Study design
We did a prospective, before–after, real-world study in the six adult ICUs (Surgical-ICU with 36 beds, Emergency-ICU with 26 beds, Neurosurgical-ICU with 21 beds, Neurology-ICU with 13 beds, Geriatrics-ICU with 20 beds and Medicine-ICU with 19 beds) of Sichuan Academy of Medical Sciences, Sichuan People's Hospital, School of Medicine, University of Electronic Science and Technology of China, a 4200-bed tertiary care teaching hospital in Chengdu in the region of Sichuan (Western China). All adult (≥ 18 years) patients with only one CVC insertion admitted during January 2019 to December 2021 were included, avoiding the possible biases caused by the interaction between different catheter insertion states, such as insertion time and insertion sites. Patients hospitalized ≤ 2 days and/or catheterized ≤ 2 days were excluded. After matching screening, the remaining participants were included in the final comparative analysis. The informed consent was waived by the local ethics committee as it was an observational study.
Study protocol
Two consecutive periods (Period 1, from January 2019 to December 2020, and Period 2, from January 2021 to December 2021) were set. Period 1, the control period, included a basic combination of prevention bundle for CVC insertion based on the published best practices, and a insertion checklist (Additional file 1: Table1) completed once a week by the part-time IPC professionals who has received unified education and has been authorized by the infection management department to stop the procedure if breaches in aseptic technique are observed. Major components of bundles included insertion qualification; hand hygiene; daily skin cleaning with 2% chlorhexidine; use of masks, caps, sterile gloves, sterile gowns, full-body sterile drapes, and the correct dressings; site disinfection with an alcoholic chlorhexidine solution containing more than 0.5% chlorhexidine gluconate (CHG) for CVC insertion; and use of the ultrasound guidance for internal jugular (IJ) catheter insertion.
Period 2, the intervention period, added cross-check and included two processes that all ICU patients with CVC insertion need to be performed: Nurses' checks for insertion practices performed by doctors, and doctors' checks for maintenance practices performed by nurses. To effectively deal with the acute phase of CABSI within 7 days after CVC insertion, we developed another standardized checklist (Additional file 1: Table 2) to be completed by the physician. To help ensure enforce infection control practices for CVC maintenance, three checks were carried out within 7 days after CVC insertion, and the time point were the first day, the fourth day and the seventh day after CVC insertion. Major components of maintenance bundles included hand hygiene; hub/connector/port disinfection; dressing replacement; clean and dry state of insertion site; and removal of blood clot at hub/connector/port.
A three-level inspector mechanism, called "Internal cross-check, Supervision of part-time IPC professionals, and External feedback from full-time IPC professionals", was established to ensure that cross-check work was subject to efficient quality control. First, if executor failed to correct the violation after the checker identified a violation, the checker should page ICU part-time IPC professionals who had been authorized by the infection management department to stop the procedure. Second, part-time IPC professionals carried out a weekly follow-up spot check on the completion of the internal cross-check. Then, the full-time IPC professionals gave results feedback once a quarter and put forward suggestions for continuous improvement.
Central venous catheters
Identical double-/triple- lumen Certofix® central venous catheters (Opaque catheter made of polyurethane with Soft-tip, Safsite® valves, Markings to verify position, Colour-coded Luer Lock connections, Fixation wing, Connecting-cable and Clip, 20 mm Lumen, B. Braun Melsungen AG, Melsungen City, Germany) were used in all patients. Catheters were inserted using the Seldinger technique and full aseptic technique by a physician who had been granted the insertion qualification by the medical department. I.V. transparent film dressing with border (3 M Tegaderm™, Neuss, Germany) was placed over the exit site at the time of insertion.
Definitions and data collection
Primary endpoint was CABSI followed by Centers for Disease Control and Prevention (CDC)/National Healthcare Safety Network (NHSN) definitions and criteria (NHSN Bloodstream Infection Event: https://www.cdc.gov/nhsn/pdfs/pscmanual/4psc_clabscurrent.pdf:) [16]. CABSI was defined as a laboratory-confirmed primary BSI that developed in patients wherein an eligible bloodstream infection-causing organism was identified and a CVC was placed at least 2 calendar days prior to the infection onset.
An independent CABSI prospective whistle-blower system, which was developed by the healthcare-associated infections quality control center in Sichuan province, was used to continuously monitor every case with CVC insertion. For each case, suspicious CABSI cases were automatically screened out by the intelligent identification program of the whistle-blower system if any of the following conditions were met: ① Fever (> 38 °C), ② Hypotension (systolic blood pressure < 90 mmHg and/or diastolic blood pressure < 60 mmHg), ③ Oliguria(< 400 ml/day), ④ Any microorganism cultured from ≥ 1 blood cultures, ⑤ CABSI cases that have been prospectively entered into the HAI electronic system by physicians.
To accomplish the study objectives, four infectious disease specialists revised the electronic medical records of all screened cases to check if all NHSN criteria were fulfilled.
In all cases the following variables were recorded: age, sex, hemodialysis, allogeneic blood transfusion (ABT), mechanical ventilation, urinary catheter insertion (UCI), tracheotomy, scheduled or non-scheduled surgery, reason for admission [International Classification of Diseases (ICD)-10 coded] [17], diabetes mellitus, hypertension, chronic obstructive pulmonary disease (COPD) and community infections.
Statistical analysis
Randomization
Propensity score matching (PSM), a method of post-randomization, was performed between intervention and control groups using specific R Package Matching version 4.9–2 to optimize inter group comparability. To minimize the impact of potential bias, we performed the following analysis: 1. Standardized mean differences (SMD) were determined the baseline characteristics that lead to incompatibility between groups. And the SMD of less than 0.1 was considered as an indicator of good balance [18]. 2. A logistic regression model was used to calculate the propensity scores. 3. A k-nearest neighbor algorithm was used to make a 1:1 match without replacement using the caliper of width equal to 0.2 of the standard deviation of the legit of the propensity scores [19]. 4. SMD was used again to determine inter group comparability after PSM.
Restricted cubic spline model
A multivariable logistic model with restricted cubic splines (RCS) was built using specific R Package Regression Modeling Strategies (RMS) version 6.2–0 to evaluate the impact of CVC-insertion in different sites on CABSI. RCS has been widely described as a valid strategy to realize the correlation analysis between continuous exposure and outcomes [20,21,22]. The spline was defined using five knots at the 5th, 25th, 50th, 75th, and 95th percentiles. The threshold was determined as the point in time with the smallest Odd Ratio (OR).
Time-to-event analysis
The Kaplan–Meier survival functions were used to estimate the cumulative hazard of CABSI. Log rank (Mantel-Cox) test was used for time-to-event analysis. For infection-free survival analysis, survival endpoints in this study were 30 days from insertion until CABSI.
Data analysis
R software (v3.6.1) under RStudio (v1.2.5001) was utilized for data analyses. Data are presented as number (corresponding percentage) for categorical variables, and as median (interquartile range, IQR) for non-normally continuous variables and as mean ± standard deviation for normally continuous variables. Using Wilcoxon rank sum test/one-way ANOVA for continuous variables and the χ2 test / Fischer’ exact test for categorical variables, data were compared between the two periods. Using specific R Package VIOPLOT version 0.3.7, Violin plots were used to visualize the incidence densities of CABSI between the two periods. All tests were 2-sided with an \(\alpha\) level of 0.05.