Contaminant Organism Growth in Febrile Infants at Low Risk for Invasive Bacterial Infection

Abstract

In this multicenter cross-sectional secondary analysis of 4042 low-risk febrile infants, nearly 10% had a contaminated culture obtained during their evaluation (4.9% of blood cultures, 5.0% of urine cultures, and 1.8% of cerebrospinal fluid cultures). Our findings have important implications for improving sterile technique and reducing unnecessary cultures.

Introduction

The evaluation of well-appearing febrile infants ≤60 days old is a common challenge for physicians in the emergency department (ED). Although the majority of these infants have a self-limited viral infection, up to 10% may have a urinary tract infection (UTI) and 2–3% may have an invasive bacterial infection (IBI), specifically bacteremia or bacterial meningitis.(1,2) Evidence-based guidelines rely on infant age and screening blood and urine tests to stratify UTI and IBI risk.(2) Infants found to be at low-risk for IBI (i.e., “low-risk febrile infants”) are not recommended to have cerebrospinal fluid (CSF) cultures obtained,(2) yet some low-risk febrile infants still do.(3) Until recently, urine and blood cultures were recommended in the initial evaluation for all febrile infants. Although guidelines continue to recommend blood cultures in all infants ≤60 days with fever, recent evidence supports limiting urine culture acquisition in this population to only those with abnormal screening urinalysis results.(2)

The recommendation to limit urine and CSF cultures stems from a recognition of the potential harm of contaminated cultures.(2) This may be of particular concern in fast-paced, high-volume environments such as the ED.(4) Furthermore, false-positive culture results often require a return visit, additional testing, the administration of unnecessary antibiotics, and hospitalization while awaiting confirmation of contamination.(5) This cascade of interventions can increase stress and cost for patients, families, and the health care system.(6)

To better understand the burden of contaminated cultures in low-risk febrile infants, we aimed to describe the proportion of contaminated blood, urine, and CSF cultures in low-risk febrile infants in a large, multicenter sample and identify factors associated with contaminant growth. To further characterize the potential impact of contaminant growth, we compared repeat ED visits and hospitalizations within 72 hours for discharged infants with and without contaminant growth.

Methods

We conducted a secondary analysis of a cross-sectional, 34-site study of well-appearing, low-risk febrile infants 29 to 60 days old. Briefly, data were collected retrospectively through electronic health record review on infants evaluated in the ED between 2018–2019, as described previously.(3) Infants were included if they had a fever (temperature of ≥38°C reported in the 24 hours before the ED visit or while in the ED) and were found to be at low risk for IBI based on their history, physical exam, and results of screening blood and urine testing, using site-specific criteria and common risk stratification algorithms, as described previously.(3) This study was approved by the institutional review boards at participating institutions. We followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines.(7)

Our primary outcome was contaminant organism growth in blood, urine, and CSF cultures obtained at the index ED visit. Blood and CSF cultures were considered contaminated if there was provider documentation that the growth was considered a contaminant, if the infant did not receive a full treatment course of antibiotics, or if there was growth of an a priori defined contaminant organism (coagulase-negative staphylococci, viridans group streptococci, or growth of multiple organisms; Table 1). Blood and CSF cultures were considered positive for bacteremia and bacterial meningitis, respectively, in the case of any other organism growth and negative if there was no growth. Urine cultures were considered positive for UTI if there was growth of at least 50,000 colony-forming units (cfu)/mL of a pathogenic organism (or at least 10,000 cfu/mL if the specimen was obtained by suprapubic catheterization), which was the accepted definition during the study time period.(8) Urine cultures were considered contaminated if there was growth of mixed flora or multiple organisms, growth of less than 10,000 cfu/mL of any organism, or if there was provider documentation that the growth was considered a contaminant. Urine cultures were considered negative if there was no organism growth. All urinalyses from this sample of low-risk infants had to be negative for leukocyte esterase, nitrite, and pyuria.(3) As such, urine cultures with 10,000 – 49,999 cfu/mL growth of a single organism were considered “likely contaminant growth” (Table 1).(2) Infants with likely contaminant growth were not included in the analyses comparing infants with negative and contaminated urine cultures, as they could not be definitely determined not to have a UTI. Infants were considered to have been evaluated during daytime hours if they presented to the ED between 4am and 4pm and nighttime hours if they presented to the ED outside of this window. For infants discharged from the index ED visit, we assessed return ED visits and readmissions within 72 hours.

Table 1.

Organism growth by culture type

	No.
Blood culture contaminant organisms 1
Coagulase-negative staphylococci	118
Viridans group streptococci	31
Multiple organisms	2
Gram positive cocci	5
Micrococcus species	5
Bacillus species	4
Diptheroids	3
Propiobacterium species	3
Staphylococcus aureus	2
Paenibacillus species	2
Neisseria sicca	1
Rothia species	1
Veillonella species	1
Unable to determine	1
Blood culture pathogens
Group B Streptococcus2	10
Escherichia coli 2	3
Streptococcus pneumoniae	3
Enterococcus faecalis 3	2
Staphylococcus aureus	1
Elizabethkingia meningoseptica	1
Chryseobacterium indologenes	1
Urine culture “likely contaminants” 4
Escherichia coli	15
Klebsiella pneumoniae	6
Enterococcus faecalis	4
Group B Streptococcus	3
Staphylococcus aureus	2
Enterobacter species	2
Lactobacillus species	1
Proteus mirabilis	1
Coagulase-negative staphylococci	1
Urine culture pathogens 5
Escherichia coli	9
Enterococcus faecalis	4
Group B Streptococcus	3
Klebsiella pneumoniae	2
Bacillus species	1
Citrobacter species	1
Lactobacillus species	1
CSF culture contaminants 1
Coagulase-negative staphylococci	7
Virdans group streptococci	3
Bacillus species	1
CSF culture pathogens (meningitis)
Escherichia coli 2	2
Group B Streptococcus2	1

¹Defined as: growth of a contaminant organism (coagulase-negative staphylococcus, viridans group streptococcus, growth of multiple organisms), provider documentation that the organism was considered a contaminant, or if the infant did not receive a full treatment course of antibiotics

²1 infant had E. coli bacteremia and meningitis; 1 infant had GBS bacteremia and meningitis

³1 infant had enterococcus bacteremia and urinary tract infection

⁴Defined as growth of 10,000–49,999 colony forming units per milliliter of a single organism in the setting of negative urinalysis (study inclusion criteria). An additional 185 infants had definite contaminant growth, defined by growth of mixed flora (n = 52) or multiple organisms (n = 6), growth of less than 10,000 colony forming unites per mililiter (n = 128), and/or documentation that the culture was considered contaminated (n = 157).

⁵Defined as growth of at least 50,000 colony forming units per mililiter (or at least 10,000 colony forming units if the sample was obtained by suprapubic aspiration)

We calculated frequencies and proportions for categorical variables. For our primary outcome, we calculated proportions at the overall and site level. We calculated chi-square statistics for comparisons between dichotomous variables. A p-value <0.05 was considered statistically significant. All data analyses were performed using R Statistical Software (v4.2.1; R Core Team 2022).

Results

Blood Cultures

Of 4042 infants meeting inclusion criteria, 3645 (90.2%) had a blood culture obtained. Twenty-two infants (0.6%) had bacteremia (Table 1). Of the 3623 infants without bacteremia (99.4%), 179 (4.9%) had a contaminated blood culture (range 0 – 42.9% by site, Figure 1a). Contaminated blood cultures were more likely in infants who presented to the ED during daytime compared to nighttime hours (5.9% vs 4.3%, p = 0.03); there was no association between day of week and the proportion of contaminated blood cultures. Compared to infants with a negative culture, infants with contaminant growth were more likely to have a repeat ED visit within 72 hours of discharge from the index ED visit (73.9% vs 7.8%, p < 0.001, Table 2).

Figure 1.

Contaminated blood and urine cultures by site

This figure shows the percentage of infants with contaminated blood (Panel A) and urine (Panel B) cultures by site. Error bars demonstrate 95% confidence intervals.

Table 2.

Characteristics of low-risk febrile infants without bacteremia, urinary tract infection, or meningitis, by culture type and result

	Blood Cultures (n = 3623)		Urine Cultures1 (n = 3653)		CSF Cultures (n = 594)
	Contaminant (n = 179)	No Growth (n = 3444)	Contaminant (n = 185)	No Growth (n = 3468)	Contaminant (n = 11)	No Growth (n = 583)
Age in days, Median (IQR)	46 (37 – 53)	45 (37 – 52)	44 (37 – 53)	45 (38 – 53)	39 (33 – 45)	39 (33 – 47)
ED revisit within 72h2, No. (%)
No	37 (26.1)3	2601 (92.2)3	129 (84.3)	2565 (89.5)	0	194 (87.0)
Yes	105 (73.9)3	219 (7.8)3	24 (15.7)	300 (10.5)	2 (100.0)	29 (13.0)
Hospital admission at ED revisit4, No. (%)
No	48 (45.7)	122 (55.7)	13 (54.2)	153 (51.0)	1 (50.0)	20 (69.0)
Yes	57 (54.3)	97 (44.3)	11 (45.8)	147 (49.0)	1 (50.0)	9 (31.0)

CSF: Cerebrospinal fluid; IQR: Interquartile range; ED: Emergency Department

¹35 infants with “likely contaminant” growth in urine culture are not included in this sample, as these cannot be definitively determined not to represent a urinary tract infection

²Of infants discharged from index ED visit (n = 3962 in blood culture sample, n = 3018 in urine culture sample, n = 225 in CSF culture sample)

³p < 0.001

⁴Of infants with ED revisit (n = 324 in blood culture sample, n = 157 in urine culture sample, n = 31 in CSF culture sample)

Urine Cultures

Overall, 3850 (95.2%) infants underwent urine testing, with 140 of these infants (3.6%) having a urinalysis obtained without a urine culture. Of the 3710 infants with a urine culture, 3571 (96.3%) were obtained by catheterization, 28 (1.0%) were bagged specimens, 3 were obtained by suprapubic aspiration (0.1%), and 108 (2.9%) were missing data regarding the method of obtaining urine. Twenty-two infants (0.6%) had a UTI (1 of whom also had bacteremia, Table 1). Another 35 infants (0.9%) had likely contaminant growth, and 185 (5.0%) had definite contaminant growth (range 0 – 33.3% by site, Figure 1b). The proportion of samples with definite contamination for specimens obtained by catheterization was similar to the overall contamination rate, though lower than for specimens obtained by bagged collection (5.0% vs. 17.8%, p = 0.01). Nine infants had contaminant growth in both blood and urine cultures; none had the same organism grow in both sites. There was no association between time of day or day of week and the proportion of infants with contaminated urine cultures. There was no difference in ED revisits between discharged infants with contaminated and negative urine cultures (Table 2).

CSF Cultures

In total, 669 infants (16.6%) had a lumbar puncture attempted, of whom 601 (89.8%) had CSF obtained in the ED. Of these, 596 had a CSF culture obtained. Two infants (0.3%) had bacterial meningitis, and 11 (1.8%) had a contaminated CSF culture. No infants with contaminant growth in the CSF had contaminant growth in blood or urine cultures. Two infants with a contaminated CSF culture were discharged from the index ED visit. Both had a return ED visit within 72 hours, and one was admitted to the hospital at that visit (Table 2).

Discussion

Nearly 10% of the low-risk febrile infants 29 to 60 days old in this multicenter sample had at least one contaminated culture obtained during their evaluation for fever. Furthermore, there was variability in blood and urine culture contamination by site, suggesting that site-specific practices may influence contamination rates. This is consistent with the limited existing literature on culture contamination in older infants and children evaluated in acute care settings, in which contamination rates vary across studies. (6,9–15) In the setting of evidence-based guidelines that drive the decision of whether to obtain cultures in young infants with fever, our findings have important implications for reducing unnecessary cultures and improving sterile technique when obtaining cultures.

The high proportion (5%) of contaminated blood cultures in our sample was associated with a clinically meaningful increase in return ED visits. In addition to being costly and increasing burden on ED resources, this can be a significant stressor to families. Importantly, the 5% contaminated blood culture rate is above the established acceptable threshold of 3%; it is also well above the 1–2% contamination rate achieved in places with pediatric and neonatal quality improvement initiatives aiming to reduce blood culture contamination.(16–19) As blood cultures are recommended for all young febrile infants,(2) these findings demonstrate a critical need for EDs to implement reliable methods for maintaining sterile technique when obtaining blood samples.

We also found a high proportion of contaminated urine cultures (5%), most documented as obtained by urethral catheterization. Although the few infants with urine culture obtained by bagged specimen had the highest rate of culture contamination, the rate of contamination was notable (5%) in specimens obtained by urethral catheterization as well. Contamination of the urine culture was not associated with an increased risk of ED revisit. As initial urine culture results often include speciation and colony counts that aid in interpretation, we hypothesize that physicians were more easily able to identify contaminants and, thus, less likely to call families back to the ED. Although the definition of a positive urine culture has been long debated,(20) it is important to note that the low-risk infants in our study all had normal urinalysis results by definition. Under the most recently published febrile infant guidelines by the American Academy of Pediatrics, urine cultures are no longer recommended in infants with normal urinalysis.(2) Though a small number of infants in this study were determined to have a UTI based on our study definition, these cases may represent asymptomatic bacteriuria and not true infection. Our findings support the ongoing process of de-implementation of routine urine cultures in febrile infants with normal urinalysis results.(2) Similarly, our findings support continued efforts to reduce unnecessary lumbar punctures in low-risk febrile infants. Although CSF contamination rates were relatively low in this sample, non-indicated lumbar punctures were performed on a meaningful proportion (~16%) of infants, reflecting the difficult task of changing behaviors and de-implementation.(21)

This secondary analysis of retrospectively collected data has limitations due to the potential for inaccurate pathogenic and contaminant organism growth classification. However, we used a multi-pronged approach for contamination based on microbiologic data obtained from the electronic health record and physician documentation. Additionally, the definition of “low-risk” that we applied to this sample was based on institutional guidelines and standardized definitions commonly used at that time. It is possible that some infants we identified as low-risk were not considered as such by their treating physician, or they may not meet low-risk criteria based on more recently published guidelines. Finally, because this was a secondary analysis of an existing dataset, we are not able to conduct a detailed analysis of the downstream interventions that occurred in response to contaminant organism growth.

The high rates of contaminated blood and urine cultures in this sample of low-risk febrile infants demonstrates a need for rigorous quality improvement efforts to minimize unnecessary cultures and enhance sterile techniques, safeguarding patient outcomes and optimizing resource utilization.

Abbreviations:

ED

emergency department

IBI

invasive bacterial infection

UTI

urinary tract infection

CSF

cerebrospinal fluid

cfu

colony-forming units