The setting was a tertiary referral hospital located in the north of Spain. The Department of Neonatology operates under the Paediatrics Clinical Management Unit and is the reference unit for its geographical region. The department has a total of 32 beds, of which 14 belong to the NICU, and 2 are equipped for patient isolation. The usual occupancy rate is 50%.

We conducted an observational and analytical case–control study.

We established the following case definitions7:

• •

  Infection by ESBL-producing K. pneumoniae: isolation of ESBL-producing K. pneumoniae from biological samples and features of infection that supported that this bacterium was the aetiological agent.

• •

  Colonisation by ESBL-producing K. pneumoniae: isolation of ESBL-producing K. pneumoniae from biological samples in the absence of features of infection.

We selected two controls for each case. We used stratified sampling to maximise the homogeneity of cases and controls in relation to sex and birth weight. The period under study went from November 3, 2014 (when the first case was detected) to February 20, 2015 (when the last affected patient was discharged). Consequently, we selected the controls from the set of newborns admitted to the NICU in the same time interval.

We selected the controls from the patients admitted to the unit in the same time period as the cases using a random number generator known as nosetup.

We performed a search of the PubMed database for articles on the subject of interest published in English in the past 10 years. We identified and reviewed 41 articles, and selected those variables found to be relevant in at least 25% of the articles for analysis. These variables were mainly related to placement of invasive devices.2,4,8–10 We also thought it would be of interest to include the variable ‘continuous positive airway pressure’ (CPAP) on account of its recent introduction in our NICUs.

With the aim of improving control of infection transmission and putting an end to the outbreak as quickly as possible, the Department of Preventive Medicine and Public Health developed a document compiling current recommendations for healthcare workers and relatives of patients.

• 1.

  An epidemic curve was produced to identify the primary case and to track the epidemiological evolution of the outbreak (Fig. 1).

  
  

  Fig. 1.

  Epidemic curve: first case → 22/11/2014; last case → 19/01/2015.

  (0.14MB).
• 2.

  Maps of the prevalence of multi-drug resistant microorganisms were drawn periodically to collect information on the evolution of the outbreak at specific time points. The maps covered both inpatient wards and NICU areas. They were shared with the medical and nursing management teams of the hospital and with physicians involved in any of the clinical departments of the hospital.

• 3.

  Communication strategy for health care professionals and relatives of patients:

  • a.

    A document containing general recommendations for outbreak control was developed and given out to professionals involved in NICU operations and other workers. The document addressed aspects related to effective communication, the microbiological strategy, contact isolation and standard precautions, safety of health care procedures, health care clothing, the physical distribution of the space, routine work protocols, cleaning protocol, hands hygiene and management of patient visits.

  • b.

    The Department of Preventive Medicine and Public Health of the hospital held 2 information sessions for the families of the patients involved in the outbreak to explain the situation and the preventive measures that it was going to implement. The purpose of these sessions was to provide direct information about the interventions that were going to be implemented and to ask for their collaboration in minimising the transmission of these sessions was to present first-hand the actions that would be carried out and to achieve their collaboration in minimising pathogen transmission.

• 4.

  Patients in the NICU were transferred to an alternative location to allow environmental decontamination of the unit with vaporised 35% hydrogen peroxide.

To determine the magnitude of the outbreak, pharyngeal and rectal samples were obtained from every patient that was staying in the NICU when the second case was detected. After that, samples were collected on a weekly basis from every patient whose previous results had been negative. Clinical samples were submitted to the National Centre of Microbiology for analysis. This laboratory characterised ESBL-producing K. pneumoniae isolates using phenotypic and molecular methods, including pulsed-field gel electrophoresis.

Samples of high-contact environmental surfaces, the hands of health care workers and disinfectants were obtained with sterile swabs and kept in liquid Amies medium. They were processed according to the recommendations of the Sociedad Española de Enfermedades Infecciosas y Microbiología Clínica in no 42 of its series of scientific documents on microbiological procedures, devoted to environmental microbiological control.11

We performed a descriptive analysis of each variable (univariate analysis), calculating the absolute and relative frequencies for qualitative variables. For quantitative variables, we obtained the mean as the measure of central tendency (with the corresponding standard deviation) as well as the minimum and maximum values.

We performed a bivariate analysis to assess potential associations in the variables under study. We used the Fisher exact test for dichotomous qualitative variables, as the sample was small and did not fulfil the necessary conditions for application of the chi square test. We used the non-parametric Mann–Whitney U test for quantitative variables, as the data did not follow a normal distribution based on the Shapiro–Wilk test (P<.001 in all four criteria and overall) and there were fewer than 30 individuals in the case group. We defined statistical significance as a P-value of less than .05.

We performed the analyses with the Statistical Package for the Social Sciences, version 18.0, and the software EPIDAT, version 3.1.

The sample consisted of 51 newborns, out of who 17 corresponded to cases and the remaining 34 to controls. The sex distribution of the sample was 59% female and 41% male. The primary case, which was also the index case, was detected on November 22, 2014, and the last case on January 19, 2015. Of the 17 cases, 12 (70.5%) were of colonisation and 5 (29.4%) of infection (bacteraemia). Seven newborns in the case group (41.2%) had been delivered by caesarean section.

As for invasive medical procedures, umbilical venous catheters were used in 12 cases (70.6%), umbilical catheters in 9 (52.9%), parenteral nutrition (PN) in 13 (76.5%), invasive mechanical ventilation (IMV) in 6 (35.3%), continuous positive airway pressure (CPAP) in 12 (70.6%) and urinary catheterization (UC) in 15 (88.2%).

When it came to intravenous therapy, 11 patients in the case group (64.7%) received intravenous antibiotherapy, and 11 fluid therapy.

The distribution of patients with positive results for ESBL-producing K. pneumoniae was the following: 10 with a positive sample from a single location (58.8%), 5 with detection of the bacteria in samples from 2 locations (29.4%), and 2 with detection of the bacteria in samples from 3 or more locations (11.8%).

The distribution of cases by birth weight was: 2 (11.8%) with birth weights greater than 2500g, 4 (23.5%) with weights between 1500 and 2500g, 5 (29.4%) with weights between 1000 and 1500g and 5 (29.4%) with weights of less than 1000g. In 1 case, there was no record of the birth weight in the chart.

The distribution by gestational age was as follows: 6 (35.3%) full term, 2 (11.8%) early term, 8 (47.1%) late preterm and 1 (5.9%) extremely preterm.

Last of all, the mean length of stay was 45.4 days (SD, ±23.6) for the overall stay in hospital and 34.5 days (SD, ±20.2) for the NICU stay.

The risk factors with a statistically significant association with colonisation or infection by ESBL-producing K. pneumoniae were: CVC (OR, 5.0 [95% CI, 1.4–17.8]; P=.016], PN (OR, 6.8 [95% CI, 1.8–25.7]; P=.006); UC (OR, 5.9 [95% CI, 1.2–30.0]; P=.028) and birth weight (P=.035) (Table 1).

We also found statistically significant differences in the mean overall length of stay and the mean length of stay in the NICU between cases and controls, with stays that were longer in the case group, by a mean of 20 days for the total hospitalisation (25 vs 45 days, P=.004) and a mean of 16 days for the stay at the NICU (18 vs 34 days, P=.002) (Table 2).

The main measures implemented were the following:

Testing found microbial resistance through production of ESBL type CTX-M-14 in all 17 cases. The molecular weight marker used in the pulsed-field gel electrophoresis tests led to the identification of 3 different profiles in the samples: 1 sample in group 1, 15 in group 2 and 1 in group 3.

All cultures of environmental samples were negative except for the cultures of a sample of topical gel and a sample taken from the hands of a health care worker, which grew the same microorganism involved in the outbreak: ESBL CTX-M-14-producing K. pneumoniae.