Central-line–associated bloodstream infections (CLABSIs) are serious life-threatening infections in the intensive care unit (ICU) setting.Reference Rosenthal, Guzman and Orellano 1 , Reference Stevens, Geiger, Concannon, Nelson, Brown and Dumyati 2 The clinical consequences of CLABSI include increased mortality, significant morbidity, and increased length of stay (LOS).Reference Stevens, Geiger, Concannon, Nelson, Brown and Dumyati 2 – Reference Higuera, Rangel-Frausto and Rosenthal 4 From an economic perspective, CLABSIs are also responsible for substantial increases in healthcare costs, as reported in both high-incomeReference Stone, Braccia and Larson 5 and resource-limited countries.Reference Higuera, Rangel-Frausto and Rosenthal 4 The burden posed by CLABSI has not yet been systematically analyzed adequately in resource-limited countries. Although hospitals in resource-limited countries do implement basic infection control programs, compliance with infection control practices varies.Reference Rosenthal 6 As reported by the International Nosocomial Infection Control Consortium (INICC) in pooled studies,Reference Rosenthal, Maki and Salomao 7 – Reference Rosenthal, Al-Abdely and El-Kholy 12 and in studies from Argentina,Reference Barnett, Graves, Rosenthal, Salomao and Rangel-Frausto 3 rates of CLABSI are 3 to 5 times higher in resource-limited countries than in more economically developed countries.
According to the literature from more economically developed countries, the incidence of CLABSI can be prevented and reduced by >30% through basic but effective measures such as those described in the bundle for CLABSI prevention developed by the Institute for Healthcare Improvement 13 : (1) hand hygiene, (2) skin antisepsis with chlorhexidine, (3) maximal barriers, (4) insertion in subclavian vein, and (5) timely central-line (CL) removal. All of these measures were associated with a reduction in the incidence density of CLABSI in developed countries.14 According to the literature from resource-limited countries, this incidence can be reduced by >50%.Reference Rosenthal, Maki and Rodrigues 15
The aim of this study was to determine the effect of the INICC Multidimensional Approach (IMA) and the use of the INICC Surveillance Online System (ISOS) for reduction of CLABSI rates in 14 ICUs of 11 hospitals in 5 Argentinean cities, and through its publication, increase and spread tools and methods to reduce this public health burden in Argentina and similar countries.
METHODS
Setting and Study Design
This prospective, cohort, before–after study was conducted in 14 adult ICUs of 11 INICC member hospitals in 5 cities in Argentina. The study period was divided into a 3-month baseline period and an intervention period starting from the fourth month of participation and extending through the remainder of the study period. Each ICU had an infection control team (ICT) comprised of infection control professionals (ICPs) and medical doctors with formal education and background in internal medicine, critical care, microbiology, and/or hospital epidemiology.
The baseline period was from January 1, 2014, to March 30, 2014. During the baseline period, only cohort HAI outcome and process surveillance in ICU patients (2 components of the IMA as described below) were conducted. The length of the baseline period was set at 3 months for the following reasons: (1) Sample size of patients and number of months of data collection during the baseline period was sufficient for comparison with data collected during the intervention period. From a statistical perspective, the issue is addressed by considering the changes in rates over time. The relatively short baseline period may have impacted the standard error of our estimates. However, this would not cause a bias in the results because there would not be systematic differences between the 2 groups. (2) Our priority was to start intervention as early as possible to achieve CLABSI rate reduction.
The intervention period included the implementation of the 6 components of the IMA, as described below. Notably, the methodology of ISOS has been used by INICC members since January 1, 2014, although the publication describing it was published in 2016.Reference Rosenthal 16
The institutional review boards of all participating hospital agreed to the study protocol, and patient confidentiality was protected by codifying the recorded information, making it identifiable only to the infection control team.
INICC Multidimensional Approach (IMA)
The IMA comprises the simultaneous implementation of the following 6 components for HAI control and prevention: (1) a bundle of infection prevention practice interventions, (2) education, (3) outcome surveillance, (4) process surveillance, (5) feedback on HAI rates and consequences, and (6) performance feedback.
Beyond the Centers for Disease Control and Prevention/National Health Safety Network (CDC/NHSN) surveillance methodology, the IMA also recommends the collection of other data essential to increasing the sensitivity of ICPs to HAIs and to prevent underreporting. 17 According to standard CDC/NHSN methods, numerators are the number of each type of HAI and denominators are device days collected from all patients as pooled data, without determining the number of device days related to a specific patient and without collecting characteristics for each patient. 17 This methodology differs from the IMA in that the design of the cohort study using the INICC methods also includes the collection of specific patient data from all patients, both with and without HAIa, as well as the collection of data regarding the risk factors of HAIs, such as invasive devices and surrogates of HAIs (which include, but are not limited to, high temperature, low blood pressure, results of cultures, and antibiotic therapy). By collecting data on all patients in the ICU, it is possible to match patients with and without an HAI using several characteristics to estimate extra LOS, mortality, and cost. The data indicated in the IMA were registered and uploaded to the ISOS.
Bundle of Infection Prevention Practice
The bundles of infection prevention practice interventions were designed following the recommendations and guidelines published by the SHEA and the IDSA published in 2008Reference Marschall, Mermel and Classen 18 and in 2014,Reference Yokoe, Anderson and Berenholtz 19 the bundle for CLABSI prevention developed by the Institute for Healthcare Improvement in 2012,Reference Wheeler, Giaccone and Hutchinson 20 and the guidelines published by the CDC-NHSN in 2011Reference O’Grady, Alexander and Burns 21 and by the Joint Commission International in 2012.Reference Barbara 22
Components of INICC the Infection Control Bundle for CLABSI PreventionReference Rosenthal 16
The infection control bundle consists of 13 elementsReference Rosenthal 16 :
-
1. Perform hand hygiene before CL insertion or manipulation.
-
2. Use maximal sterile barrier precautions during CL insertion.
-
3. Use a chlorhexidine-based antiseptic for skin preparation.
-
4. Remove CL as early as possible, when not necessary, by means of the daily assessment of the necessity of catheter, thereby aiming at the reduction of CL device utilization ratio (DUR).
-
5. Change administration set every 96 hours unless used for fat, nutrition, or blood products. In these cases, it is changed every 24 hours according to the date on the administration set.
-
6. Use sterile gauze or transparent sterile dressing to cover the insertion site, maintain optimal condition of sterile dressing, change gauze every 48 hours, and change the transparent dressing every 7 days.
-
7. Perform daily bathing with 2% chlorhexidine-impregnated washcloth.
-
8. Avoid insertion of CL in the femoral vein in adult patients.
-
9. Use an all-inclusive catheter cart or kit.
-
10. Avoid using single-use vials more than once.
-
11. Disinfect line hubs, needleless connectors, and ports before accessing the CL.
-
12. Use a split septum as an IV connector.
-
13. Use a collapsible, nonvented, closed-system bag as an IV fluid container.
Education
Education sessions were provided on a monthly basis to all healthcare workers (HCWs) in the participating ICUs on the infection control measures contained in this INICC infection control bundle for CLABSI prevention.Reference Monahan and Fisher 23
Outcome surveillance
Prospective, active outcome surveillance through the ISOS allowed the classification of cohort surveillance data into specific module protocols that apply the CDC/NHSN definitions published in January 2013. 17 These site-specific criteria included reporting instructions that provided full explanations integral to their adequate application. 17
Process surveillance
Process surveillance was performed through the ISOS modules, including the monitoring of compliance with the components of the INICC infection control bundle for CLABSI prevention described here.Reference Rosenthal 16
Feedback on DA-HAI rates and consequences
The ICPs generated reports through the ISOS. The ICU HCWs received feedback on DA-HAI rates and their consequences at monthly meetings held by ICPs, who shared and discussed the results generated by the ISOS. These reports contained several charts and tables showing a running record of the monthly cohort surveillance data, including patient characteristics, such as age and sex, proportion, and pooled means of DA-HAIs, DURs, microorganism profiles, bacterial resistance, and benchmarks of these rates against standards from the CDC-NHSN report of 2013,Reference Dudeck, Edwards and Allen-Bridson 24 which was the last INICC report of 50 countries including Turkey, India, Colombia, and Mexico.Reference Rosenthal, Al-Abdely and El-Kholy 12
Performance feedback
At monthly meetings, performance feedback was provided by ICPs to HCWs working in the ICU by communicating and reviewing the rates resulting from process surveillance.Reference Rosenthal 16 The ICPs shared the report of 32 charts generated through the ISOS, which contained data regarding compliance with the respective elements of the bundle.
Data Collection and Analysis
The ISOS meets the INICC protocol and CDC-NHSN criteria, which were followed by the ICPs who collected daily data on CLABSIs and denominator data, patient days, and specific device days in the ICUs.
These data were uploaded to the ISOS and were used to calculate CLABSI rates per 1,000 CL days and the CL utilization ratio, according to the following formulas: (1) CL days consisted of the total number of days of patient exposure to CL; (2) CL DUR equals the total number of CL days divided by the total number of bed days; (3) CLABSI rate per 1,000 CL days was calculated according to CDC/NHSN formula. 17 We applied CDC/NHSN definitions for CLABSI published by the CDC/NHSN in 2013. 17
Statistical Methods
The ISOS version 3.0 (Buenos Aires, Argentina) was used to calculate CLABSI rates and DUR.
Patient characteristics were compared using the Fisher exact test for dichotomous variables and an unmatched Student t test for continuous variables. P values<.05 by 2-sided tests were considered significant.
We conducted 3 types of analyses to evaluate the impact of our intervention on CLABSI rates: First, we performed an analysis to compare the data of the first 3 months (ie, the baseline period) with the remaining pooled months (the intervention period), using relative risks (RR), 95% confidence interval (CI), and P value. Second, to analyze progressive CLABSI rate reduction, we divided the data into the baseline period followed by an intervention period comprising a 9-month and a 12-month follow-up period. We compared the CLABSI rates for each follow-up period with the baseline CLABSI rate. We calculated the incidence density rates (IDR), IDR ratios, and IDR reduction to account for the CLABSI rate reduction.
Third, we estimated the effect of the intervention on CLABSI using a logistic regression model. A set of covariables was included to account for potential confounding and effect modification. A backward procedure that compares nested models using the Akaike information criterion (AIC) was carried out to get the final set of significant covariables. Collinearity among independent variables was measured using the variance inflation factor (VIF). We calculated the odds ratio (OR) and 95% confidence intervals (CI) for the intervention and other independent variables. The effectiveness of the intervention was calculated using the formula: (1 − OR) ×100, where OR is the adjusted odds ratio estimated by the model. All statistical analyses were performed using R version 3.2.2 software (R Foundation for Statistical Computing, Vienna, Austria). 25
To analyze saved days per year due to reduction of CLABSI, we calculated the average number of CLABSIs per month per ICU during the baseline and intervention periods. We multiplied both numbers by 12 to calculate the number of extra bed days per year due to CLABSI. The difference between both numbers represent the saved bed days per year applying INICC strategy.
To analyze the cost, we multiplied saved days by bed-day cost.
RESULTS
During the study period, we recorded a total of 3,940 patients, hospitalized in for 27,941 days, with a total of 20,777 CL days, in the following types of ICU: medical surgical (n=11), cardiothoracic (n=1) neurosurgical (n=1), and trauma (n=1). Some patient characteristics, such as sex, were similar during both periods, whereas mean age was lower during the intervention. DUR was similar in both periods (Table 1).
TABLE 1 Patient Characteristics, Device Use, and Central Line-Associated Bloodstream Infection Rates, Compliance With Care Bundle in Baseline Period and Intervention Period
NOTE. CI, confidence interval; RR, relative risk; CLABSI, central line-associated bloodstream infection; CL, central line; SD, standard deviation.
a Bed-days are the total number of days that patients are in the ICU during the selected period.
b CL-days are the total number of days of exposure to central line by all the patients in the selected population during the selected period.
Regarding the results of the bundle components, we registered high levels of compliance with the following measures: (1) hand hygiene; (2) daily assessment of the necessity of catheter; (3) number of cases in which the dates of insertion were written in the administration set of the patient or the dressing; (4) presence of sterile dressing; (5) condition of dressings (ie, evaluating whether the dressing was clean, dry, and correctly adhered to the insertion site); (6) use of single-use flushing, (7) daily bathing with a 2% chlorhexidine-impregnated washcloth, and (8) use of a closed IV set connector (Table 1).
During the baseline period, we recorded 5,118 CL days. There were 49 CLABSIs, for an overall baseline rate of 9.6 CLABSIs per 1,000 CL days. During the intervention period, we recorded 15,659 CL days. The rate of CLABSIs per 1,000 CL days was reduced to 4.1 CLABSIs per 1,000 CL days in the second year, accounting for a 57% cumulative CLABSI rate reduction (IDR, 0.43; 95% CI, 0.3–0.6; P<.001) (Table 2 and Figure 1).
FIGURE 1 Central line-associated bloodstream infection rate reduction of each intensive care unit by length of participation.
TABLE 2 Central Line-Associated Bloodstream Infection Rates Stratified by Length of Participation of Each Intensive Care Unit (ICU)
NOTE. INICC, International Nosocomial Infection Control Consortium; CLABSI, central line-associated bloodstream infection; CL, central line; IDR, incidence density rate; CI, confidence interval.
The results of the logistic regression model showed a significant reduction in the CLABSI risk in patients during the intervention period, when controlling for the number of CL days. The adjusted effectiveness of the intervention was 63% (95% CI, 46%–75%). No significant interaction was detected between the intervention and the number of CL days. Collinearity indices in the final model were low, indicating the absence of multicollinearity among the independent variables (Table 3).
TABLE 3 Results of the Logistic Regression Model Showing the Effect of the INICC Intervention on the CLABSI Rates
NOTE. CI, confidence interval; Coeff, β coefficient of the logistic regression; CLABSI, central line-associated bloodstream infection;
INICC, International Nosocomial Infection Control Consortium; OR, odds ratio.
a Adjusted OR for the logistic regression model including the 3 variables in the table.
b Central-line days is the total number of days of exposure to a central line by all the patients in the selected population during the selected period.
c For a unit increase in the predictor.
During the baseline period, the average number of CLABSIs per ICU was 1.2 per month and 16 per year. During the intervention, on average, there were 0.4 CLABSIs per month and 5 CLABSIs per year.
Total extra days per year due to CLABSI were 315 days during baseline and 108 days during the intervention. The average extra LOS per patient with CLABSI was 22.5 days, and the cost per bed day was US $1,000. Total saved days were 207 per ICU per year, amounting to US $207,000 saved per ICU per year.
The microorganism profile is shown in Table 4. The predominant microorganisms during the entire study period were Klebsiella pneumoniae and Staphylococcus aureus.
TABLE 4 Microorganisms Related to Central Line-Associated Bloodstream Infections in Intensive Care Units in Phase 1 (Baseline Period) and Phase 2 (Intervention Period)
DISCUSSION
This study was conducted to assess the effect of the IMA in the ICUs in 5 cities in Argentina. Compared with international CLABSI rates, the baseline rate in this study (6.9 per 1,000 CL days) was similar to the last international INICC report 50 countries for 2010–2015 (4.1 CLABSIs per 1,000 CL days).Reference Rosenthal, Al-Abdely and El-Kholy 12 By contrast, the baseline rate of CLABSI found in this study was significantly higher than the US rate of 0.8 CLABSIs per 1,000 CL days determined by the CDC/NHSN for 2013Reference Dudeck, Edwards and Allen-Bridson 24 ; a rate of 1.4 per 1,000 CL days was reported by the German surveillance system KISS.Reference Geffers and Gastmeier 26 Within the scope of studies addressing the burden of CLABSIs in Argentina, in a study conducted by Rosenthal et alReference Rosenthal, Guzman and Crnich 27 in medical-surgical ICUs, the CLABSI rate per 1,000 CL days was 30.3, whereas this rate was 14.2 in coronary care units.Reference Rosenthal, Guzman and Crnich 27
In our study, some patient characteristics, such as sex and DUR, showed similar intrinsic patient risk in both study periods. The high CLABSI rate determined in our ICUs at baseline was reduced from 9.6 to 4.1 per 1,000 CL days (IDR, 0.43; 95% CI, 0.3–0.6; P<.001), showing a 57% CLABSI rate reduction. This reduction can be associated with the implementation of the IMA because the results of the bundle components showed high levels of compliance with the following measures: (1) hand hygiene; (2) daily assessment of the necessity of catheter; (3) number of cases in which the dates of insertion were written in the administration set of the patient or the dressing; (4) presence of sterile dressing; (5) condition of dressings; (6) use of single-use flushing, (7) daily bathing with a 2% chlorhexidine-impregnated washcloth, and (8) use of closed IV set connector.
The logistic regression analysis showed that the intervention led to a significant reduction in CLABSI risk, and this relationship was modulated by the number of CL days. Moreover, based on the model’s results, the effectiveness of the intervention was 63%, which means there was a significant reduction in the patient CLABSI risk associated with the INICC method.
References from the literature showing a similar reduction can be mentioned from previous studies performed by the INICC that implementation of a 4- or 6-component multidimensional approach for CLABSI resulted in significant reductions in rates of CLABSI. In Latin America, a study in Argentina showed a 76% reduction (46.63 vs 11.10 CLABSIs per 1,000 CL days).Reference Rosenthal, Guzman, Pezzotto and Crnich 28 In Europe, a study in Turkey showed a 47% reduction (22.7 to 12.0 CLABSIs per 1,000 CL days).Reference Leblebicioglu, Ozturk and Rosenthal 29 In India,Reference Jaggi, Rodrigues and Rosenthal 30 a study showed a 39% reduction (6.4 CLABSIs to 3.9 CLABSIs per 1,000 CL days). In addition, multinational studies showed a 33% reduction in adult ICUs (14.5 vs 9.7 CLABSIs per 1,000 CL days)Reference Rosenthal, Maki and Rodrigues 15 and a 51% reduction in pediatric ICUs (10.7 vs 5.2 CLABSIs per 1,000 CL days).Reference Rosenthal, Ramachandran and Villamil-Gomez 31
During the baseline period, the average number of CLABSIs per ICU was 1.2 CLABSIs per month and 16 CLABSIs per year. During the intervention, on average, there were 0.4 CLABSIs per month and 5 CLABSIs per year.
The average extra LOS per patient with CLABSI was 22.5 days, with a cost per bed day of US $1,000. During intervention, 207 days and US $207,000 were saved per ICU per year. In a previous study conducted in Argentina, the mean extra LOS for patients with CLABSI compared with patients without CLABSI was 11.9 days, and the mean extra cost was US $4,888.42.Reference Rosenthal, Guzman, Migone and Crnich 32
Similarly, according to a study conducted by Hollenbeak, the cost of CLABSIs in the US ICU setting was between US $33,000 and US $44,000 in the general adult ICU, between US $54,000 and US $75,000 in the adult surgical ICU, and ~US $49,000 in the pediatric ICU. CLABSIs were associated with reimbursement that was more than US $26,000 less than costs.Reference Hollenbeak 33
Regarding the microorganism profile, we identified a predominance of Klebsiella pneumoniae and Staphylococcus aureus, which have previously been identified to be among the most common causative pathogens in ICUs in Argentina.Reference Barnett, Graves, Rosenthal, Salomao and Rangel-Frausto 3
Study Limitations
First, our findings cannot be generalized to all ICU patients in Argentina. However, the IMA was fundamental to reducing the incidence of CLABSIs in these ICU settings. Second, the 4-month baseline period may have led to an overestimation of the effect of the intervention. Nevertheless, during baseline period, the sample size was adequate, and the confidence intervals for the baseline rate were narrow. Third, due to budget restrictions, some bundle components were not included in process surveillance. Finally, the results of the microorganism profile should be considered with caution due to the small sample size.
ACKNOWLEDGMENTS
The authors thank the many healthcare professionals at each member hospital who assisted with surveillance in their hospitals. The authors also thank Mariano Vilar and Débora López Burgardt, who work at INICC headquarters in Buenos Aires, and the INICC Advisory Board, Country Directors, and Secretaries who generously supported this unique international infection control network: Hail M. Alabdaley, Yassir Khidir Mohamed, Safaa Abdul Aziz AlKhawaja, Amani Ali El-Kholy, Aamer Ikram, Vineya Rai, María Isabel Villegas Mota, Souha S. Kanj, Hakan Leblebicioglu, Yatin Mehta, Bijie Hu, Lul Raka, Najiba M Abdulrazzaq, Sergio Cimerman, Alfonso J. Rodríguez-Morales, Sofía del Carmen González Collantes, Javier Eduardo Desse, Hernán Diosnel Rodríguez Enciso, Nguyen Viet Hung, Wing Hong Seto, Anucha Apisarnthanarak, Toshihiro Mitsuda, Syed Sattar, William Rutala, William R. Jarvis, Russell N. Olmsted, Carla J. Alvarado, Catherine Murphy, Dennis Maki, Nicholas Graves, Patricia Lynch and Didier Pittet.
Financial support: The funding for the activities carried out at INICC headquarters was provided by the corresponding author, Victor D. Rosenthal, and the Foundation to Fight Against Nosocomial Infections.
Potential conflicts of interest: All authors report no conflicts of interest related to this article.
