Trends of Racial/Ethnic Disparities in Pediatric Central Line–Associated Bloodstream Infections

We performed a retrospective cohort study, using data assembled from the Pediatric Health Information System (PHIS) database covering the period from January 1, 2016, through March 31, 2021. The PHIS collects deidentified discharge-level data from 49 free-standing tertiary children's hospitals affiliated with the Children's Hospital Association. A variety of individual-level characteristics are recorded in the database, including sociodemographic information, payer status, All Patient Refined Diagnosis-Related Group codes; and International Classification of Diseases, 10^th Revision, Clinical Modification codes. Hospitals also report hospital-specific resource utilization and date-stamped billing data using Clinical Transaction Classification (CTC) codes (IBM Watson, Armonk, NY). To maintain database quality, reporting is subjected to data consistency reviews, coding consensus meetings, hospital-specific data quality audits, and data quality reports. The institutional review board at our institution approved the protocol for this study. We adhered to strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines for cohort studies.

Children eligible for inclusion were patients <18 years of age who were admitted to the emergency department or inpatient ward and underwent CVC placement. We identified children who received a CVC using previously described criteria based on CTC codes, as outlined in the supplemental material (Supplemental Table 2).¹⁷  In brief, CTC codes allowed for the identification of the CVC insertion procedure. They also distinguished between peripheral and central access location and the number of lumens of the central line. Patients with missing information specifying race/ethnicity were excluded (3.8%). Given that <5% of data for race/ethnicity was absent, we deemed it unlikely that this missing information would alter the results.

The primary exposure in this study was race/ethnicity, which was operationalized as a categorical variable with four mutually exclusive groups (non-Hispanic White, non-Hispanic Black, Hispanic, and Other). The Other group included patients of Asian, Pacific Islander, American Indian, Alaska Native, or other races/ethnicities.

The primary outcome was the risk-adjusted CLABSI rate according to race or ethnicity. CLABSIs were identified using the International Classification of Diseases, 10^th Revision, Clinical Modification code T80.211A.

Baseline differences between children of different race/ethnicity were described by determining the frequency (percentage) for categorical variables and median (interquartile range) for continuous variables. We estimated the risk-adjusted CLABSI rate by race/ethnicity using mixed-effects log-binomial models, accounting for hospital clustering, and controlling for the following variables: age at admission, sex (male versus female), median household income of the patient’s residence ZIP code (a surrogate for socioeconomic status), CVC type, and presence of a chronic complex medical condition. Chronic complex medical condition was defined according to Feudtner and colleagues as “any medical condition that can be reasonably expected to last at least 12 months (unless death intervenes) and to involve either several different organ systems or 1 organ system severely enough to require specialty pediatric care and probably some period of hospitalization in a tertiary care center.”¹⁸  These chronic conditions have been derived from specific clinical diagnoses/comorbidities, identified using appropriate International Classification of Diseases coding. We did not include insurance status in the multivariable model because of its correlation with income. We compared the risk-adjusted rate of CLABSI by estimating the risk ratios (RRs) with corresponding 95% confidence intervals (95% CIs). Differential trends between race/ethnic groups in the risk-adjusted rate of CLABSIs during the study period were explored by including a two-way interaction term between quarter (continuous) and race/ethnicity, while adjusting for baseline individual and hospital characteristics. We used Stata version 16 (StataCorp, College Station, TX: Stata Press) to perform all data management and analyses.

Table 1 summarizes the characteristics of the study population. A total of 226 802 children were admitted to 49 United States hospitals and underwent CVC placement between 2016 and 2021, of whom 121 156 (53.4%) were White, 40 589 (17.9%) were Black, and 43 374 (19.1%) were Hispanic. White children had the highest rate of private insurance (53.4%), whereas Black children had the highest rate of public insurance (77.5%). Children of Other race/ethnicity and White children were more likely to belong to the highest family income quartile (according to ZIP code) compared with Black and Hispanic children. Black children were more commonly in the lowest family income quartile (36.2%). Hispanic children had the highest rate of cardiovascular (15.5%) and hematology/oncology-related admissions (8.1%). White children had the highest rate of digestive/metabolic-related admissions (12.9%). (Table 1)

A CLABSI occurred in 4763 of the 226 802 patients, resulting in an overall CLABSI rate of 2.1%. Compared with White children, the unadjusted relative risk of CLABSI was 1.32 (95% CI, 1.22–1.43; P < .01) for Black children, 1.10 (95% CI, 0.99–1.21; P = .07) for children of Other race, and 1.21(95% CI, 1.12–1.31; P < .01) for Hispanic children. The adjusted RR of CLABSI relative to White children was 1.27 (95% CI, 1.17–1.37; P < .01) for Black children, 1.09 (95% CI, 0.99–1.21; P = .09) for children of Other race, and 1.16 (95% CI, 1.08–1.26; P < .01) for Hispanic children (Fig 1). Throughout the study period, CLABSI rate decreased over time for children of White and Black race (Fig 2). Specifically, the quarterly decline in CLABSI rate was 0.01 percentage points for White children (P < .01) and 0.02 percentage points for Black children (P < .01). The risk-adjusted decline in CLABSI rate was not statistically significant for Hispanic and Other children (Fig 3). Throughout the study period, both Black and Hispanic children had consistently higher CLABSI rates compared with White children (Figs 1, 2, and 3). Indeed, the rate of CLABSIs in Black and Hispanic children in the first quarter of 2021 was higher than the rate in White children in 2016 (Figs 1 and 2). Despite the overall decreasing trends, we found no statistically significant evidence that the gap in CLABSI rate narrowed over time between race/ethnicity groups, as illustrated by the nonsignificant interaction between quarter and race/ethnicity (Figs 1 and 2).

In this study, we evaluated a large multi-institutional cohort of children admitted to hospitals from across the United States over the past 5 years. The results showed that despite an overall declining rate of CLABSIs, Black and Hispanic children had a consistently higher rate of CLABSIs throughout the study period. Notably, the rate of CLABSIs in Black and Hispanic children in the first quarter of 2021 was higher than the rate of CLABSIs in White children at the start of the study period (2016). We also found no evidence that the racial/ethnic gap in CLABSI rate narrowed over the study period. Hospital-associated CLABSIs are regarded as “never events” across United States health care systems. This ambitious stance has led to policy changes inspiring health care institutions to use quality improvement initiatives to tackle serious patient safety issues.¹⁹  However, it is clear that the benefits of such initiatives vary substantially across race and ethnicity.

A study of pediatric hematology/oncology patients with CVCs receiving home health care from 1992 to 1994 identified that Black children had more than six times the rate of ambulatory CLABSIs than White children.²⁰  An intervention to improve ambulatory CLABSI rates in this population reduced CLABSIs by more than 50% in White children, but more than tripled CLABSI rates in Hispanic children, and had no effect on CLABSI rates in Black children. The authors felt that cultural, ethnic, and language differences of the children’s parents were primary drivers of the disparity, with additional contributions from socioeconomic status and education level. The authors suggested prospective studies to further evaluate this inequity, but nearly three decades later, we are no closer to narrowing the racial/ethnic disparity in central line infections. Our findings echo the disparity noted in ambulatory CLABSIs. Our study is also consistent with a recent retrospective study of more than 79 000 all-payer hospital discharges, which revealed differences in composite healthcare–associated infections between racial/ethnic groups.²¹  Our results are also consonant with the results of Romano et al²²  which showed that the rate of postoperative sepsis and/or infection secondary to medical care was higher in Black patients.

CLABSI is considered an eminently preventable hospital-acquired infection that, despite significant efforts, remains a collective and challenging national problem.²³  This is the reason that central line insertion and maintenance bundles have been widely implemented by the Institute for Healthcare Improvement. These bundles are a set of evidence-based practices involving implementation of stringent infectious control processes, workflow changes, and, most importantly, behavior modifications through health care provider education, training, performance assessment, compliance audits, and feedback.^24,25  Our findings raise the question as to why racial/ethnic disparities in CLABSIs exist nearly two decades after initiation of central line bundle practices, which are known to decrease the rate of CLABSIs.^3,24,26  Compliance with CVC maintenance bundles has been demonstrated to be more important in pediatric CLABSI prevention than compliance with CVC insertion bundles in the intensive care unit setting.²⁷  This contrasts with CLABSI prevention in adult intensive care units, which more strongly correlates with insertion bundle compliance. Maintenance bundles emphasize nursing procedures associated with CVC accessing and medication administration, such as catheter site care and dressing changes; catheter hub, cap, and tubing changes; and infusion tubing assessment and replacement.^26,27  Therefore, our findings of racial/ethnic disparities in CLABSIs should prompt a closer look at maintenance bundle compliance in minority patients. A recent single-institution retrospective study found an increased rate of CLABSIs among Black children.²⁸  Analysis of the data revealed variations in interpreter utilization and line maintenance practices by race/ethnicity, which were felt by the investigators to be contributors to the disparity. It is possible that such differential bundle compliance may reflect healthcare provider implicit biases, which are “attitudes [and] stereotypes that impact understanding, actions, and decisions in an unconscious manner.”²⁹  Although healthcare professionals consciously strive to provide excellent and equitable care, when in situations of stress, competing demands, and fatigue, they may rely more on cognitive shortcuts and are therefore susceptible to the influence of unconscious prejudices.²⁹  A systematic review conducted by Hall and colleagues found that most healthcare providers, including nurses, physicians, and students of both disciplines, appear to have implicit bias in terms of positive attitudes toward White patients and negative attitudes toward people of color.³⁰  Future studies should investigate both insertion and maintenance bundle compliance by patient race/ethnicity to help determine if bias may underlie the disparity in pediatric CLABSI.

This study has some limitations that must be considered when interpreting its findings. First, because line-dwell time could not be abstracted from the PHIS, we are unable to report CLABSI risk in the traditional outcome measure: number of events per 1000 patient-days. We cannot, however, think of a reason that reporting our findings as event rate rather than the traditional number of events per 1000 patient-days would explain away our findings. Additionally, this study is based on a retrospective design; therefore, granular information is limited regarding patient-level clinical characteristics, procedural details, spoken language preference and interpreter use, and palliative or quality-of-life issues. However, this type of limitation is inherent to composite variables and very common in large multi-institutional clinical databases. Given the observational design of our study, we cannot entirely exclude the potential for misclassification bias and coding errors, neither of which are likely to be differential by race or ethnicity. Similarly, information was unavailable regarding the environmental setting or acuity under which the CVCs were inserted. Specific details regarding the CVC type, location, tunneling, and insertion site were also limited, all of which may affect CLABSI outcomes.³¹  There is conflicting evidence on the CLABSI risk associated with different catheters. Traditionally, it is thought that the femoral vein insertion site carries the highest risk for infection and nontunneled catheters are higher risk than PICC lines, with PICC lines being higher risk than implanted ports.^32–34  However, studies demonstrating findings to the contrary exist.^35–37  Given the ambiguity of CLABSI risk associated with each of these central line variables, we felt it appropriate to include patients with all types of CVCs in our analysis. Although we do not have specific information about the CVC, we found no empirical evidence suggesting that children of minority race receive different types of CVCs that their White peers. We also have no reason to believe that the setting or acuity of CVC placement should be racially distributed.

The reason for central line placement could not be determined from the data source. Relatedly, mucosal-barrier injury (MBI) bloodstream infections, or those bloodstream infections related to translocation of bacteria because of breaches in the gastrointestinal and/or oropharyngeal mucosa, were not distinguished from other CLABSIs.³⁸  It is estimated that nearly one-half of all inpatient CLABSIs may be subcategorized as MBIs, with such symptoms/signs as neutropenia and volume/frequency of diarrhea used to identify mucosal barrier injury.^38,39  Indeed, the organisms most commonly isolated from MBI CLABSIs and non-MBI CLABSIs do tend to differ.³⁹  However, standard CLABSI prevention bundles have been shown to reduce MBI CLABSIs, suggesting that the pathogenesis of MBI CLABSIs is still largely related to breaches in insertion and maintenance care of the catheter.⁴⁰  Though we cannot say whether the rate of MBI CLABSIs is racially distributed, we believe that the implications of our findings remain: interventions to ensure that CLABSI prevention bundles are equitably used are urgently needed.

This study did not evaluate ambulatory CLABSIs or CLABSIs that occurred in an episode of care other than the hospitalization corresponding to CVC insertion. Ambulatory CLABSIs account for a significant burden of central line–associated cost and morbidity⁴¹  and without this information, our analysis gives an incomplete picture of the racial/ethnic disparity in CLABSIs. However, by limiting our study to inpatient CLABSIs that developed during the same encounter in which the line was placed, we eliminate the confounders related to environmental contamination. For example, ambulatory CLABSI occurrence has been linked to parental and patient language barriers (and likely difficulty understanding maintenance bundle instructions) and poor continuity of care,^20,42  both of which may be more likely in minority patients and those of low socioeconomic status.⁴³  Additionally, the access to and quality of resources such as home health care may affect CLABSI development.^41,44  Ambulatory patients with CVCs may require access of the line in a variety of settings including home, clinic, laboratory, radiology, or a dialysis center. Compliance with CVC access bundles across these settings is likely to vary.⁴¹ 

Hospitals vary in the degree to which they treat minority and underserved pediatric populations. To account for case mix, our analysis adjusted for a variety of patient- and hospital-level factors, including median income for ZIP code of the patient’s residence and hospital location. However, these social determinants of health are a reality of medical care in the United States and are likely to play some role in the observed CLABSI disparities. Nevertheless, it is important to note that the patients in this cohort were inpatient during their care, thereby removing many of the factors related to access to care (transportation, proximity, compliance) from the picture.

Another limitation of this study is related to the categorical racial/ethnicity designations in PHIS, which are based on parental reports. Census Bureau data from 2015 indicated that although almost 15% of United States infants were multiracial, only one-third of multiracial adults considered themselves as such, with the remainder identifying with a single race or ethnic group.⁴⁵  Thus, studies based on parental designation of their child's race may not account for the child's multiracial and ethnic background. Nevertheless, it important to note that race/ethnicity is a purely social construct, and the standard practice in the United States is to assign the race or ethnicity of the non-White parent to the child.⁴⁶  Furthermore, regardless of how individuals classify themselves, the societal classification is what truly matters because that is what determines any implicit or explicit bias affecting treatment.

The burden of CLABSIs among children is significantly higher in Black and Hispanic children than in their White peers. We also found no evidence that the racial/ethnic gap in CLABSI rate narrowed over the study period. This is yet another indicator of clear health care disparities in the United States. Future studies should determine the disparities in CLABSI risk among specific subpopulations, including hematology/oncology and intestinal failure patients, as well as investigate CLABSI bundle compliance by patient race/ethnicity.

CI

confidence intervals

CLABSI

central line–associated bloodstream infections

CTC

Clinical Transaction Classification

CVC

central venous catheter

MBI

mucosal-barrier injury

PHIS

Pediatric Health Information System

RR

relative risk