University Hospital Virgen del Rocío (Seville, Spain) is a 1367 bed-centre with two adult ICUs, which have six multibed open wards for medical and surgical patients (50 beds), and two wards for trauma and neurosciences areas divided into rooms for one patient each (18 beds). This is a reference centre for severe trauma, neurosurgery, burn patients and solid organ and bone marrow transplantations. During the studied period, the hospital and ICUs had, respectively, an average of 52,593 and 3379 inpatient episodes per year.
This is a quasi-experimental intervention study, based on open cohorts of all the patients admitted to the hospital from the 1st January 2009 to the 31stAugust 2017, having a pre-intervention period of 46 months and a post-intervention period of 58 months.
Infection control team
A multidisciplinary team, composed of an infection control and hospital hygiene specialist, a microbiologist, a pharmaceutical specialist, an infectious diseases physician and three nurses with expertise in infection control, designed and led the intervention program.
A multifaceted program was implemented. All the measures agreed the current recommendations of the Centers for Disease Control and Prevention (CDC) .
An exhaustive environmental cleaning policy was instituted. The disinfectant used were a hypochlorite-based disinfectants; for the handling of not-disposable medical products in which this disinfectant was not appropriate, a wipe disinfectant containing benzalkonium chloride and propane-1,2-diol was used. Medical products were exclusive of each colonized of infected patient. Disposable medical products were discarded once the patient did not need them any more following the CDC recommendations , and not-disposable products were cleaned following the procedures previously described.
To verify the environmental decontamination procedure, two specific checklists were created: one for the general environmental cleaning and the other for medical equipment disinfection. These checklists had to be filled by the cleaning staff every time that any cleaning procedure was performed, and at least twice each day; the percentage of compliance was monitored weekly. The infection control team met several times with the cleaning staff for educational purposes and feedback of the results.
Every ICU ward was sequentially closed for a terminal cleaning at the beginning of the program, and periodically at least thrice a year. Afterwards, every bedside of the ICU and the rooms in other wards from which a colonized patient was discharged underwent terminal cleaning as well.
Hand hygiene instruction and surveillance
Hand hygiene education courses were provided to the ICU staff the first weeks after the XDR-Ab outbreak, and periodically afterwards, emphasizing on training on newly incorporated staff. The hand hygiene training was based on the performance of periodic workshops in which the five moments for hand hygiene were remembered and the technique was trained using an UV glow box. During the first year, structured observation of hand hygiene compliance during the daily care activity was performed weekly by the nurses of the infection control team in the ICU, following the recommendations and observation tool of the WHO . Afterwards, this observation was performed periodically (at least once each month). The percentage of compliance was included in the weekly reports.
Antimicrobial stewardship program
Until the outbreak of XDR-Ab was controlled, colistin was restricted, needing pre-authorization and being revised the indication and duration of every colistin course. Afterwards, the educational stewardship program previously set in the hospital  was enhanced in the ICUs, adding a weekly feedback of the antibiotic consumption data to the staff.
Isolation and contact precautions
Contact precautions, following the CDC recommendations , were already mandatory for patients carrying MDR-Ab. In addition, we implemented the surveillance of its compliance and displayed information posters in every affected unit.
Contact precautions were maintained through the whole admission period for patients with MDR-Ab.
Active surveillance for MDR-Ab colonization
All patients admitted to adult ICUs were screened weekly. Surveillance cultures were obtained by rectal and pharyngeal swabs.
A weekly report was made with the evolution of the selected indicators. It includes the level of compliance of the measures and the results achieved. Definitions and indicators are described below.
Regular meetings with the staff of the affected areas
The infection control team met weekly with the staff of the ICUs (physicians, nurses, assistant nurses, hospital attendants, housekeepers) to discuss the results of the weekly report, and to take additional measures when necessary. Additionally, the medical and nurse directors were informed by weekly email for the whole period.
Definitions and indicators
A. baumannii was defined as MDR or XDR following the criteria of Magiorakos et al. , which consider that A. baumannii is MDR when non-susceptible to ≥1 agent in ≥3 antimicrobial categories (including aminoglycosides, carbapenems, fluorquinolones, antipseudomonal penicillins+β-lactamases inhibitors, extended spectrum cephalosporins, trimethoprim-sulphamethoxazole, ampicillin-sulbactam, polymixins and tetracyclines) and XDR when non-susceptible to ≥1 agent in all but ≤2 of the same categories. If the susceptibility to one antimicrobial was not tested, it was classified as “resistant” (see below). The main outcome was the incidence density of MDR-Ab in clinical samples, defined as the number of patients newly infected or colonized with MDR-Ab per 1000 patient-days. Surveillance samples were excluded because of their absence in the pre-intervention period. An isolate of MDR-Ab was considered as a new episode after 365 days of the last positive culture in the same patient or when the initial infection was cured and MDR-Ab was not isolated in three rectal swabs performed in three consecutive weeks. Hand hygiene compliance was expressed as the percentage of correct actions over all the observed ones . The appropriateness of the isolation and contact precautions was evaluated with a form designed for this purpose. The global compliance was the percentage of patients with a correct response to all the items over the total of patients under contact precautions observed. Environmental and medical equipment cleaning compliance was evaluated by checklists designed for this purpose, completed by the responsible staff (housekeepers and assistant nurses, respectively). They were stated as the percentage of correct actions of each item over the total. Antimicrobial consumption was measured in Daily Defined Doses (DDD) per 1000 patients-days. Hand hygiene, contact precautions and cleaning compliance were surveyed just in the ICUs; all the others in the whole hospital, analysing separately the ICUs.
Active surveillance for MDR-Ab colonization was performed in all patients admitted to the ICU with at least prior 48 h hospitalization . The culture media used was Brilliance™ CRE Agar, a chromogenic screening plate for the detection of carbapenem-resistant strains with high sensitivity and specificity and whose results are available in just 18–24 h. Each isolate was identified by the Matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF MS- Brucker®). Susceptibility testing to colistin and meropenem of the isolates was studied by E-test© (AB Biodisk, Sweden) according to the manufacturer’s recommendations. Susceptibility to other antibiotics was performed by using commercial microdilution methods (PMicroScan combo NC58. BeckamCoulter. USA). Amikacin, ampicillin/sulbactam, imipenem, minocycline and tigecycline were tested in all the isolates. Due to the generalized resistance to some groups of antibiotics (fluorquinolones, antipseudomonal penicillins+β-lactamases inhibitors, extended spectrum cephalosporins, trimethoprim-sulphamethoxazole and tetracycline) among the isolates of A. baumannii from our hospital in previous years, the susceptibility to them was tested just in a variable proportion of isolates. The breakpoints used were those recommended by EUCAST in each time frame .
Molecular typing using PFEG was performed for the initial outbreak investigation, as mentioned above, and to the new isolates of the outbreaks that occurred after the eradication of the endemic. Plug preparation, lysis, cell washing, restriction digestion (60 U of ApaI), and electrophoresis were performed as previously described . PFGE was performed by using a clamped homogeneous electric field electrophoresis (CHEF) DRIII apparatus (Bio-Rad Laboratories, Hercules, CA). The conditions employed were as follows: temperature of 14 °C, voltage of 6 V/cm, run time of 19 h, and switch time of 5 to 20 s. The images obtained were processed with Bio- Rad Molecular Imager® GelDoc™ XR+ with Lab™ Software. PFGE clustering was determined by using the unweighted-pair group method with arithmetic averages (UPGMA) and by using Dice’s coefficient. The tolerance was set at 1%. All calculations were performed by using InfoQuest software (Applied Maths, Saint-Martens-Latem, Belgium). The results of the PFGE typing were compared according to the Tenover criteria .
To estimate changes in the observed trends we used a joinpoint regression analysis , with previous evaluation of homocedasticity and existence of autocorrelation for each variable using SPSS Version 19.0 (Armonk, NY: IBM Corp.). These models give a double result: they identify the time point in which the trend changes and they also estimate the observed trend in each time interval. A maximum of three turning points were searched for each regression analysis. Statistical significance was set in an alpha error of 0.05. The software used was Joinpoint Regression Program, Version 126.96.36.199 - June 2017 (Statistical Methodology and Applications Branch, Surveillance Research Program, National Cancer Institute, Bethesda, MD, USA). In addition, we performed an interrupted time series (ITS) analysis with R version 3.4.3.