Background and aim
Central line-associated bloodstream infections (CLABSI) in intensive care units (ICU) are significant healthcare-associated infections (HAI), resulting in increased mortality, length of stay, and healthcare costs. Optimizing infection prevention and control and antimicrobial stewardship practices is essential to prevent HAI and the emergence of antimicrobial resistance. Over the last 2 decades, there have been extensive efforts to prevent CLABSI using multimodal care bundles Reference Buetti, Marschall and Drees1,Reference Resar, Griffin, Haraden and Nolan2 and as part of hospital quality improvement initiatives. Reference Ista, van der Hoven and Kornelisse3,Reference Blot, Bergs, Vogelaers, Blot and Vandijck4
Victoria is the second most populous Australian jurisdiction with a population of over 6.7 million. Hospitals have participated in a state surveillance program for HAI since 2002, coordinated by the Victorian Healthcare Associated Infection Coordinating Centre (VICNISS). Although resistant pathogens have been reported in hospital settings within this region, Reference Cronin, Poy Lorenzo and Olenski5,Reference Worth, Harrison and Dickinson6 it is not clear if emergent pathogens have specifically contributed to CLABSI burden, and no recent evaluation of CLABSI pathogen antimicrobial susceptibility has been performed. A timely review of CLABSI pathogens is beneficial to identify future opportunities for CLABSI prevention. Reference Novosad, Fike and Dudeck7
The aim of this study was to describe CLABSI incidence, pathogens responsible for infection, and antimicrobial susceptibility of pathogens reported between 2011 and 2022 in adult ICU in Victoria, Australia.
Methods
Definitions
The VICNISS CLABSI surveillance module, as previously described, Reference Worth, Spelman, Bull, Brett and Richards8 is based on methods employed by the Centers for Disease Control and Prevention’s National Healthcare Safety Network. 9 A CLABSI event is a primary bloodstream infection in a patient with a central venous catheter (CVC) in place for 2 or more consecutive days prior to the event and where pathogen and clinical criteria are met.
Data collection
Continuous surveillance is undertaken in hospitals by infection prevention and control staff trained in VICNISS data submission. CLABSI event data include hospital and ICU admission dates, patient demographics, clinical and laboratory criteria, organism, and antimicrobial susceptibilities. Denominator data include a number of ICU patients and those with a CVC in place; these are used to determine the device utilization ratio (DUR) and CLABSI rate per 1,000 CVC days.
Analysis
Data from 2011 to 2022 were analyzed for changes in rates, pathogens, and antimicrobial resistance. CVC surveillance data and patient demographics are summarized annually using the median and interquartile ranges for continuous data with categorical data described by frequencies and percentages. Yearly DUR values were calculated by dividing the number of CVC days by the number of patient days. Annual CLABSI rates, derived by dividing the total number of CLABSI events by the total number of CVC days, were calculated per 1,000 CVC days, with 95% Poisson exact confidence intervals. Mixed effects negative binomial regression was used to assess whether any time trend was present in the annual CLABSI rates. Robust standard errors were applied to account for misspecifications, and the hospital identifier was modeled as a random effect to adjust for intrahospital correlation. The total number of CVC days per period was fitted as the exposure.
Pathogen data from blood cultures, grouped into 3-year epochs (2011–2013, 2014–2016, 2017–2019, and 2020–2022) and described by frequencies and percentages, were analyzed to identify the most common organisms and antimicrobial resistance. Changes in pathogen incidence over time were evaluated using mixed effects Poisson regression with robust standard errors, the hospital identifier as a random effect, and the total number of CVC days per period fitted as the exposure. Mixed effects negative binomial regression was applied where overdispersion was present.
To assess changes in pathogen-specific antimicrobial resistance, Poisson regression was used where the total number of the specific pathogen per epoch formed the exposure. For Staphylococcus aureus, any reported resistance to methicillin, oxacillin, flucloxacillin, or cefoxitin was classified as methicillin-resistant S. aureus (MRSA). For Enterococcus faecium or Enterococcus faecalis, reported resistance to vancomycin was classified as vancomycin-resistant E. faecium or E. faecalis. For Escherichia coli and Klebsiella pneumoniae, reported resistance to ceftriaxone or ciprofloxacin and fluconazole for Candida spp. was classified as resistant to these agents. Susceptibility analysis of fungal isolates over time was not carried out as insufficient results were available in the study extract.
The effect size for each regression model was quantified as the incidence risk ratio. Linear and nonlinear models were compared using the likelihood ratio test to determine the best fit for the observed data.
All statistical analyses were performed using STATA/SE 18.0 (StataCorp LP, College Station, TX, USA). 10
Ethics
This study was approved by the human research ethics committee associated with the administration of the VICNISS program (RMH QA2022086).
Results
Between 2011 and 2022, 608 CLABSI events were reported over 751,350 CVC days from 47 participating facilities (Table 1). Of the 28 public and 19 private hospitals that participated, 22 (78.6%) and 15 (78.9%), respectively, reported at least 1 event. The median age of patients was 54.9 years (interquartile range: (24, 68)) with 65.8% males and with a median length of stay in ICU before CLABSI event of 8 days (Supplementary file 1).
Table 1. Central venous catheters under surveillance and central line-associated bloodstream infections outcomes, 2011–2022
Note. CLABSI, central line-associated bloodstream infections; DUR, device utilization ratio.
CLABSI rates and DUR
Overall CLABSI incidence was 0.81 per 1,000 CVC days; a 49.3% rate reduction was observed from 2011 to 2022 (1.39 to 0.70 per 1,000 CVC days) (Figure 1). The largest reduction in CLABSI rates was observed in the period 2011–2014, after which a plateauing of incidence has been observed. For the period 2011–2022, the DUR decreased by 15.4% from 0.67 to 0.56, with the largest reduction observed from 2011 to 2013. Overall, the DUR was 0.57.
Figure 1. Central line-associated bloodstream infection rates and device utilization ratio, 2011–2022.
CLABSI pathogens
Overall, 690 pathogens consisting of 72 unique species were responsible for the 608 reported CLABSI events. Most events were monomicrobial (88.0%), with an increase in the proportion of gram-positive organisms causing CLABSI over time. By pathogen, the most common by rank order were coagulase-negative Staphylococci (CNS), Candida spp., S. aureus, E. faecalis, E. faecium, and E. coli (Table 2). The proportion of CNS-causing CLABSI events increased by 69.0% from 2011 to 2022, with the largest increase observed from time epoch 1 (2011–2013) to 2 (2014–2016). This trend was not observed for other organisms.
Table 2. Central line-associated bloodstream infection pathogens and time trends, 2011–2022
Note. IRR, incidence risk ratio.
^ Denominator is the total number of pathogens by time epoch.
* No Candida auris reported.
Antimicrobial resistance
The proportion of antimicrobial-resistant pathogens by time epoch is presented in Table 3.
Table 3. Antimicrobial susceptibility for top-ranked pathogens responsible for central line-associated bloodstream infections, Victorian adult intensive care unit patients, 2011–2022
Note. IRR, incidence risk ratio; S. aureus, Staphylococcus aureus; S-DD, susceptible-dose dependent; E. faecalis, Enterococcus faecalis; E. faecium, Enterococcus faecium; E. coli, Escherichia coli; K. pneumoniae, Klebsiella pneumoniae.
Gram-positive organisms
For CLABSI events due to S. aureus, an approximate 33% decrease in methicillin resistance was observed for every increase in 1 time epoch. On the other hand, for events due to E. faecium, an approximate 4% increase in vancomycin resistance was observed for every increase in 1 time epoch. No analysis was possible for E. faecalis due to insufficient data.
Gram-negative organisms
For CLABSI events due to E. coli, an approximate 12% increase in ceftriaxone resistance was observed for every increase in 1 time epoch. For CLABSI events due to K. pneumoniae, no meaningful result was possible due to insufficient data. No carbapenem resistance was reported in gram-negative pathogens.
Discussion
This study provides the most comprehensive report of CLABSI burden and pathogen susceptibility in Australia to date, with an analysis of time trends over the last decade. The data set is representative of adult ICU in Victoria, which comprises approximately 20% of ICU beds in Australia. Reference Litton, Huckson and Chavan11 This report provides new insights into previously reported data from 2009 to 2013. Reference Worth, Spelman, Bull, Brett and Richards8 We demonstrate a decreasing incidence of CLABSI in adult ICU settings, which plateaued in 2015. Although overall incidence has decreased, the burden of infections due to CNS has proportionally increased. Notably, we did not identify any significant time trend increases in antimicrobial-resistant organisms, including MRSA, vancomycin-resistant E. faecium, and extended-spectrum beta-lactamase (ESBL)-producing Enterobacterales. These findings are relevant for identifying priorities and suitable approaches to CLABSI prevention in Victorian adult ICU.
The trend we observed of decreasing central-line infections in adult ICU is consistent with trends reported internationally. Reference Novosad, Fike and Dudeck7,12 This observed reduction in CLABSI rates corresponded to the introduction of CLABSI surveillance in Victorian hospitals in 2002, Reference Worth, Spelman, Bull, Brett and Richards8 the inclusion of CLABSI as a key performance indicator for Victorian hospitals in 2008, 13 and the promotion and implementation of bundles of care for prevention by local hospital-level initiatives and release of guidelines on management of central lines by Australia New Zealand Intensive Care Society in 2012. Reference Entesari-Tatafi, Orford, Bailey, Chonghaile, Lamb-Jenkins and Athan14,15 Reported rates are much lower than other countries with less developed programs of surveillance and infection prevention. Reference Mathur, Malpiedi and Walia16,Reference Zeng, Wu, Li and Jia17
Other studies have reported changes in patterns of pathogens in non-Australian ICU over time. US ICU participating in Centers for Disease Control and Prevention’s National Healthcare Safety Network surveillance reported a significant reduction in S. aureus incidence rates from 2006 to 2010 Reference Fagan, Edwards, Park, Fridkin and Magill18 and a significant reduction in CNS and S. aureus with an increase in Candida spp./yeast CLABSIs in adult ICU from 2011 to 2017. Reference Cronin, Poy Lorenzo and Olenski5 In US adult ICU in 2017, the most frequently reported pathogens were Candida spp./yeast (25.1%), followed by Enterococcus spp. (16.9%), CNS (13.8%), and S. aureus (9.1%) Reference Weiner-Lastinger, Abner and Edwards19 . We previously reported significant reductions in Enterococcus spp., S. aureus, and CNS in the VICNISS adult ICU cohort from 2009 to 2013, with the most frequently identified pathogens being Enterococcus spp. (26.3%), Candida spp. (15.4%), S. aureus (13.3%), and CNS (10.4%). Reference Worth, Spelman, Bull, Brett and Richards8
From 2011 to 2022, we observed a reduction in the total number of pathogens in adult ICU corresponding to a decrease in CLABSI rates. However, we report significant increases in the proportion of CNS with over 1 in 5 CLABSI events attributed to CNS pathogens since 2014. These events were identified using internationally accepted surveillance criteria specifically developed to distinguish CLABSI events from potential contaminants (ie, requirement for 2 or more positive blood cultures to confirm infection). The contribution of CNS pathogens to CLABSI events is comparable to, or higher than, those reported by others using comparable surveillance methodology (the United States 14%, Scotland 19%, Belgium 19%). Reference Novosad, Fike and Dudeck7,20 Potentially contributing factors include lapses in infection prevention measures such as hand hygiene, aseptic technique, and transmission-based precautions, which are amenable to targeted interventions.
We have not observed increasing rates of antimicrobial-resistant MRSA or E. coli; however, we report persistently high rates of vancomycin-resistant E. faecium in the Victorian ICU. The proportion of vancomycin-resistant E. faecium isolated from our CLABSI cohort between 2017 and 2022 (61.5%–62.5%) is much higher than national data in 2020–2021 (12.5%–15.2%) but similar to data from their Victorian laboratories (61.6%–64.2%), where two-thirds are vanB type 21 . The proportion of E. coli resistance to ceftriaxone isolated from our CLABSI cohort between 2017 and 2022 (20.0%–22.2%) is higher than that reported in the Victorian cohort from national surveillance data in 2020–2021 (13.0%–17.0%) 21 . The proportion of MRSA isolated from our CLABSI cohort between 2017 and 2022 (15.8%–20.0%) is similar to that reported in the Victorian cohort from national surveillance data in 2020–2021 (approximately 12%–15%) 21 . Our reports of higher rates of resistance in E. faecium and E. coli than national surveillance reporting reflect different patient cohorts; although our data are derived from clinical infections attributed to healthcare intervention, national surveillance data are derived from sentinel pathology services around Australia and include hospital or community isolates and clinical or surveillance isolates. As community MRSA rates in Victoria are similar to hospitals, although ESBL E. coli and VRE rates are more common in hospitals, the differences in patient population are likely to partially account for our higher rates of resistance compared with national surveillance.
Our findings support the use of targeted approaches to further reduce CLABSI rates in Victorian ICU. CLABSI prevention approaches have involved the use of multimodal intervention bundles focused on CVC insertion and maintenance practices. Reference Blot, Bergs, Vogelaers, Blot and Vandijck4,Reference Entesari-Tatafi, Orford, Bailey, Chonghaile, Lamb-Jenkins and Athan14 Not all bundles incorporate equivalent components, and these may need to be reviewed to determine if they are effective and optimal for addressing current-era pathogens and antimicrobial resistance. The risk of CLABSI caused by skin organisms such as CNS and S. aureus can potentially be mitigated by improvements in hand hygiene, aseptic technique, and chlorhexidine bathing. CLABSI caused by Enterobacterales may warrant novel approaches to active surveillance and isolation policies, environmental cleaning, and antimicrobial stewardship to further mitigate risk. The use of antiseptic or antimicrobial-impregnated devices is currently recommended as an additional strategy reserved for adult ICU settings with high CLABSI rates. Reference Buetti, Marschall and Drees1 New interventions introduced into ICU settings to reduce the risk of HAI require careful consideration of their potential impact on environmental antimicrobial resistance against individual patient benefit. An example of this would be selective decontamination of the digestive tract in the treatment of critically ill patients. A meta-analysis Reference Hammond, Myburgh and Seppelt22 showed that selective decontamination of the digestive tract with intravenous antibiotics reduced the risk for in-hospital mortality, ventilator-associated pneumonia, and ICU-acquired bacteremia.
Our evaluation of antimicrobial resistance in CLABSI pathogens can be used to assist local CLABSI treatment approaches. International guidelines support the use of empirical vancomycin for suspected device-related bloodstream infections, which is underpinned by our data where vancomycin would adequately cover many gram-positive organisms including MRSA and CNS but not VRE. We propose that the use of antibiotic agent/s to empirically cover antimicrobial-resistant organisms such as VRE and ESBL Enterobacterales should be based on local antimicrobial susceptibility data. Our findings highlight the need for healthcare facilities in our region to review local epidemiology and antibiotic susceptibility of infections to support appropriate prescribing practices.
A limitation of our study is the use of phenotypic antimicrobial resistance reports, rather than genotyping, to define antimicrobial-resistant pathogens responsible for CLABSI. Furthermore, we were not able to report on resistance trends in intrinsically sensitive Candida spp. such as C. albicans, due to missing data on fluconazole susceptibility. Although the evaluation of the relationship between CVC insertion and maintenance practices and infection burden was beyond the scope of the current study, we acknowledge this as a priority for future research to identify the impact of clinical practice on infection burden and pathogens.
Optimal infection prevention and control practices and implementing antimicrobial stewardship are essential measures to prevent HAI and the emergence of antimicrobial resistance in ICU. Further understanding of the burden and incidence of CLABSI, and changes in pathogen and resistance patterns, contributes to inform targets for future CLABSI prevention approaches. Although CLABSI rates in our region have continued to decline over recent years, the proportion attributable to CNS has increased, and novel approaches are required to specifically reduce risks for CNS infection.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/ice.2024.132.
Acknowledgments
We thank VICNISS participants for their contributions to HAI surveillance and improving the quality of health care and patient safety.
Financial support
None.
Competing interests
None.
