once. HCWs were defined as all personnel including doctors, nurses and other people (cleaning staff, administration staff, porters, nutritionist) working in one of the two ICUs during the study period.
Screening method
From patients enrolled, screening cultures were obtained from throat and rectum or stools by experienced ICU nurses who had been trained for the task of taking the samples, on the day of admission, at the time of discharge from the ICU, and weekly if the patient was admitted for seven days or more. Sampling was performed using sterile cotton-tipped swabs, and swabs were transported to the laboratory in Amies transport medium (Oxoid, Basingstoke, UK). The swabs in medium were transported in clean, closed boxes at ambient temperature to the laboratory on the same day. All swabs were processed in the laboratory within 24 h.
Clinical samples were collected on indication from patients under aseptic precautions from the lower respiratory tract, blood, urine, tissue, or wound.
Environmental samples were taken from various sites, including washbasins, bed rails, bedside cabinet tables, ventilators, and monitor screens (Additional file 1: Table S1), with sterile cotton-tipped swabs and placed in Amies transport medium.
All HCWs working in one of the ICUs were sampled once over the course of one month (September 2013) with sterile cotton-tipped swabs, which were transported to the laboratory in Amies transport medium.
Microbiological methods
Isolation and identification
In the Clinical Microbiology Laboratory of Faculty of Medicine, Universitas Indonesia, Jakarta, each swab was placed in 5 ml trypticase soy broth (TSB) supplemented with cefotaxime 2 mg/L plus vancomycin 50 mg/L and incubated overnight. The next day, a loop of broth (10 μl) was subsequently subcultured onto MacConkey agar (Oxoid) and incubated aerobically at 37 °C for 16–24 h, following which identification using the VITEK2® system (bioMérieux, Lyon, France) and susceptibility testing of colonies suggestive of A. baumannii-calcoaceticus complex was performed. All swabs, i.e. from patients, healthcare workers (HCWs), and environmental screening were processed in the same way.
Blood cultures were collected in BACTEC® (BD, Franklin Lakes, NJ, USA) bottles as per manufacturer’s instructions at the discretion of attending clinicians with a minimum of 10 ml of blood collected from at least two puncture sites. Other clinical specimens were inoculated onto blood and MacConkey agar plates and incubated for 24 h at 37 °C. Subsequently, all colonies that had been cultured were examined for morphology by Gram stain and identified using the VITEK2® system.
Strains were stored in duplicate in −80 °C in TSB with glycerol 10%. One tube of each strain was sent to the Department of Medical Microbiology and Infectious Diseases, Erasmus MC, Rotterdam, the Netherlands, which laboratory holds an ISO 15189 accreditation, for further analysis. The other tube of each strain remained in the Indonesian laboratory. In the Netherlands, the identity of strains was confirmed using matrix-assisted laser desorption/ionisation (Maldi Biotyper, Bruker Microflex LT, London, UK).
The quality control strains used for this part of the study in Indonesia were Escherichia coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853, in the laboratory in Erasmus MC multiple quality control strains were used.
Antimicrobial susceptibility testing
The susceptibility of the screening isolates to imipenem and meropenem was investigated by standard Kirby-Bauer disc diffusion technique using Mueller-Hinton agar plates (BD). For the isolates from clinical cultures, approximation of the minimum inhibitory concentrations (MICs) of antibiotics was determined by the VITEK2® system. Carbapenem zone sizes and MICs were interpreted according to EUCAST (2013) using the following breakpoints: meropenem zone size <21 mm and MIC >2 mg/L, imipenem zone size <23 mm and MIC >2 mg/L. [7] For this part of the study, quality control strains as described above were used.
DNA extraction and carbapenemase gene detection
DNA from the isolates was extracted by a cell lysis step and boiling using the InstaGene Matrix (Bio-Rad Laboratories, USA) according to the manufacturer’s instructions. PCR-based detection of Ambler class B metallo-beta-lactamases (blaNDM), class D beta-lactamases (blaOXA-23-like, blaOXA-24-like, blaOXA-51-like and blaOXA-58-like) and ISAba1 were carried out using a T3000 Thermocycler (Biometra-Whatman, Goettingen). The upstream location of the ISAba1 insertion element of the blaOXA-23-like gene was demonstrated by using the ISAba1 forward primer and the blaOXA-23-like reverse primer. PCR primers and reaction conditions for PCR were as described previously [8,9,10,11]. Amplified PCR products were resolved by electrophoresis at 250 V for 30 min on 1.5% agarose gels with 0.5 x Tris (89 mM)-boric acid (89 mM)-EDTA(2 mM) buffer containing SyBr® Safe DNA Gel Stain and visualized under UV light and photographed. In each run, a positive and negative control was included.
Clonal relatedness
Raman spectroscopy (SpectraCell RA® Bacterial Strain Analyzer, RiverD International BV, Rotterdam, The Netherlands) was applied as a first typing method [12, 13]. All isolates were grown overnight on trypticase soy agar (TSA; BD). Samples were prepared and submitted to spectrometry as described previously [13]. Raman light scatterings were analyzed by SpectraCellRA software version 1.9.0.13444:24. The similarity between pairs of spectra was calculated using the squared Pearson correlation coefficient (R2-values), multiplied by 100 and expressed as a percentage. The similarity threshold for this study was set at 91% so that two isolates with an R2 below this threshold were considered to be different and were designated different Raman types. Two isolates with an R2-value between 91% and 100% were considered indistinguishable and were considered to have the same Raman type. Correlation matrices displayed as 2D plots diagram were created using MATLAB version 7.1 (The MathWorks, Natick, MA, USA).
Multilocus sequence typing (MLST) was used as a second typing method for a subset of isolates, including isolates from the largest clones of Raman spectroscopy, and all isolates from blood cultures (one per patient). These isolates were subjected to whole genome sequencing (WGS) using Illumina chemistry. MLST typing results were deduced from the WGS data and assigned based on the Oxford database (pubmlst.org/abaumannii).
Statistical analysis
Statistical analyses were done using SPSS Version 24.0 (SPSS, Chicago, IL, USA). Patients admitted to adult ICU were compared to ER-ICU using Chi square or Fisher’s Exact and Mann-Whitney as appropriate. One-way ANOVA was used to compare patient characteristics according to their A. baumannii-calcoaceticus complex status. Univariate and multivariate analyses were performed to establish risk factors associated with mortality using a multivariate logistic regression model with backward selection and inclusion of variables with a p-value <0.1 in the univariate analysis. Cox proportional regression was used to analyse risk factors for length of stay. Kaplan-Meier method was performed to construct survival curves. P-values of less than 0.01 were considered significant [14].