Outcomes after bloodstream infection in hospitalized pediatric hematology/oncology and stem cell transplant patients

Abstract

Background:

Pediatric hematology/oncology (PHO) patients receiving therapy or undergoing hematopoietic stem cell transplantation (HSCT) often require a central line and are at risk for bloodstream infections (BSI). There are limited data describing outcomes of BSI in PHO and HSCT patients.

Methods:

This is a multicenter (n = 17) retrospective analysis of outcomes of patients who developed a BSI. Centers involved participated in a quality improvement collaborative referred to as the Childhood Cancer and Blood Disorder Network within the Children’s Hospital Association. The main outcome measures were all-cause mortality at 3, 10, and 30 days after positive culture date; transfer to the intensive care unit (ICU) within 48 hours of positive culture; and central line removal within seven days of the positive blood culture.

Results:

Nine hundred fifty-seven BSI were included in the analysis. Three hundred fifty-four BSI (37%) were associated with at least one adverse outcome. All-cause mortality was 1% (n = 9), 3% (n = 26), and 6% (n = 57) at 3, 10, and 30 days after BSI, respectively. In the 165 BSI (17%) associated with admission to the ICU, the median ICU stay was four days (IQR2–10). Twenty-one percent of all infections (n = 203) were associated with central line removal within seven days of positive blood culture.

Conclusions:

BSI in PHO and HSCT patients are associated with adverse outcomes. These data will assist in defining the impact of BSI in this population and demonstrate the need for quality improvement and research efforts to decrease them.

1 |. INTRODUCTION

Pediatric hematology/oncology (PHO) patients and those undergoing hematopoietic stem cell transplantation (HSCT) often have a central line for medication administration, transfusions, and frequent blood draws and are at risk for bloodstream infections (BSI).^1–4 BSI lead to prolonged hospitalization, intensive care unit (ICU) admissions, extensive antibiotic treatment, increased mortality, and potentially avoidable use of healthcare resources.^5–10

In 2008 the Centers for Medicare and Medicaid Services implemented the Hospital-Acquired Conditions Initiative to improve value and patient safety.¹¹ The initiative decreased reimbursement rates for preventable conditions, which included central line–associated bloodstream infections (CLABSI).¹² BSI occurring in hospitalized patients (day of hospitalization ≥ 3) with a central venous catheter are classified by the National Health Safety Network (NHSN) as either a primary laboratory-confirmed bloodstream infection (CLABSI), secondary BSI, or a common commensal single-positive blood culture (SPBC) which are often referred to as contaminents.^2,5,13 To address nonpreventable BSI thought to occur from translocation of bacteria through a damaged mucosa, in 2015 the NHSN adopted a subclassification of CLABSI defined a new category of infection termed mucosal barrier injury laboratory-confirmed bloodstream infection (MBI-LCBI).^5,6 The NHSN classification of BSI is somewhat complicated, relying on multiple variables including type of organism, secondary site-specific cultures or imaging, absolute neutrophil count (ANC) at the time of infection, and presence of gastrointestinal graft-versus-host disease (GVHD) with associated diarrhea.¹⁴

We have previously made a case that tracking all BSI, not just CLABSI, in PHO and HSCT patients is a patient-centered, clinically relevant approach that could help better assess across-center and within-center differences in infection rates.¹³ This includes tracking not only CLABSI (stratified by MBI-LCBI vs non–MBI-LCBI) and secondary BSI, but also SPBC with NHSN-defined common commensals. The inclusion of SPBC is relevant since these can be real BSI in immunocompromised patients.^3,5 Although there has been progress in the reduction of CLABSI rates using quality improvement methodologies across pediatric centers,^15–18 the majority of data describing outcomes after BSI in these high-risk populations is limited to single-center reviews.

The overall objective of this study was to describe the outcomes of PHO and pediatric HSCT patients who develop any category of healthcare-associated bloodstream infection (those occurring on or after the third day of hospitalization and up to two days after discharge home) as part of a multicenter quality improvement collaborative.

2 |. METHODS

2.1 |. Study design and data source

Starting November of 2009, the Children’s Hospital Association, working with healthcare providers from across the United States, developed a quality improvement network called the Childhood Cancer and Blood Disorder Network that focused on eliminating CLABSI in inpatient PHO and HSCT patients. Starting in 2011, this network started tracking all BSI as per the NHSN definition.^13,17 Thirty-two centers participated in the Childhood Cancer and Blood Disorder Network during the study period.¹⁷ Participating institutions submitted monthly deidentified BSI data to the Children’s Hospital Association, which included patient and organism details related to each BSI event but did not include BSI outcome data.

2.2 |. Participating centers

Seventeen of the 32 centers in the Childhood Cancer and Blood Disorder Network participated in this retrospective analysis of BSI outcomes. Using the reported data, as well as any data maintained at the center to identify individual patients, a systematic chart review was used to determine specific outcomes as described below. Institutional review board approval was obtained from the individual sites and all patient data were deidentified.

2.3 |. Study population

The study population included BSI from November 1, 2011, through April 30, 2015, that were reported to the Children’s Hospital Association collaborative by the 17 participating centers. BSI were included in the analysis as long as NHSN classification and outcomes (death, ICU admission, and central line removal within seven days) were obtained. BSI occurring in the ICU were excluded in the analysis. Additionally, ambulatory BSI, defined as a positive culture within two calendar days from hospital admission and two days after discharge per NHSN guidelines,^5,19 were excluded from analysis. BSI were classified by the individual institutions at the time of infection according to the NHSN criteria at that time.¹⁹ Importantly, SPBC with a common commensal organism) were included in the analysis. Standard NHSN criteria at the time of infection directed if the infection was a new BSI or a continuation of the already classified BSI (not included). Primary analysis was based on the BSI, not the patient, and patients could be included in the analysis more than once.

2.4 |. BSI-associated variables

For each BSI, we evaluated demographic data of the patient at the time of positive culture. This included gender, age, diagnosis, relapse status, time from hospital admission to BSI, the number of central lines, type and location of the central line, and organism(s) isolated from the positive culture. We also included ANC at the time of BSI, the presence of mucositis, diarrhea, and/or skin breakdown. HSCT patients were identified as such if they had undergone an HSCT within 100 days of infection. HSCT patient demographics included preparative regimen, stem cell source, matched versus unmatched donor, time from transplant day to development of BSI, and day +100 GVHD grade.²⁰

2.5 |. Outcome variables

Definitions of adverse outcome included all-cause mortality within 30 days of positive culture, transfer to the ICU within 48 hours of the positive culture, and central line removal within seven days of the positive culture. Outcomes were not mutually exclusive, and all outcomes were “all-cause.” Additionally, we described the percentage of BSI in patients who died within 3 and 10 days after positive culture. In the patients who died within 30 days of BSI, evaluation included if resuscitation limits or a Do Not Resuscitate order was in place at the time of the positive blood culture.

2.6 |. Bloodstream infection classification

Laboratory-confirmed bloodstream infections were prospectively classified according to applicable NHSN definitions in use during the study period as either CLABSI or secondary BSI.¹⁹ Common commensal organisms, as defined by the NHSN (e.g., coagulase-negative staphylococci and viridans group streptococci), cultured from a single blood culture with no other organisms isolated were captured as SPBC.^5,21

2.7 |. MBI-LCBI–like

In January of 2015, the NHSN broadened the CLABSI definition to include the MBI-LCBI subcategorization.⁶ For the purpose of this analysis, an MBI-LCBI–like definition was created and retrospectively applied to describe MBI-LCBI versus non-MBI-LCBI CLABSI. A BSI that met the CLABSI or secondary BSI criteria was defined as MBI-LCBI–like if the BSI organism was defined by the NHSN as an MBI-organism^5,6,21 and neutropenia, an ANC less than 500 cells/mm³, was present on the day of the positive culture.

2.8 |. Statistical analysis

Variables were described with medians and interquartile ranges. Univariate patient demographic data were compared between groups using the Fisher exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. The Fisher exact test was used to compare outcomes between BSI subcategories as well as those with an ANC less than 500 cells/mm³ at the time of the infection.

3 |. RESULTS

A total of 957 BSI were included in the analysis. Seventy-one percent (n = 677) of BSI occurred in PHO patients, 29% (n = 280) were in HSCT patients. Six hundred thirty-eight CLABSI (67%), 142 (15%) secondary BSI, and 177 (18%) SPBC were included (Figure 1). Secondary BSI occurred later during hospitalization at a median of 20 days (IQR 10–56) after hospitalization, with CLABSI occurring at 15 days (IQR 10–23), and SPBC at 11 days (IQR 4–22) (P < 0.001). Thirty-seven percent of all reported BSI (n = 354 of 957) were associated with at least one adverse outcome, with a total of 425 adverse events reported.

FIGURE 1.

Bloodstream infections from 17 centers (November 1, 2011, through April 30, 2015)

3.1 |. Pediatric hematology/oncology (n = 677)

Patient demographics can be found in Table 1. Nearly half of BSI (43%, n = 289) occurring in patients with an ANC of 0 at the time of infection, and 68% (n = 460) had ANC < 500. The median time from day of hospitalization to infection was 13 days (IQR 6–24) (Table 1). Sixty-six percent (n = 445) of the 677 BSI in PHO patients were classified as a CLABSI, 14% (n = 92) were classified as a secondary BSI, and 20% (n = 147) were classified as an SPBC (Figure 1).

TABLE 1.

Demographics of PHO and SCT patients after bloodstream infection (n = 957)

BSI in PHO patients (n = 677)	N (%)
Male	359 (59%)
Median age at time of BSI in years (IQR)	8.7 (3.2–14.7)
Median days post admission to BSI (IQR)	13 (6–24)
Median ANC at time of infection (IQR)	0 (0–500)
Number of central venous catheters
One	634 (93%)
Two or more	43 (7%)
Diagnosis
Acute myelogenous leukemia	196 (33%)
Acute lymphoblastic leukemia	138 (23%)
Non-CNS solid tumor	81 (13%)
CNS tumor	25 (4%)
Immune deficiency	22 (4%)
Lymphoma	20 (3%)
Nonmalignant hematology	17 (3%)
Bone marrow failure	9 (2%)
Unknown	96 (16%)
Relapsed disease	186 (27%)
BSI in Pediatric HSCT Patients (n = 280)
Male	161 (58%)
Median age at time of BSI in years (IQR)	7.9 (2.9–14.8)
Median days post admission to BSI (IQR)	17 (12–37)
Median days post SCT to BSI (IQR)	8 (5–28)
Median ANC at time of BSI (IQR)	0 (0–300)
Number of central venous catheters
One	196 (70%)
Two or more	84 (30%)
Diagnosis
Leukemia	108 (39%)
Immune deficiency	32 (11%)
CNS tumor	24 (9%)
Solid tumor	20 (7%)
Lymphoma	15 (5%)
Bone marrow failure	10 (4%)
Unknown	71 (25%)
Preparative regimen
Myeloablative	189 (68%)
Reduced intensity	69 (25%)
Unknown	22 (7%)
Donor
Allogeneic	218 (83%)
Autologous	62 (17%)
Donor relationship (allogeneic only, n = 218)
Unrelated donor	148 (68%)
Related donor	70 (32%)
Match (allogeneic only, n = 218)
Full match donor	108 (50%)
Mismatched donor	84 (39%)
Unknown	26 (11%)
Day 100 GVHD score (allogeneic, n = 218)
Grade 0	78 (36%)
Grade 1	7 (3%)
Grade 2	8 (3%)
Grade 3	23 (11%)
Grade 4	8 (4%)
Undocumented	94 (43%)

Abbreviations: ANC, absolute neutrophil count; BSI, bloodstream infections; CNS, central nervous system; GVHD, graft-versus-host disease; HSCT, hematopoietic stem cell transplant; IQR, interquartile range.

In BSI occurring in PHO patients, all-cause mortality (3 days, 10 days, and 30 days), ICU admission, and central line removal within seven days were higher in the secondary BSI group than with CLABSI or SPBC. In BSI occurring in HSCT patients, outcomes were similar in all three categories (CLABSI, secondary BSI, and SPBC) (Table 2). Thirty-day all-cause mortality (P = 0.027) and intensive care admission (P = 0.013) were significantly increased in PHO patients with ANC < 500 cells/mm³ at the time of infection (Table 3).

TABLE 2.

Outcomes of bloodstream infections (n = 957) occurring in PHO and HSCT patients

PHO patients (n = 677)	CLABSI (n = 445)	Secondary BSI (n = 92)	SPBC(n = 140)	P value
All-cause mortality
Death within three days (n = 6)	2 (0.4%)	4 (4%)	0	<0.003
Death within 10 days (n = 18)	7 (2%)	9 (10%)	2 (1%)	<0.001
Death within 30 days (n = 42)	24 (5%)	14 (15%)	4 (3%)	<0.001
Intensive care admission (n = 119)	81 (18%)	22 (24%)	16 (11%)	0.03
Line removal within seven days (n = 146)	93 (21%)	35 (35%)	18 (13%)	<0.001
Pediatric HSCT Patients (n = 280)	CLABSI (n = 193)	Secondary BSI (n = 50)	SPBC (n = 37)	P value
All-cause mortality
Death within three days (n = 3)	2 (1%)	0	1 (3%)	0.69
Death within 10 days (n = 8)	5 (3%)	1 (2%)	2 (5%)	0.59
Death within 30 days (n = 15)	10 (6%)	3 (6%)	2 (5%)	0.99
Intensive care admission (n = 46)	32 (17%)	9 (18%)	5 (14%)	0.85
Line removal within seven days (n = 57)	42 (22%)	12 (24%)	3 (8%)	0.13

Abbreviations: BSI, bloodstream infection; CLABSI, central line-associated bloodstream infection; HSCT, hematopoietic stem cell transplant; PHO, pediatric hematology/oncology; SPBC, single positive blood culture.

TABLE 3.

Univariate analysis of outcomes based on ANC (n = 957)

BSI occurring in PHO patients (n = 677)	ANC < 500 cells/mm3 (n = 460)	ANC ≥ 500 cells/mm3 (n = 217)	P value
All-cause mortality within 30 days (n = 42)	35 (8%)	7 (3%)	0.027
Intensive care admission (n = 119)	92 (20%)	27 (12%)	0.013
Line removal within seven days (n = 146)	100 (22%)	46 (21%)	0.92
BSI occurring in pediatric HSCT patients (n = 280)	ANC < 500 cells/mm3 (n = 188)	ANC ≥ 500 cells/mm3 (n = 92)	P value
All-cause mortality within 30 days (n = 15)	13 (7%)	2 (2%)	0.16
Intensive care admission (n = 46)	36 (19%)	10 (11%)	0.088
Line removal within seven days (n = 57)	47 (25%)	10 (11%)	0.007

Abbreviations: ANC, absolute neutrophil count; BSI, bloodstream infection; HSCT, hematopoietic stem cell transplant; PHO, pediatric hematology/oncology.

3.2 |. Pediatric hematopoietic stem cell transplant (n = 280)

Patient demographics can be found in Table 1. Sixty-seven percent (n = 188) of BSI occurred in patients who had an ANC < 500 at time of infection. The majority, 78% (n = 218), of BSI in this group were in patients who underwent an allogeneic transplant. Of these, 68% occurred in patients who underwent myeloablative conditioning. Of the 218 BSI in the allogeneic HSCT recipients, 148 (68%) received an unrelated donor graft (Table 1). Sixty-nine percent (n = 193) of BSI in HSCT recipients were classified as a CLABSI, 18% (n = 50) were classified as a secondary BSI, and 13% (n = 37) were classified as an SPBC (Figure 1).

In BSI occurring in HSCT patients, outcomes were similar in all three categories (CLABSI, secondary BSI, and SPBC) (Table 2). Line removal within seven days was significantly increased (P = 0.007) when BSIs occurred in pediatric HSCT patients with ANC < 500 cells/mm³ at the time of infection (Table 3).

3.3 |. Other outcomes

When BSI were associated with admission to the ICU (n = 165, 17%), the median duration of stay was four days (IQR 2–10). Fourteen (26%) of the 57 patients who died within 30 days of BSI had advanced directives in place at the time of infection.

3.4 |. MBI-LCBI–like demographics and outcomes

Forty-three percent (n = 407) of BSI met criteria for MBI-LCBI–like based on the organism and ANC at the time of the positive culture. In this group, the mean age was 9.8 years (95% CI 9.2–10.6) at time of infection versus a mean age of 8.8 years (95% CI 8.3–9.3) in those without an MBI-LCBI (P = 0.12). The average time to infection from admission in this category was 30 days (95% CI 26–34) compared with 25 days (95% CI 21–28) in non-MBI-LCBI infections (P = 0.02).

Thirty-day all-cause mortality was similar in MBI-LCBI–like BSI (7%) versus non-MBI-LCBI BSI (5%). Central line removal within seven days was also similar (24% vs 19%, P = 0.12). However, admission to the ICU was significantly higher in MBI-LCBI–like infections (21% vs 15%; P = 0.012) (Table 4).

TABLE 4.

Comparison of outcomes in between bloodstream infections classified as MBI-LCBI-like and non-MBI-LCBI in PHO and HSCT patients (n = 957)

	MBI-LCBI-like BSI (n = 407)	Non-MBI-LCBI BSI (n = 550)	P value
All-cause mortality
Death within three days (n = 9)	4 (1%)	5 (1%)	0.91
Death within 10 days (n = 26)	15 (4%)	11 (2%)	0.11
Death within 30 days (n = 57)	28 (7%)	29 (5%)	0.29
Intensive care admission (n = 165)	85 (21%)	80 (15%)	0.012
Line removal within seven days (n = 203)	96 (24%)	107 (19%)	0.12

Abbreviation: BSI, Bloodstream infection; MBI-LCBI, mucosal barrier injury laboratory-confirmed bloodstream infection.

3.5 |. Organisms identified

A total of 1075 organisms were isolated in the 957 infections. Six percent (n = 60) grew two or more organisms at the time of BSI. A list of the organisms can be found in Table 5.

TABLE 5.

Organisms isolated in BSI and outcomes occurring in PHO and HSCT patients who developed a BSI (n = 957)

Organism isolated	Number of isolates (n = 1075)
Acinetobacter species	7 (0.6%)
Bacillus species	43 (4%)
Bacteroides species	3 (0.3%)
Candida species	54 (5%)
Capnocytophagia	1 (0.1%)
Citrobacter freundii	6 (0.6%)
Corynebacterium species	5 (0.5%)
Entercoccus galinarum	16 (1%)
Enterococcus faecalis, vancomycin resistant	31 (3%)
Enterococcus faecalis, vancomycin susceptible	57 (5%)
Enterobacter aerogenes	3 (0.3%)
Enterobacter cloacae	45 (4%)
Escherichia coli	89 (8%)
Gram positive coccus, not otherwise specified	14 (1%)
Klebsiella oxytoca	12 (1%)
Klebsiella pneumonia	78 (7%)
Lactobacillus species	9 (0.8%)
Micrococcus species	2 (0.2%)
Other	100 (9%)
Pseudomonas aeruginosa	56 (5%)
Rothia	1 (0.1%)
Serratia marcescens	3 (0.3%)
Staphylococcus aureus, methicillin resistant	20 (2%)
Staphylococcus aureus, methicillin sensitive	56 (5%)
Staphylococcus species, coagulase negative	150 (14%)
Stenotrophomonas maltophilia	10 (1%)
Streptococcus agalactiae	1 (0.1%)
Streptococcus pneumoniae	7 (0.7%)
Streptococcus pyogenes	2 (0.2%)
Streptococcus species	25 (2%)
Streptococcus viridans	167 (16%)
Yeast, not otherwise specified	2 (0.2%)

4 |. DISCUSSION

In this study, we evaluated the outcomes of 957 blood culture–positive BSI that occurred in PHO and pediatric HSCT patients over a 4.5-year period across 17 institutions. Our data show that nearly 40% of all BSI were tied to at least one adverse outcome, which constitutes a significant burden to the PHO and pediatric HSCT population. These data will assist in determining the need for quality improvement and research efforts to decrease BSI in this patient population.

Six percent (n = 57) of the BSIs included in the analysis occurred in patients who died within 30 days of the infection. To account for disease progression as the cause of death, we determined which patients had advanced directives in place at the time of infection (n = 14). It is difficult, however, to measure the impact of BSI on survival; in HSCT patients, BSI are an independent risk factor for mortality, independent of death directly related to infection^2,22–26; and in PHO patients, BSI are associated with increased mortality.^{2,3,24,27–34} It is possible patients who develop BSI develop perturbations in gut microbiome secondary to antibiotic use, along with intestinal inflammation.^35–37 This can lead to intestinal domination of pathogenic bacteria^38–41 and translocation of bacteria through the mucosal barrier causing further infection (MBI-LCBI).^7,42,43 Interestingly, 43% of the blood cultures we evaluated met criteria for MBI-LCBI–like BSI. MBI-LCBI are a substantial burden in the PHO/HSCT population and lead to significant morbidity and mortality and healthcare resource utilization.² Fluoroquinolone prophylaxis has been shown to decrease bacteremia in pediatric oncology patients⁴⁴; however, further understanding of the pathophysiology and prevention of MBI-LCBI is needed.⁴⁵

Seventeen percent of BSI occurred in patients who were subsequently admitted to the ICU with a median ICU stay of four days, and 21% of BSI were associated with central line removal. BSI have been shown to be associated with increased length of hospital stay as PHO patients with CLABSI have an average length of stay 21 days longer than PHO patients without CLABSI.^46,47 A recently published meta-analysis of healthcare-associated infections revealed that CLABSI are associated with increased healthcare-associated infection costs, averaging over $45 000 per event.⁴⁸ This is further supported by a recent cost-analysis evaluation of BSI in ambulatory PHO and HSCT patients who demonstrated a cost of over $40 000 for each BSI.⁴⁹ Finally, Wilson et al utilized propensity scoring with matched cases, while controlling for other covariates, and defined the attributable cost of a CLABSI to approximately $70 000 per event in PHO patients.⁴⁷ Primary prevention of BSI, through research and quality improvement methodologies, is likely a substantial fiscal strategy for institutions.

An important goal of our study was to evaluate the outcomes of hospitalized PHO and pediatric HSCT patients with SPBC, our data demonstrate that the incidence of adverse outcomes after SPBC is not trivial. Common commensals with multiple positive blood cultures are classified as true pathogens; however, those with only one positive blood culture (SPBC) are considered contaminants by the NHSN.^14,50 In the immunocompromised patient population, however, common commensals are often the pathogenic organism.^2,51,52 It is important to note that although SPBC are not routinely followed as a patient safety measure, adverse outcomes may occur in the PHO and HSCT populations after these BSI. Further investigation is needed to delineate mechanisms for clinicians to determine which SPBC are true pathogens versus blood culture contamination.

There are a number of limitations to this study. First, the outcome review of BSI was done retrospectively across multiple institutions, which carries inherent limitations. Each BSI was categorized as an independent event; however, repeat episodes in the same patient do not necessarily constitute independent events, especially if the central line was left in place, although the institutions followed the existing NHSN classification guidelines at the time of the BSI. Also, there were no uninfected controls and we were unable to identify patients who had more than one infection, thus limiting our ability to fully evaluate attributable outcomes in this cohort. Additionally, it is difficult to ascertain if BSI are a direct causation of adverse outcomes, particularly mortality. Finally, retrospective classification of MBI-LCBI is imperfect, likely over calling some MBI-LCBI (only using one day of neutropenia instead of two) and under call others (not applying the gastrointestinal GVHD criteria). Despite these limitations, the data clearly show a heavy burden on healthcare utilization from BSI that merits further attention toward potential strategies to reduce these infections.

In summary, the data show adverse outcomes in 37% of BSI occurring in hospitalized PHO and pediatric HSCT patients. CLABSI (including MBI-LCBI), secondary BSI, and SPBC lead to significant morbidity, mortality, and healthcare resource utilization in this patient population. As evidenced by the data, reduction in the frequency of BSI in at-risk populations should be a major public health and scientific priority.

Abbreviations:

ANC

absolute neutrophil count

BSI

bloodstream infection

CLABSI

central line–associated bloodstream infection

HSCT

hematopoietic stem cell transplant

ICU

intensive care unit

IQR

interquartile range

MBI-LCBI

mucosal barrier injury laboratory-confirmed bloodstream infection

NHSN

National Health Safety Network

PHO

pediatric hematology/oncology

SPBC

single-positive blood culture

DATA AVAILABILITY

The data that support the findings of this study are available from the corresponding author, CED, upon reasonable request.