Active surveillance testing to reduce transmission of carbapenem-resistant, gram-negative bacteria in intensive care units: a pragmatic, randomized cross-over trial

Study design

We conducted a pragmatic, cluster-randomized, nonblinded cross-over study in the included randomized ICUs between June 2019 and June 2020. We included 6 adult ICUs in a tertiary care hospital, Seoul, South Korea: two medical ICUs (23 beds), two surgical ICUs (26 beds), a cardiac ICU (16 beds), and a cardiothoracic surgery ICU (15 beds) in a tertiary care hospital. The study was approved by the physicians and nurse team leaders of each ICU and the institutional review board (IRB no. 2019–0274). The requirements for informed consent were waived. ICUs were randomly assigned to perform active surveillance testing (intervention) or use standard precautions (control) during the initial 6-month study period (period 1), followed by a 1-month washout period, and alternative during the second 6-month period (period 2). Randomization of ICU was performed by SPSS for Windows software, version 21 (SPSS Inc., Chicago, IL, USA). The microbiology laboratory processed surveillance specimens using standard culture-based identification of CRGNB. Patients with histories of CRGNB colonization or infection were placed under contact precautions at the time of admission.

Active surveillance and contact precautions

In the intervention period, stool or perirectal swabs for CRPA, CRAB, and CRE surveillance cultures and sputum, or endotracheal cultures for CRPA or CRAB, were obtained from patients within 2 days of their admission to the ICU and weekly thereafter. In the intervention period, preemptive isolation and contact precautions were implemented at admission, and if the initial surveillance test was negative, contact precautions were ceased, and standard precautions were continued. If the initial surveillance test or subsequent surveillances or clinical culture tests were positive for CRGNB, isolation and contact precautions were continued until 3 negative consecutive test results were obtained. In the control period, surveillance testing was not performed, and if clinical specimens were positive for CRGNB, contact precautions were implemented. During both the intervention and control periods, daily chlorhexidine-bathing was performed in all ICUs, and contact precautions were required in patients with MRSA and VRE colonization or infection. In period 2 (from April to June 2020), universal use of personal protective equipment (PPE) (gown, glove, KF94 mask, and face shield or goggle) was implemented for response to COVID-19 pandemic when caring patients in ICUs. During the whole study period, hand hygiene compliance was observed 4 times by a year by the infection control team staff, and the results by units were disclosed to all hospital staffs. Promotions for improving the compliance of hand hygiene included frequent monitoring and real-time feedback by infection control leader in ICU nursing team, and hospital-wide rewards given to the units with high hand hygiene compliance.

If outbreaks of CRGNB occurred, surveillance and post-outbreak surveillance in the control period were permitted.

Definition

An event was defined as a positive result for CRGNB from a clinical culture. The event date was the date of the earliest positive clinical culture. A patient was classified as having a new event if they had stayed in the ICU > 2 days, had no history of colonization or infection during the previous year, had no positive clinical culture within 2 days after admission to the ICU, and if admitted to an intervention ICU, a negative surveillance culture was obtained within 2 days of admission. Days at risk were calculated from the date of the third day in the ICU through the event date or the date two days after discharge from the ICU, whichever was later.

The primary outcome was the ICU-level incidence of new events per 1000 ICU patient-days at risk. Secondary ICU-level outcomes were the incidences of new events with CRPA, CRAB, or CRE calculated separately and the incidences of hospital-acquired bloodstream infections, catheter-related bloodstream infections, urinary tract infections, catheter-associated urinary tract infections, pneumonia, ventilator-associated pneumonia, and in-ICU mortality. We also performed subgroup analysis of individual ICUs for new events per 1000 ICU patient-days at risk. In addition, we compared new events per 1000 ICU patient-days at risk between periods 1 and 2. For the evaluation of economic impacts, we also compared the lengths of hospital and ICU stays and the costs of hospitalization between the intervention and control periods.

Outbreak was defined as ≥ 3 cases of acquisition of CRGNB within 2 weeks. If surveillance and post-outbreak surveillance were performed in the control period because of a CRGNB outbreak, we excluded the ICU in the modified intention-to-treat (mITT) analysis.

Statistical analysis

Based on the acquisition rate of CRGNB from 2016 to 2018 in ICUs of our hospital, we assumed a mean baseline incidence of CRGNB colonization or infection of 8 per 1000 patient-days; between-cluster variance would be 0.4, and the average amount of time a patient spent in the ICU would be 10 days. This study was designed to achieve 80% power for detecting a reduction in acquisition of 40% in the intervention period with a 2-sided type I error of 5%. According to these assumptions, the estimated sample size was 2400 patients (200 per cluster; a total of 12 clusters with one cross-over of 6 ICUs) [5].

Categorical variables were analyzed using the chi-square or Fisher’s exact test, as appropriate. Normally and non-normally distributed continuous variables were analyzed by Student’s t test and the Mann–Whitney U test, respectively. The primary analysis was a comparison of the primary outcomes between the intervention and control periods using an unadjusted Poisson regression model according to the mITT. All statistical analyses were performed using SPSS for Windows software, version 21 (SPSS Inc., Chicago, IL, USA) and MedCalc Statistical Software version 18.10.2 (MedCalc Software bvba, Sotend, Belgium) with P < 0.05 considered statistically significant.

留言 (0)

沒有登入
gif