Characterizing Patients Presenting on Hospital Admission With Central Line–Associated Bloodstream Infections: A Multicenter Study

Abstract

Background

There are no systematic measures of central line–associated bloodstream infections (CLABSIs) in patients maintaining central venous catheters (CVCs) outside acute care hospitals. To clarify the burden of CLABSIs in these patients, we characterized patients with CLABSI present on hospital admission (POA).

Methods

Retrospective cross-sectional analysis of patients with CLABSI-POA in 3 health systems covering 11 hospitals across Maryland, Washington DC, and Missouri from November 2020 to October 2021. CLABSI-POA was defined using an adaptation of the acute care CLABSI definition. Patient demographics, clinical characteristics, and outcomes were collected via record review. Cox proportional hazard analysis was used to assess factors associated with the all-cause mortality rate within 30 days.

Results

A total of 461 patients were identified as having CLABSI-POA. CVCs were most commonly maintained in home infusion therapy (32.8%) or oncology clinics (31.2%). Enterobacterales were the most common etiologic agent (29.2%). Recurrent CLABSIs occurred in a quarter of patients (25%). Eleven percent of patients died during the hospital admission. Among patients with CLABSI-POA, mortality risk increased with age (hazard ratio vs age <20 years by age group: 20–44 years, 11.2 [95% confidence interval, 1.46–86.22]; 45–64 years, 20.88 [2.84–153.58]; ≥65 years, 22.50 [2.98–169.93]) and lack of insurance (2.46 [1.08–5.59]), and it decreased with CVC removal (0.57 [.39–.84]).

Conclusions

CLABSI-POA is associated with significant in-hospital mortality risk. Surveillance is required to understand the burden of CLABSI in the community to identify targets for CLABSI prevention initiatives outside acute care settings.

Graphical Abstract

This graphical abstract is also available at Tidbit: https://tidbitapp.io/tidbits/characterizing-patients-presenting-on-hospital-admission-with-central-line-associated-bloodstream-infections-a-multicenter-study-8a995da0-a857-4c84-81e9-2a247ee878c5?utm_campaign=tidbitlinkshare&utm_source=IO

Open in new tabDownload slide

In acute care settings, central line–associated bloodstream infections (CLABSIs) are a primary cause of preventable healthcare-associated infections, resulting in extended hospital stays, increased hospital costs, and increased mortality rates [1–3]. CLABSIs in acute care hospitals are publicly reportable [4]; since public reporting began, national and local prevention initiatives led to a 50% decrease in CLABSI rates [5].

However, because national estimates of CLABSIs focus on acute care hospitals [5], they underestimate the total CLABSI burden in healthcare. Central venous catheters (CVCs) are increasingly maintained outside the hospital, such as in skilled nursing facilities, long-term acute care facilities, hemodialysis facilities, outpatient infusion centers, ambulatory surgical centers, and homes [6]. CLABSI surveillance efforts outside acute care hospitals are limited. Hemodialysis facilities report dialysis events as defined by the National Healthcare Safety Network (NHSN; defined as bloodstream infections, antimicrobial agent starts, or pus, redness, or increased swelling at the vascular access site) as a quality metric [7, 8], but only 11% of dialysis events are reported [9]. Our group previously published a validated home infusion CLABSI definition [10, 11], but this definition has not yet been widely implemented. Meanwhile, long-term acute care and skilled nursing facilities report CLABSI to the NHSN, but reporting is not mandatory [12].

Despite these efforts focusing on reporting outside of acute care hospitals, there is significant room for improvement in surveillance, and other approaches are needed. It is reasonable to assume that the majority of patients with CLABSI in the community would present to an acute care hospital for evaluation and treatment. Therefore, measuring CLABSI present on admission (CLABSI-POA) may help estimate the burden of CLABSI in the community, capturing the different locations where CLABSI may occur, and allowing for targeting of prevention efforts. The aim of our study was to determine the frequency of CLABSI-POA and risk factors for mortality among patients with CLABSI-POA.

METHODS

Study Design

This study was a retrospective cross-sectional study. It was approved by the Johns Hopkins University School of Medicine, Washington University School of Medicine, and University of Maryland School of Medicine institutional review boards.

Setting

We included patients from 3 large academic health systems in Maryland, Washington, DC, and Missouri between November 2020 and October 2021. These included 2 hospitals in health system A (a tertiary adult and a tertiary pediatric academic medical center), 4 hospitals in health system B (a tertiary adult/pediatric academic medical center and 3 adult community hospitals), and 5 hospitals in health system C (a tertiary adult/pediatric academic medical center, a community-based academic medical center, and 3 adult community hospitals).

Population

Patients with CVCs and positive blood cultures within 72 hours of hospital admission were considered for possible inclusion. All health systems used the Epic medical record system. Patients of all ages were considered eligible.

Definition of CLABSI-POA

Patients who would have otherwise met the 2021 National Healthcare Safety Network (NHSN) CLABSI criteria, including criteria for common commensals, but who had the first positive blood culture within the first 3 calendar days of hospital admission were considered to have CLABSI-POA [5]. Port catheters were included if there was documentation of the port being accessed or deaccessed within a month or if the port was being accessed on presentation to the hospital. Positive blood cultures that met the criteria for NHSN Repeat Infection Timeframe were excluded [13]. At each health system, a trained infection preventionist (H. S. or S. M.) or a trained infectious diseases physician (P. R. C. or S.K.) reviewed patients to determine whether they met the 2021 NHSN criteria for CLABSI [5].

Data Collection

Data collected via record review included demographic information, healthcare location (ie, hemodialysis facility, home with support from home nursing or home infusion agency, prison or jail, oncology clinic, outpatient infusion center, skilled nursing facility, subacute rehabilitation facility, or transfer from another hospital), clinical information, information about the CVC (ie, peripherally inserted central catheter, hemodialysis catheter, tunneled CVC, or port), and organisms recovered from blood cultures. In addition, we recorded a history of prior CLABSI, or catheter-related bloodstream infection (CRBSI) based on clinical documentation. We did not perform separate surveillance to confirm whether these prior infections met NHSN CLABSI criteria. It was recorded whether the admission with CLABSI-POA was within 30 days of discharge from another hospital. Pitt bacteremia scores were calculated on data collected within 24 hours of hospital admission [14], and we determined whether patients were admitted to a general or intensive care unit. Outcome data included outcome of hospitalization (eg, discharge to home, discharge to skilled nursing facility or subacute rehabilitation, discharge to hospice, or death), and in-hospital, 30-day, and 6-month all-cause mortality rates.

Statistical Analysis

Descriptive statistics were used to evaluate demographic and clinical characteristics of the patients with CLABSI-POA and the microbiology of CLABSI-POA, including the presence of multidrug-resistant (MDR) organisms (MDROs) (ie, methicillin-resistant Staphylococcus aureus [MRSA], vancomycin-intermediate or vancomycin-resistant S. aureus, vancomycin-resistant Enterococcus, extended-spectrum β-lactamase­–producing [Enterobacterales spp.; MDR] Pseudomonas, carbapenem-resistant Enterobacterales spp., MDR Acinetobacter, and Candida auris). As appropriate, χ² or Fisher exact tests were used to describe differences by health system and by those who had had a prior CLABSI or CRBSI. Kaplan-Meier plots were used to describe the survival probability of the study patients over time. Multivariable Cox regression analysis was performed to determine demographic and clinical variables associated with mortality within 30 days (ie, days to death). This multivariable analysis included demographic and clinical variables that were considered for possible inclusion based on the goodness of fit of the final model. Goodness-of-fit indices included the Akaike information criterion, the Wald test, and the likelihood ratio test.

RESULTS

Over a 12-month study of 11 hospitals, 461 patients were identified as having CLABSI-POA. Of these, 236 (51.2%) were male, and 414 (89.8%) were adults (Table 1). Hospital system B had no young children with CLABSI-POA. The location where patients maintained their CVCs varied by health system; systems A and B had the largest number of patients presenting from home infusion therapy (42.9% and 29.6% respectively), while system C had the largest number from oncology clinics (42.9%). Overall, 38.2% of the patients had received chemotherapy in the 6 months, and 23.0% had received parenteral nutrition (PN) in the month before the CLABSI-POA event. Half of the patients (49%) presented within 30 days of their most recent admission. Tunneled CVCs were the most common type of CVC (29.9%; Supplementary Table 1).

Overall, Enterobacterales were the most common etiologic organism (29.2%; Table 2 and Supplementary Table 2), followed by coagulase-negative staphylococci (17.3%). Fifty-two cases of CLABSI-POA included ≥1 MRDO; 1 was caused by 2 MRDOs.

One hundred thirteen patients (24.5%) had a history of prior CLABSI/CRBSI (Table 3). The prior CLABSI/CRBSI occurred a median of 92 days (interquartile range, 43–269) before the current event. Ten patients (8.8%) had ≥5 prior CLABSIs/CRBSIs. Patients with recurrent CLABSI/CRBSI were more likely to have received PN (41.2% vs 17% in those without recurrence) or home infusion therapy (56.1% vs 25.1%) in the month before the current CLABSI event. Fewer patients with prior CLABSI/CRBSI had received chemotherapy within 6 months (24.6% vs 42.7% of those without prior CLABSI/CRBSI) or had received care in an oncology clinic (14.9% vs 36.6%).

Table 4 describes severity of illness at presentation and health outcomes. About 1 in 5 patients were admitted to the intensive care unit (2.2% for pediatric and 21.9% for adults). The median Pitt bacteremia score was 1 (interquartile range, 0–2), and the median length of hospital stay was 8 (5–14) days. Of the patients, 17.5% (n = 81) died within 30 days of hospital admission. 10.6% (n = 49) died during the acute hospital stay, and another 20.2% (n = 93) died within 6 months after hospital discharge (Figure 1).

In a multivariate analysis, risk factors for death within 6 months in patients with CLABSI-POA (Table 5 and Supplementary Table 3) included older age groups compared with those <20 years of age (hazard ratio for age 20–44 years, 11.21 [95% confidence interval, 1.1.46–86.22]; age 45–64 years, 20.88 [2.84–153.58]; and age ≥65 years, 22.50 [2.98–169.93]) and lack of insurance (2.46 [1.08–5.59]). Removal of CVC during admission was associated with decreased mortality rate (hazard ratio, 0.57 [95% confidence interval, .39–.84]). Other variables were not found to be significantly associated with mortality risk.

DISCUSSION

We examined the number and characteristics of patients presenting with CLABSI in 11 hospitals in 3 health systems across 2 states and the District of Columbia. We found that CLABSI-POA events appear to be frequent. Overall, 461 CLABSI-POA cases were identified over the 1-year study period. As it is unknown how many patients during this time had CVCs in the community, we were unable to calculate a CLABSI-POA per CVC-day rate. A large proportion (38%) of patients with CLABSI-POA had recently received chemotherapy, while almost 1 in 4 had received PN in the prior month. One in 5 had a previous CLABSI or CRBSI. We found that 11% of patients with CLABSI-POA died during that hospital stay, and 18% died within 30 days after hospital admission. While this calculation is not directly comparable to published measures of CLABSI-attributable mortality rate, as it may have included patients who died from other causes, it is Interesting to note that patients with CLABSI in acute care hospitals have a 12%–15% in-hospital attributable mortality rate [15]. Compared with the 461 patients with CLABSI-POA in the 1-year study period, the hospitals discharged a total of 200 396 patients in 2022. In addition, CLABSI rates in the hospitals ranged from standardized infection ratios [13] of 0.273 to 1.777. Resources to prevent CLABSI outside acute care hospitals should be increased.

Understanding who is presenting with CLABSI-POA is important to inform prevention strategies. Most patients in this cohort were admitted from home and were receiving home infusion therapy. While our group has shown that CLABSI rates in home infusion therapy are approximately one-fifth to one-fourth that in acute care settings [11], home infusion therapy is commonly provided. The National Home Infusion Association estimates that 3 million patients received home infusion therapy in 2019—a 300% increase since the prior decade [6]. The data presented here suggests that home infusion therapy should be a focus of infection prevention. In acute care hospitals, CLABSI rates decreased once reporting of CLABSIs was mandated; mandating reporting of CLABSIs in other healthcare locations could lead to a similar decrease [4, 5].

A third of the patients in this study received care in an oncology clinic. Prior work in ambulatory oncology has focused primarily on preventing CLABSI in pediatric oncology populations [16, 17]. While ambulatory pediatric oncology patients have a lower rate of CLABSIs than hospitalized pediatric oncology patients [18], as the majority of pediatric and adult cancer care is provided in ambulatory settings [19], resources should also focus on ambulatory care.

A significant proportion of patients in this study had CVCs for outpatient hemodialysis. CVC-associated infections are the main infectious cause of hospitalization in patients receiving hemodialysis [20]. Bloodstream infections in hemodialysis have been used by the Centers for Medicare and Medicaid Services as part of the end-stage renal disease Quality Incentive Program since 2012 [7, 21], and dialysis facilities currently report device events (which include bloodstream infections) to the NHSN. Among submitted dialysis device events data, 63% of bloodstream infections occurred in patients with a CVC, access-related bloodstream infections are highest in CVCs compared with arteriovenous fistulae or arteriovenous grafts (1.21 per 100 patient-months), and 81.8% initiate dialysis with a CVC [8, 22]. Minimal communication between dialysis facilities and infection prevention staff in hospitals (leading to a lack of awareness for the reason for a hospital admission on the part of the dialysis facility) and lack of dialysis staff training in surveillance may contribute to underreporting of dialysis events [9]. Additional approaches are needed to confirm rates of device events as well as to prevent CLABSIs. For example, avoidance of long-term CVCs and preferential choice of arteriovenous fistulae or arteriovenous grafts is considered a quality metric [21].

Significantly, half of the patients with CLABSI-POA presentations were admitted within 30 days of their most recent hospitalization. Prior research has shown that 27.9% of patients with CLABSIs were readmitted within 30 days [23]. As health systems look for ways to reduce 30-day readmission rates, focusing on educating patients and caregivers on CLABSI prevention strategies before hospital discharge could be an important strategy.

In this cohort, Enterobacterales spp. were the most frequent pathogen, followed by coagulase-negative staphylococci and S. aureus. Enterobacterales isolation among this population is significant in an era of increasing drug resistance. Furthermore, strategies for preventing CLABSIs due to Enterobacterales are less clear. Given the high frequency of patients receiving PN, it is possible that some of these patients may be experiencing intestinal failure, and some cases of CLABSI-POA caused by Enterobacterales spp. may be misclassified as such in the setting of intestinal translocation. Alternatively, there may be the possibility of tap water or sink contamination of the CVC during handwashing or bathing [24], or there may be contamination during patient or caregiver preparation of PN. More research is needed to identify strategies to prevent Enterobacterales-related CLABSIs in community settings. For example, the role of appropriate ostomy care in preventing CVC contamination could be explored.

MDROs accounted for one-third of the isolates in our study. MRSA was the most commonly identified MDRO in both our study and the 2011–2014 NHSN report [25]. Understanding which organisms are implicated in CLABSI-POA and the likelihood of drug resistance can aid in empiric antibiotic choice for a patient presenting with CLABSI-POA. Strategies such as chlorhexidine bathing protocols could be useful in preventing MRSA infections in this cohort [26].

Increasing age and lack of insurance were associated with greater mortality risk after presentation with CLABSI-POA, while removal of CVC was associated with decreased mortality risk. CVC removal is recommended for many organisms causing CLABSI (eg, S. aureus or fungal infections), but catheter salvage is an important and often successful approach in many patients who have a need for long-term access and limited availability for new CVCs [27–29]. Such patients (for example, those requiring hemodialysis or PN) may also be at higher mortality risk. Meanwhile, unlike in other literature [30], S. aureus infections were not associated with mortality risk; it is possible that patients with S. aureus bacteremia in this cohort had fewer comorbid conditions than others in the cohort (eg, cancer or requirement for PN). Most prior work investigating predictors of mortality rate among patients with CLABSI has focused on intensive or acute care settings [31, 32]. In these studies, cancer and age were associated with increased mortality rate.

Based on the findings from this study, CLABSI prevention strategies need to be adapted to healthcare locations outside acute care hospitals. Healthcare workers in all care settings require training in CLABSI prevention. Outside of acute care (and particularly with home infusion, infusion centers, and hemodialysis), patients and their caregivers play a significant role in CVC maintenance. Patients in these settings and their caregivers must learn to keep the CVC dressing clean and dry [24]. For the patients and caregivers especially, it is essential to identify and address barriers to CVC care, audit and observe CVC care practice, and train all family members who may be directly and indirectly involved in line care.

Our study had several limitations. First, outside of acute care hospitals, there are no regional or health system–based lists of all patients with CVCs. CVCs may be placed in acute care settings, in clinics, in radiology suites, surgery centers, and even homes, with no health system–wide lists of these CVCs. Outside of acute care settings, there are no standardized ways to document CVC removal. Because of this, we were unable to calculate a catheter-day denominator to present a CLABSI-POA rate. Second, the study period covered an early portion of the coronavirus disease 2019 (COVID-19) pandemic. Nationally, this was a time of increased acute care CLABSIs [33]. It is unclear what the role of COVID-19 may have been in CLABSIs outside acute care settings. Third, if patients sought care outside the hospitals in the study or did not present to an acute care hospital, the CLABSI-POA measure may have underestimated CLABSIs in the community. Fourth, we did not capture the cause of death or treatment received for the CLABSI, which may have affected mortality outcomes. Fifth, while the study was multicentered and in multiple states, it used large academic healthcare centers and affiliated community hospitals, so it may not be representative of hospitals nationally. Sixth, we had limited data on any prior CRBSI or CLABSI as we looked at the presence of CRBSI or CLABSI as documented in a historical problem list or history of present illness. We were not able to confirm either the diagnosis or management of the prior CRBSI or CLABSI or whether the current CLABSI was a relapse of a prior CLABSI or CRBSI.

In conclusion, we noted frequent CLABSI-POA in this study, and patients with CLABSI-POA had a high in-hospital and 6-month mortality rates. Expanding infection surveillance resources to community settings is needed to understand the frequencies and rates of CLABSI in the community and are an important step in tracking infections and the impact of prevention strategies. Clinicians such as those in the emergency room should have a high index of suspicion for CLABSI among symptomatic patients presenting with a CVC. Dedicated efforts and improved knowledge and practices are also needed for the healthcare workers, patients, and their caregivers. CLABSI surveillance and prevention initiatives should be expanded beyond acute care settings.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author Contributions. H. S., P. R. C., S. M., and S. C. K. reviewed the patients for the study. Y. J. H., A. G., and S. C. K. performed the statistical analysis. O. O. S. and S. C. K. wrote the manuscript. All authors conceptualized the study design and reviewed and edited the manuscript for publication.

Disclaimer. The content is solely the responsibility of the authors and does not necessarily represent the official view of the funding agency.

Data availability. Data are not publicly available.

Financial support. This work was supported by the Centers for Disease Control and Prevention's Epicenters Program (grant U54CK000617 to S. E. C. and C. R.).

References

Author notes