Comparison of catheter-related bloodstream infection rates in pediatric patients receiving parenteral nutrition with soybean oil-based intravenous fat emulsion versus a mixed oil fat emulsion

Abstract

Study Objective:

To compare rates of catheter-related bloodstream infections (CR-BSI) in pediatric patients who received parenteral nutrition (PN) with either soybean oil-based intravenous fat emulsion (SO-IFE) or mixed oil IFE (MO-IFE). We hypothesized that the use of MO-IFE would be independently associated with reduced infection rates compared to SO-IFE.

Design:

Retrospective cohort study.

Setting:

Tertiary referral children’s hospital and its associated gastrointestinal rehabilitation clinic (01/01/2015 – 07/31/2019).

Patients:

Days of IFE exposure were counted for patients aged < 18 years on IFE initiated during the review period, who had a central venous catheter (CVC) placed for PN administration, received IFE at least three times weekly, and for at least 7 days.

Measurements:

The primary outcome included total and categorical CR-BSI rates expressed as the average with standard error (SE) number of infections per 1000 fat emulsion days. The following categories were specified: Candida albicans, non-albicans Candida spp., coagulase-negative Staphylococcus (CoNS), Enterobacterales, methicillin-resistant S. aureus, methicillin-susceptible S. aureus, and Pseudomonadales. Average infection rate comparisons were quantified as incidence rate ratios (IRR) using generalized linear mixed modeling with a Poisson distribution.

Main Results:

743 SO-IFE and 450 MO-IFE exposures were reviewed from 1131 patients, totaling 37,599 and 19,796 days of therapy, respectively. From those found significantly different, the average rate of infections with CoNS was 3.58 (SE 0.5)/1000 days of SO-IFE and 1.39 (SE 0.45)/1000 days of MO-IFE (IRR [95% confidence interval (CI)]: 0.27 [0.16–0.46]; p < 0.01). Total average rates of infection were 7.33 (SE 0.76)/1000 days of SO-IFE and 4.52 (SE 0.75)/1000 days of MO-IFE (IRR [95% CI]: 0.60 [0.44–0.81]; p < 0.01). Other factors associated with higher infection rates include female gender, neonatal age, and inpatient-only IFE exposure.

Conclusion:

Receipt of MO-IFE was associated with lower rates of CoNS and total CR-BSIs compared to SO-IFE in pediatric patients. These findings could have major implications on IFE selection for pediatric patients receiving PN.

1 |. INTRODUCTION

1.1 |. Background

Catheter-related bloodstream infections (CR-BSIs) are a frequent complication for patients receiving parenteral nutrition (PN) with central venous catheters (CVCs) placed for administration.¹ Despite several preventative measures taken over time to reduce CR-BSI incidence, receipt of PN remains an independent risk factor for infection regardless of age or setting.^2–5 Specific PN composition is tailored to meet each patient’s nutritional needs, but generally contain amino acids, dextrose, electrolytes, fat emulsion, and minerals. Macronutrient composition is likely an important factor as increased infection risk has been associated with increased parenteral calories.⁶

The most common microbes associated with CR-BSIs are coagulase-negative staphylococci (CoNS), Staphylococcus aureus, Candida species (spp.), and Enterobacterales as noted in multiple clinical practice guidelines.^7–8 These microbes are capable of robust biofilm formation on indwelling medical devices (e.g., CVCs) commonly used in vulnerable patient populations. This highly organized, multicellular biofilm structure is typically encased in a dense extracellular matrix that impedes removal of microorganisms from intravenous lines and further exacerbates antimicrobial treatment failure.^9–11 Of the macronutrients, prior evidence has suggested that receipt of intravenous fat emulsion (IFE) is associated with increased infection incidence likely secondary to the promotion of biofilm formation.^12–14

Soybean-oil IFE (SO-IFE) was the first IFE approved in 1975 and has been the predominant intravenous fat source in the United States.¹⁵ However, starting in 2016, newer IFEs have been approved including mixed-oil IFE (MO-IFE) and fish-oil based IFE (FO-IFE).¹⁶ Most studies associating IFE or PN receipt to CR-BSI incidence in the United States were done during SO-IFE predominance. Thus, there is limited evidence examining whether newer fat emulsions impact CR-BSI rates.

Recent in vitro investigations have demonstrated that biofilm formation propensity varies in the presence of fatty acids.^17,18 SO-IFE and FO-IFE were observed to stimulate C. albicans biofilm formation; whereas MO-IFE inhibited biofilm formation comparatively by approximately 50%. This was attributed to the presence of capric acid in MO-IFE which dose-dependently reduces biofilm formation through inhibition of hyphae formation and elongation.¹⁷ For S. aureus, an in vitro model demonstrated reduction in biofilm formation with exposure to herring oil, which has an abundance of docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA).¹⁸ Notably, DHA and EPA are in MO-IFE and FO-IFE, but absent in SO-IFE.¹⁵ These findings suggest that newer IFEs and their varied fatty acid composition may impact CR-BSI rates and the distribution of microbes encountered.

1.1 |. Objectives

The objective of this study was to compare CR-BSI rates in pediatric patients receiving PN with SO-IFE or MO-IFE following adjustment for other characteristics potentially impacting CR-BSI rates. We hypothesized that the use of MO-IFE would be independently associated with reduced infection rates compared to SO-IFE.

2 |. METHODS

2.1 |. Setting

This was a retrospective cohort study of pediatric patients who received PN at Le Bonheur Children’s Hospital (LBCH), a tertiary care children’s hospital in Memphis, TN, or had outpatient PN management by the Children’s Intestinal Rehabilitation Clinic at Le Bonheur (CIRCLe). Inpatients who received PN at LBCH primarily included those requiring intensive care (i.e., cardiovascular, neonatal, and pediatric intensive care units) or outpatients with intestinal failure. LBCH does not directly manage hematology/oncology patients unless they require management for acute problems not directly focused on their hematologic/oncologic condition or admission to an intensive care unit. Following approval, MO-IFE was used in select patients for PN-associated liver disease (PNALD) resolution. After January 2018, MO-IFE was the IFE formulation of choice for patients receiving PN at LBCH following reports that demonstrated reduced rates of developing PNALD compared to SO-IFE.^19,20 LBCH and CIRCLe are part of Methodist Le Bonheur Healthcare (MLH), a comprehensive health system serving the greater Memphis area that uses electronic health records (EHRs) which integrate information from inpatient and outpatient encounters within MLH. The study protocol underwent expedited review and was approved by the University of Tennessee Health Science Center (UTHSC) Institutional Review Board.

2.2 |. Participants and Study Design

Inpatient and outpatient IFE exposure from January 1, 2015 to July 31, 2019 was reviewed. Days of exposure were counted if patients were aged < 18 years at IFE initiation during the review period, had a CVC placed for PN administration, received IFE at least three times weekly, and for at least for 7 days. CR-BSI occurrence was noted during the exposure period and identified by collection of a positive blood culture with no other identifiable source of infection and receipt of at least 7 days of appropriate antimicrobial therapy, in congruence with a prior definition.^21,22 LBCH inpatients received IFE separately from amino acids and dextrose (2-in-1); however, outpatients may have received admixtures containing all three macronutrients depending on outpatient pharmacy compounding practice and patient/parent preference. Ethanol lock therapy (ELT) is typically employed in patients with long-term CVCs and recurrent CR-BSIs, which applies to the PN patients managed as outpatients.²³ Of note, no changes in practice with antimicrobial lock use, management of CR-BSI caused by CoNS, and neonatal intensive care unit (NICU) antifungal prophylaxis occurred during the review period. Referring institutional exposure and infection occurrence were also recorded if PN with IFE was started at a separate institution prior to being transferred to LBCH, the same fat emulsion was continued within 48 hours of transfer to LBCH, and microbiologic culture information from the referring institution was available for review. Exposure was followed until IFE discontinuation at LBCH, transfer to another facility outside of MLH with no readmittance, or loss of follow-up with CIRCLe.

2.3 |. Outcomes and Variables

Overall IFE exposure during the review period was recorded in days and grouped as SO-IFE or MO-IFE for each patient. CR-BSIs that occurred during the exposure period were counted and assigned to the IFE administered at the time of infection. As patients who received both formulations were included, total IFE exposure and CR-BSI data were recorded separately for each formulation. Total CR-BSI count was noted between the two formulations and major categories of infections caused by C. albicans, non-albicans Candida spp. (NAC), coagulase-negative Staphylococcus (CoNS; e.g., S. epidermidis, S. capitis, S. haemolyticus, S. hominis, and S. lugdunensis), Enterobacterales (e.g., Enterobacter spp., Escherichia coli, Klebsiella spp., Proteus spp.), Enterococcus spp., methicillin-resistant S. aureus (MRSA), methicillin-susceptible S. aureus (MSSA), or Pseudomonadales (e.g., Acinetobacter spp., Pseudomonas spp.). Polymicrobial infections were only counted once for analyses regarding total infections but were included for each categorical infection rate analysis based on involved pathogens. The primary outcome was the categorical and total CR-BSI rates between IFE formulations. Assessing the primary outcome as a rate was chosen to normalize the duration of IFE exposure to CR-BSIs as similarly done for other investigations examining infection occurrence with consideration of varying durations of CVC placement.^2,24 Additional factors recorded included patient age at the start of IFE therapy during the review period; categorized as neonate, infant, child, or adolescent based on United States Food and Drug Administration (FDA) groupings of pediatric subpopulations; gender; and inclusion of outpatient and referring institutional exposure.²⁵

2.4 |. Statistical Methods

Cohort characteristics by IFE course were initially evaluated as proportions with chi-square test for categorical variables or median and interquartile ranges (IQR) with Mann-Whitney U-test for continuous variables. Infection rates are reported as averages with standard error (SE).To account for multiple exposures per patient and variability in infection risk if exposure were continued from referring institutions, both were incorporated as random effects to a generalized linear mixed model (GLMM) with a Poisson distribution and a log link function that assessed differences in rates of infections between the two formulations as incidence rate ratios (IRR) with 95% confidence intervals (CI). To account for differences in the duration of IFE exposure between patients, the models were offset by the duration of exposure. Unadjusted models were developed first to determine differences in infection rates using IFE formulation as the only fixed effect. The significance level was set at two-sided p < 0.05. For infection categories that demonstrated a statistically significant difference in the rates of infections between the two formulations on unadjusted models; age category, gender, and outpatient PN receipt were incorporated as additional fixed effects for model adjustment. Analyses and graphics development were performed in R (version 4.2.1; 2022; The R Foundation for Statistical Computing; Vienna, Austria) and SAS (version 9.44; 2022; SAS Institute Inc.; Cary, North Carolina).

3 |. RESULTS

3.1 |. Exposure Characteristics

Based on our inclusion criteria, 743 SO-IFE and 450 MO-IFE exposures were reviewed from 1131 patients, totaling 37,599 and 19,796 days of therapy, respectively. 62 patients received both formulations during the study period with sub-totals of 12,490 days of SO-IFE and 8,696 days of MO-IFE. Summary statistics by the IFE course are outlined in Table 1. Proportions of exposure characteristics and median duration of exposure did not significantly differ between the two IFE groups.

TABLE 1.

Characteristics by type of fat emulsion course

	SO-IFE (n=743)	MO-IFE (n= 450)	p-value
Age at start of IFE therapy, n (%)			0.07
Neonate (age < 28 days)	498 (67%)	277 (61.6%)
Infant (age ≥ 28 days/< 2 years)	129 (17.4%)	106 (23.6%)
Child (age ≥ 2 years/< 12 years)	65 (8.7%)	37 (8.2%)
Adolescent (age ≥ 12 years/< 18 years)	51 (6.9%)	30 (6.7%)
Male, n (%)	398 (53.6%)	255 (56.7%)	0.29
Outpatient PN, n (%)	39 (5.2%)	18 (4%)	0.26
Referring institution exposure, n (%)	37 (5%)	18 (4%)	0.43
IFE exposure (days), median (IQR)	17.5 (11–37)	18.5 (10.5–41)	0.86

Abbreviations: IFE, intravenous fat emulsion; MO-IFE, mixed oil-IFE; PN, parenteral nutrition; SO-IFE, soybean oil-IFE

3.2 |. Main Results and Outcomes

Total and categorical CR-BSI counts by IFE formulation are provided in Table 2 and infection rates are illustrated in Figure 1. Overall, average infection rate was 6.27 (SE 0.55)/1000 fat emulsion days, and infections were most frequently caused by CoNS, Enterobacterales, and S. aureus at average rates of 2.75 (SE 0.35), 1.25 (SE 0.21), and 0.87 (SE 0.17)/1000 fat emulsion days, respectively. For patients who received both formulations, total average infections rates were 4.85 (SE 1.16)/1000 fat emulsion days on SO-IFE and 3.26 (SE 0.89)/1000 fat emulsion days on MO-IFE, and CoNS average infections rates were 2.35 (SE 0.82)/1000 fat emulsion days and 0.31 (SE 0.2)/1000 fat emulsion days on SO-IFE and MO-IFE, respectively. Forty polymicrobial infections (12.4% of all infections) were noted, with 32 occurring during SO-IFE administration. The most common microbes isolated from polymicrobial infections included CoNS, Enterobacterales, and Enterococcus spp. in 18, 15, and 12 infections, respectively.

TABLE 2.

Total^a and categorical catheter-related bloodstream infection counts

Infection Categories	Count on SO-ILE (n = 238)	Count on MO-IFE (n=85)
Candida albicans	15	3
Coagulase-negative Staphylococcus	102	18
Enterobacterales	62	28
Enterococcus spp.	19	12
Methicillin-resistant S. aureus	13	12
Methicillin-susceptible S. aureus	24	7
non-albicans Candida spp.	15	8
Pseudomonadales	8	3

Abbreviations: IFE, intravenous fat emulsion; MO-IFE, mixed oil-IFE; SO-IFE, soybean oil-IFE

^aTotal counts not summative of categories as co-infections counted as one

FIGURE 1.

Total^a and categorical average catheter-related bloodstream infection rates based on IFE formulation^b

Abbreviations: CA, Candida albicans; CoNS, coagulase-negative Staphylococcus; EB, Enterobacterales; EC, Enterococcus; IFE, intravenous fat emulsion; MO-IFE, mixed oil-IFE; MRSA, methicillin-resistant S. aureus; MSSA, methicillin-susceptible S. aureus; NAC, non-albicans Candida spp.; PM, Pseudomonadales; SO-IFE, soybean oil-IFE

*, P < 0.05 using unadjusted Poisson regression models

^aTotal rates not summative of categories as co-infections counted as one

^bRates expressed as average +/− SE

From those found significantly different on unadjusted models, the average rate of infections with CoNS was 3.58 (SE 0.5)/1000 days of SO-IFE and 1.39 (SE 0.45)/1000 days of MO-IFE (IRR [95% confidence interval (CI)]: 0.27 [0.16–0.46]; p < 0.01). Total average rates of infection were 7.33 (SE 0.76)/1000 days of SO-IFE and 4.52 (SE 0.75)/1000 days of MO-IFE (IRR [95% CI]: 0.60 [0.44–0.81]; p < 0.01). Other infection rates were not statistically different between the two IFE formulations. However, most other infection rates were numerically lower with MO-IFE receipt compared to SO-IFE (Figure 1). Tables 3 and 4 summarize the mixed-effects Poisson regression models incorporating exposure characteristics as possible associations to total and CoNS CR-BSI infection rates.

TABLE 3.

Associations with total CR-BSIs on adjusted Poisson regression models

Characteristic	Incidence Rate Ratio (95% Confidence Interval)	p-value
Age at start of IFE therapy
Neonate	Reference
Infant	0.98 (0.74–1.31)	0.90
Child	0.49 (0.26–0.93)	0.03
Adolescent	0.62 (0.33–1.18)	0.15
Male	0.75 (0.59–0.96)	0.02
Outpatient PN	0.67 (0.48–0.95)	0.03
MO-IFE	0.60 (0.44–0.81)	< 0.01

Abbreviations: CR-BSI, catheter-related bloodstream infection; IFE, intravenous fat emulsion; MO-IFE, mixed oil-IFE; PN, parenteral nutrition

TABLE 4.

Associations with CoNS CR-BSI on adjusted Poisson regression models

Characteristic	Incidence Rate Ratio (95% Confidence Interval)	p-value
Age at start of IFE therapy
Neonate	Reference
Infant	1.03 (0.67–1.56)	0.91
Child	0.46 (0.18–1.14)	0.09
Adolescent	0.14 (0.02–1.04)	0.05
Male	0.97 (0.68–1.39)	0.86
Outpatient PN	0.34 (0.2–0.6)	< 0.01
MO-IFE	0.27 (0.16–0.46)	< 0.01

Abbreviations: CoNS, coagulase-negative Staphylococcus; CR-BSI, catheter-related bloodstream infection; IFE, intravenous fat emulsion; MO-IFE, mixed oil-IFE; PN, parenteral nutrition

For both CoNS and total CR-BSIs, receipt of MO-IFE and inclusion of outpatient PN were independently associated with reduced infection rates. On adjusted models, patients who received MO-IFE were 73% less likely to experience a CoNS CR-BSI (Table 4), and 40% less likely to experience any CR-BSI than those who received SO-IFE (Table 3). Total average infection rates for patients with outpatient IFE exposure and only inpatient exposure were 5.11 (SE 1.09) and 6.33 (SE 0.58)/1000 fat emulsion days, respectively; meanwhile, CoNS average infection rates were 0.81 (SE 0.28)/1000 fat emulsion days for outpatient IFE and 2.85 (SE 0.37)/1000 fat emulsion days for only inpatient IFE exposure. Male gender and child age- compared to neonates- were also associated with reduced total average infection rates in the adjusted model. The rate of infections in male patients was 5.18 (SE 0.64)/1000 fat emulsion days and in female patients was 7.58 (SE 0.95)/1000 fat emulsion days. Rates of infections for patients who started PN as neonates, infants, children, and adolescents were 6.31 (SE 0.7), 7.65 (SE 1.36), 4.87 (SE 1.74), and 3.65 (SE 1.21)/1000 fat emulsion days, respectively.

4 |. DISCUSSION

4.1 |. Key Results

To the best of our knowledge, this is the first study comparing CR-BSI rates in pediatric patients receiving PN with commercially available SO-IFE and MO-IFE. Consistent with our hypothesis, the overall infection rate was lower in patients receiving MO-IFE as part of their PN compared to SO-IFE. This difference was largely attributable to lower rates of infections caused by CoNS during the administration of MO-IFE compared to SO-IFE. Age, gender, and outpatient PN receipt also impacted infection rates.

4.2 |. Generalizability and Limitations

As previously stated, there is limited evidence examining whether newer fat emulsions impact CR-BSI rates. However, previous investigations noting differences in infection incidence with different IFE compositions have been reported.²⁶ Notable fatty acids present in MO-IFE and absent in SO-IFE include medium chain triglycerides (MCT; e.g., capric and caprylic acid), DHA, and EPA.¹⁵ Comparisons of outcomes in patients who received SO-IFE compared to those who received IFE formulations with MCT were unable to determine differences in infection incidence.^27–30 Studies comparing outcomes with receipt of FO-IFE compared to SO-IFE also produced inconsistent findings related to infection occurrence.^31–36 However, these studies solely examined inpatient exposure in adults and most did not establish infection differences as a primary end point, reported on any type of infection, included septic patients prior to IFE exposure, used small sample sizes, or only examined outcomes to a pre-determined duration.

Aside from literature in adult patients, a recent report examined differences in the incidence of varied complications associated with MO-IFE versus SO-IFE in NICU patients.³⁷ Of these, the incidence of sepsis was reduced in patients who received MO-IFE compared to SO-IFE. However, sepsis incidence was noted via chart documentation without microbiologic confirmation. The primary end point of the study was the incidence of PNALD with sepsis incidence as secondary, and adjustment based on the duration of IFE was not performed. Our study was able to address most of the previously described deficiencies through rate-based analysis focusing on overall IFE exposure, obtaining a sufficient count of infections to allow for adequate statistical testing, and inclusion of multiple settings. A proportional analysis was another option to assess outcomes and help determine appropriate empiric antimicrobial choices for possible infections.²² However, rate-based analyses are often utilized for assessing impact of interventions on CR-BSIs over extended time periods to account for differences in exposure and multiple infections per patients.^2,24

Our findings are in line with in vitro investigations suggesting that the presence of select fatty acids influences microbial virulence and propensity for causing CR-BSIs for C. albicans and S. aureus.^17,18 Among the microbes analyzed in the present analysis, CoNS demonstrated differences in infection rates between SO-IFE and MO-IFE; however, C. albicans and S. aureus did not, despite the inclusion of MCT and fish oil in MO-IFE. Our working definition for CR-BSIs may have been sensitive for infections from common commensals such as CoNS, but our practice for the management of CR-BSI by common commensals did not change during the review period.^21,22 Alternatively, this was a single-center study, so the number of other infections, especially with C. albicans and S. aureus, was low and likely unable to detect significant differences, despite numerically lower infection rates with MO-IFE. This could be addressed by capturing more infections through multi-center extension.

Other notable findings from the adjusted model for total CR-BSIs included differences in infection rates based on age and gender. The first is rationalized by the greater proportion of patients who were started on PN as neonates regardless of IFE formulation and previous associations demonstrating increased infection risk in neonates compared to children.^3,38 The second is noteworthy as previous univariate and multivariate assessments have not associated differences in infection risk with gender.^3,5,38 The possible reasons behind these associations require further investigation.

Lastly, an intriguing finding was the significantly reduced rates of infections with receipt of outpatient PN. As previously mentioned, the majority of our PN patients managed as outpatients received prophylactic ELT, which could rationalize a lower rate of infections. Inpatient risk factors may have influenced CR-BSI occurrences, such as additional intravenous medication receipt, critical illness impact on the immune system, and post-operative status. Other time-varying covariates and factors not effectively captured with rate-based analyses including central line type, duration of placement, and insertion site; ELT frequency; indication for PN; and prior antimicrobial exposure change with each infection and may confound study findings.^3,4,5,38 Nonetheless, both patients who received IFE as outpatients and solely as inpatients were analyzed and inpatients did not routinely receive ELT or other antimicrobial lock therapy, minimizing the impact of ELT receipt on our findings. Other limitations not previously discussed include the single center setting, lack of broad applicability to hematology/oncology patients, and retrospective cohort design. In addition, we included multiple patient types with regards to age and referring institutional exposure which poses additional confounders, but these confounders were controlled with adjusted multivariable regression models to increase the applicability of our findings to a larger group of pediatric patients.

5 |. CONCLUSION

This study found that receipt of MO-IFE was associated with lower rates of CoNS and total CR-BSIs compared to SO-IFE in pediatric patients. The lack of statistical differences found with other groups of pathogens may be due to a small number of included infections, and additional characteristics should be assessed with each infection to address potential confounders. Given the difference in rates of infections between the two IFE formulations, in vitro investigations with CoNS are warranted to determine the influence that select fatty acids present in MO-IFE have on CoNS biology and biofilm formation relevant to CR-BSIs. Together, these findings could have major implications on IFE selection for pediatric patients receiving PN, empiric antimicrobial regimens used for CR-BSIs, and antimicrobial lock formulations to aid infection prevention.