United States’ Emergency Department Visits for Fever by Young Children 2007–2017

Introduction Our goal in this study was to estimate rates of emergency department (ED) visits for fever by children <2 years of age, and evaluate frequencies of testing and treatment during these visits. Methods We performed a cross-sectional study of ED encounters from 2007–2017 using the National Hospital Ambulatory Medical Care Survey, a cross-sectional, multi-stage probability sample survey of visits to nonfederal United States EDs. We included encounters with a visit reason of “fever” or recorded fever in the ED. We report demographics and management strategies in two groups: infants ≤90 days in age; and children 91 days to <2 years old. For patients 91 days to <2 years, we compared testing and treatment strategies between general and pediatric EDs using chi-squared tests. Results Of 1.5 billion encounters over 11 years, 2.1% (95% confidence interval [CI], 1.9–2.2%) were by children <2 years old with fever. Two million encounters (95% CI, 1.7–2.4 million) were by infants ≤90 days, and 28.4 million (95% CI, 25.5–31.4 million) were by children 91 days to <2 years. Among infants ≤90 days, 27.6% (95% CI, 21.1–34.1%) had blood and 21.3% (95% CI, 13.6–29.1%) had urine cultures; 26.8% (95% CI, 20.9–32.7%) were given antibiotics, and 21.1% (95% CI, 15.3–26.9%) were admitted or transferred. Among patients 91 days to <2 years in age, 6.8% (95% CI, 5.8–7.8%) had blood and 7.7% (95% CI 6.1–9.4%) had urine cultures; 40.5% (95% CI, 40.5–40.5%) were given antibiotics, and 4.4% (95% CI, 3.5–5.3%) were admitted or transferred. Patients 91 days to <2 years who were evaluated in general EDs had higher rates of radiography (27.1% vs 15.2%; P<0.01) and antibiotic utilization (42.3% vs 34.2%; P<0.01), but lower rates of urine culture testing (6.4% vs 11.6%, p = 0.03), compared with patients evaluated in pediatric EDs. Conclusion Approximately 180,000 patients ≤90 days old and 2.6 million patients 91 days to <2 years in age with fever present to US EDs annually. Given existing guidelines, blood and urine culture performance was low for infants ≤90 days old. For children 91 days to <2 years, rates of radiography and antibiotic use were higher in general EDs compared to pediatric EDs. These findings suggest opportunities to improve care among febrile young children in the ED.

What was the research question?Our goal was to report rates of presentation and testing among children <2 years with fever in US emergency departments.
What was the major finding of the study?A lower proportion of infants ≤90 days in age are evaluated for infections.Testing in older children may be high.

How does this improve population health?
Findings have implication for quality improvement efforts: more testing is needed among young infants, whereas some testing among older children may be of low value.
5][6][7][8][9] In contrast, the incidence of bacteremia in children 3-36 months of age is lower, allowing for selective testing and treatment. 10,11In such patients, routine blood culture is generally not recommended. 12owever, for febrile children older than three months of age, cross-sectional studies estimate that the overall incidence of UTI remains high (between 3-8%). 13,14Therefore, it remains important for providers to remain vigilant in evaluating for UTI.Across both age groups (<90 days and 91 days <2 years), routine use of chest radiographs is generally not recommended. 15espite extensive research performed on the risk stratification of febrile children, few epidemiological data are available describing the frequency of presentation of this condition to acute care settings and rates of testing performed.7][18] Our primary objective was to estimate the rate of ED visits for fever by infants ≤90 days, and children 91 days <2 years of age.Our secondary objective was to evaluate frequencies of blood and urine culture acquisition, radiographs, and antibiotic administration in this population and compare the management of pediatric patients 91 days <2 years between pediatric and general EDs.

METHODS
We performed a cross-sectional analysis of the National Hospital Ambulatory Medical Care Survey (NHAMCS), a nationally representative sample survey conducted annually by the Centers for Disease Control and Prevention National Center for Health Statistics (NCHS). 19NHAMCS is a cross-sectional probability sample survey of ED encounters to nonfederal and short-stay hospitals in the United States.Research with NHAMCS is approved by NCHS Ethics Review Board.
We included ED encounters from 2007-2017.We evaluated two cohorts, given the disease prevalence and evidence-based management strategies: a) infants ≤90 days of age; and b) children 91 days <2 years.We identified patients with fever as those encounters with either 1) a reason for visit code (RFV) classified as "fever" (RFV 1010.0) or "feeling hot" (RFV 1012.2); or 2) a documented temperature in the ED of 100.4°F (38.0°C) or greater.NHAMCS does not document the route of temperature acquisition.
We abstracted the following: demographics; testing (including blood culture, urine culture, radiographs); antibiotics (in ED and/or prescribed); disposition; and diagnoses.We classified EDs as pediatric if >75% of encounters were by patients under 18 years of age. 20Results were provided using survey-weighting procedures accounting for the NHAMCS sampling design, with 95% confidence intervals (CI). 21We assessed presentation rates using quasibinomial regression.For patients 91 days to <2 years old, we compared rates of testing and treatment between pediatric vs non-pediatric EDs using the Rao-Scott adjusted chi-squared test.We assessed SBI rates among infants <90 days, and rates of SBI, UTI, pneumonia, and otitis media among children 91 days to <2 years in age (Supplementary Table 1). 16,22,23Estimates with fewer than 30 records or with a relative standard error >30% were considered unstable. 24We conducted analyses using the survey package 25 in R, version 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria).
To evaluate specific rates of presentation and testing in 0-28, 29-60, and 61-90 day age groups, we conducted an exploratory analysis.For this, we broadened our inclusion to the years 2002-2017 in order to obtain sufficient numbers of raw patients to generate reliable estimates.
In our exploratory analysis for febrile infants ≤90 days old for the years 2002-2017, rates of blood cultures and urine cultures were similar between those 0-28 days and 29-60 days (Supplementary Table 2).A higher proportion of patients in older subgroups were discharged from the hospital.

DISCUSSION
In this nationally representative sample of ED encounters, approximately 180,000 infants ≤90 days and 2.6 million children 91 days to <2 years old presented annually for fever.One-third of febrile infants ≤90 days had blood and urine cultures, while 7% of older febrile children had blood cultures.Given higher rates of bacteremia in febrile infants <90 days old, routine acquisition of blood and urine cultures is recommended by guidelines. 85][6][7][8] While rates of bacteremia in infants 61-90 days old may be lower than rates in 0-60 days old infants, data from one recent prospective study suggest that the prevalence of bacteremia even in the third month of life is still high (1%). 26ne study limited to pediatric hospitals found the rate of culture acquisition among febrile infants ≤90 days was 69% for blood and 75% for urine cultures. 16ur investigation found a low frequency of culture acquisition (27.6% having blood cultures, and 21.3% having urine cultures).However, as the rate of SBI in our study was 9% in this age group, comparable to prior research, 3,7 our findings suggest a need for education and quality improvement.Quality-based measures, such as the recently reported Reducing Excessive Variability in the Infant Sepsis Evaluation, which includes clinical algorithms, order sets, education, and a mobile phone application for the management of febrile infants, can reduce variability with respect to hospital admission and lengths of stay. 27acteremia is relatively uncommon among infants >90 days of age.In one multicenter review of 57,000 blood cultures from children 3-36 months of age, rates of bacteremia were <0.5%. 11However, we observed that blood culture performance in this group was high (at approximately 1 in 14) and approached the rates of urine culture.Frequent use of blood cultures in this setting may lead to downstream effects, such as false positives and repeated testing. 28Our findings may represent adherence to older guidelines recommending empiric treatment for occult bacteremia in patients with fever.The 2003 American College of Emergency Physicians guidelines provided "Level B" evidence supporting empiric antimicrobial use for children having fever without a source. 29Acknowledging lower rates of bacteremia in the post-pneumococcal vaccine era, specific recommendations regarding empiric antimicrobial use were removed in a 2016 update to this guideline. 15Given the prevalence of viral infections in febrile children, 30 a large number of patients may receive antibiotics unnecessarily.Educational sessions and individualized audits may be beneficial in limiting unnecessary antibiotic use. 31

LIMITATIONS
Our findings carry limitations, including potential errors with respect to documentation, abstraction, and coding. 32ome variables were not present during the entire study period.In addition, we were unable to provide reliable estimates for some tests, or obtain testing trends over time.Indications for performing particular testing and antibiotic prescribing were not available in this dataset.In particular, we were unable to directly correlate antibiotic use for specific infectious diagnoses.

CONCLUSION
Approximately 180,000 children ≤90 days old and 2.6 million children between 91 days and <2 years present to US EDs annually with fever.Fewer than 1/3 of infants ≤90 days were evaluated with blood and urine cultures, which appears to be low.Blood culture testing and antibiotic use among children 91 days to <2 years appear to be high, in light of practice guidelines.These findings suggest important opportunities to improve the care of febrile children in the ED.
Western Journal of Emergency Medicine Ramgopal et al.United States' ED Visits for Fever by Young Children 2007-2017

Table 1 .
Continued.CI, confidence interval; PA, physician's assistant; NP, nurse practitioner.*Raw counts are the number of actual encounters available within the NHAMCS dataset; these are used with encounter-level survey weights to generate estimates and percents with confidence intervals. 21† Urine and lumbar puncture data were only available for years 2012-2016.‡ Calculated from a low number of raw counts or with a high relative standard error, which may lead to estimate instability per the National Center for Health Statistics guidelines.

Table 2 .
Testing and treatment of febrile children 90 days to <2 years of age, by type of emergency department.
CI, confidence interval; PA, physician's assistant; NP, nurse practitioner.*P-values assessed by Rao-Scott adjusted Pearson chi-squared statistic.† Urine and lumbar puncture data were only available for years 2012-2016.‡ Calculated from a low number of raw counts or with a high relative standard error, which may lead to estimate instability per the National Center for Health Statistics guidelines.Volume 21, no.6: November 2020 United States' ED Visits for Fever by Young Children 2007-2017 Ramgopal et al.