Trends in Management of Simple Febrile Seizures at US Children’s Hospitals

We performed a retrospective cross-sectional analysis of children diagnosed with SFS in the ED. Data were obtained from the Pediatric Health Information System (PHIS), an administrative database that contains inpatient, ED, ambulatory surgery, and observation encounter-level data from not-for-profit, tertiary care pediatric hospitals in the United States affiliated with the Children’s Hospital Association (Lenexa, KS). Data quality and reliability are assured through a joint effort between the Children’s Hospital Association and the participating hospitals. Portions of the data submission and data quality processes for the PHIS database are managed by IBM Watson Health (Ann Arbor, MI). Participating hospitals provide discharge and/or encounter data, including demographics, diagnoses and procedures, as well as resource use data (eg, pharmaceuticals, imaging, and laboratory studies). Data are deidentified and given unique identifiers to allow for tracking across multiple encounters.¹⁴  We included 49 hospitals contributing data for any period of time during the 15-year study period. This study was approved by the institutional review board at the study institution with a waiver of informed consent.

We included encounters for children presenting to an ED for their first visit with a diagnosis of SFS between January 1, 2005, and December 31, 2019. All subsequent visits were excluded because PHIS allows tracking of the same patients across multiple encounters. Using the criteria of SFSs per the AAP, only patients aged 6 to 60 months were included.¹⁰  The International Classification of Diseases, Ninth Revision (ICD-9), and International Classification of Diseases, 10th Revision (ICD-10) diagnosis codes of 780.31 and R56.00, respectively, corresponding to SFSs were used to identify eligible encounters. We included any visit with an ICD-9 and/or ICD-10 code for SFS as either a primary or subsequent diagnosis. We excluded visits with a codiagnosis of complex febrile seizure (ICD-9 780.32; ICD-10 R56.01), head trauma (ICD-9 800.0–804.9, 850.0–854.19, 950.1–950.3, 995.55, and 959.01; ICD-10 S00–S09.93 and T74.4), posttraumatic seizure (ICD-9 780.33; ICD-10 R56.1), bacterial meningitis (ICD-9 and ICD-10 codes as below), and hemiplegia (ICD-9 342 and 344.89; ICD-10 G81 and G83.84). Patients who died during the ED encounter (AAP guidelines apply to non–critically ill children), patients transferred to the PHIS hospital (evaluation before transfer cannot be determined), and children with preexisting complex chronic conditions were also excluded.¹⁵ 

Patient data, including demographics (age and sex), visit details (month, year, and season of presentation), and hospital region (northeast, south, midwest, and west) were collected. Diagnostic tests, including complete blood counts, serum chemistry, urinalysis, and brain computed tomography (CT) performed during the visit, as well as antibiotic administration, were measured. To ensure we accurately identified all children in whom an LP was performed, we defined the performance of an LP by either presence of the Current Procedural Terminology code for the procedure or laboratory coding for cerebrospinal fluid culture.¹⁶  We defined hospital admission as admission to the ICU, inpatient unit, or observation status. We measured return ED visits for each patient within 3 calendar days of the index ED encounter.

Our primary outcome was performance of LP because this was the principal change in the 2011 guideline. Our secondary outcomes were performance of hematologic and urine testing, neuroimaging, hospital admission, intravenous antibiotic use, mean cost per hospital encounter, and delayed diagnosis of bacterial meningitis over a 3-day revisit period. We defined a case of delayed diagnosis of bacterial meningitis as children who did not have an LP performed at their index encounter and returned within the 3-day revisit period with subsequent diagnosis of bacterial meningitis.

We used the following ICD-9 and/or ICD-10 codes to identify patients with a diagnosis of bacterial meningitis: bacterial meningitis and Haemophilus meningitis (ICD-9 320.0; ICD-10 G00.0), pneumococcal meningitis (ICD-9 320.1; ICD-10 G00.1), streptococcal meningitis (ICD-9 320.2; ICD-10 G00.2), staphylococcal meningitis (ICD-9 320.3; ICD-10 G00.3), meningitis due to other specific bacteria (ICD-9 320.8, 320.81, and 320.89; ICD-10 G00.8 and G00.9), meningitis in other bacterial disease (ICD-9 320.7; ICD-10 G01), bacterial meningitis not otherwise specified (ICD-9 320.9 and 320.82; ICD-10 G00.9 and G04.2), meningococcal meningitis (ICD-9 036.0; ICD-10 A39.0), tuberculous meningitis (ICD-9 013.0–013.06; ICD-10 A17.0), syphilis meningitis (ICD-9 091.91 and 094.2; ICD-10 A51.41 and A52.13), and Salmonella meningitis (ICD-9 003.21; ICD-10 A02.21).¹⁶  We defined a case of bacterial meningitis as having the following: (1) any 1 of the ICD-9 and/or ICD-10 diagnosis codes above, (2) LP performed, and (3) either a minimum of 7 days of intravenous antibiotics or placement of a peripherally inserted central catheter. This definition was used to address the limitations of coding and to improve the specificity with which we could identify true cases of bacterial meningitis.¹⁶ 

Cost data were calculated directly from charges reported by PHIS hospitals. Costs were obtained by using the total adjusted costs (ratio-cost-charge based). Hospitals contribute charges to PHIS using billing data that are then converted to costs by using hospital- and department-specific cost-to-charge ratios. Costs are also adjusted for regional wage-price indices to facilitate comparison of costs across geographic regions of the country. We adjusted costs for inflation over the study period using the Consumer Price Index for Medical Care, as compiled by the US Bureau of Labor and Statistics, such that all costs were adjusted to 2019 US dollars.¹⁷  Costs analyzed included all costs for the encounter for febrile seizure, including ED and hospital costs (if admitted to the hospital or observation status). Mean charges were chosen to account for the significantly higher costs associated with hospital admission.

We described patient characteristics using medians and interquartile ranges (IQRs) for continuous variables and proportions and 95% confidence intervals (CIs) for categorical variables. Because the 2011 guidelines made specific recommendations for LP in the youngest infants, we categorized patients aged 6 to <12 months and ≥12 to 60 months for subanalyses of rates of LP. We compared patient characteristics between the preperiod and postperiod using Wilcoxon rank tests for continuous variables and χ² tests for categorical variables. To assess for trends in rates of our outcomes, we performed interrupted time series analyses, with an indicator of outcome on the y-axis and a variable for continuous time by month before and after guideline publication on the x-axis. The output from this model provided the slope for the preintervention and postintervention regression lines. The data for the beginning and end study years were aggregated for rates of diagnostic testing, medication administration, hospital admission, and cost and compared by using t-tests for costs and χ² test for categorical outcomes. Statistical analyses were performed by using SAS software (SAS Institute Inc, Cary, NC).

We identified 192 348 ED visits by children with a diagnosis of SFS, of whom 142 121 met all inclusion and no exclusion criteria (Fig 1).

Demographics and seasonality of ED visits are summarized in Table 1. Median age at presentation was 20.8 months (IQR 15.2–29.6 months), and 60 255 (42.4%) patients were female. A total of 51.4% of patients were White and 23.2% Black, and 25.0% identified as Hispanic. A greater proportion of visits for SFS occurred during the winter and spring months (32.2% and 26.6%, respectively). A total of 49 668 (35.0%) children presented before and 92 453 (65.0%) presented after publication of the AAP guideline. No substantive differences in patient demographics were noted between the preguideline and postguideline periods.

The percentage of patients undergoing LP declined over the study period (Fig 2). LPs were performed in 11.6% (95% CI, 10.8% to 12.4%) of children in 2005 and 0.6% (95% CI, 0.5% to 0.8%; P < .001) in 2019. The greatest reduction in rate of LPs was found in children aged 6 to <12 months: 28.3% (95% CI, 25.0% to 31.8%) to 1.0% (95% CI, 0.5% to 1.8%; P < .001). Among children aged 12 months to 60 months, 9.4% (95% CI, 8.7% to 10.2%) had an LP performed in 2005 compared with 0.6% (95% CI, 0.5% to 0.8%; P < .001) in 2019.

The use of diagnostic testing also decreased over the study period (Fig 3). Rates of head CT decreased from 10.6% (95% CI, 9.8% to 11.4%) in 2005 to 1.6% (95% CI, 1.4% to 1.9%; P < .001) in 2019. In 2005, complete blood counts were performed in 38.8% (95% CI, 37.5% to 40.0%) of visits but declined to 10.9% (95% CI, 10.3% to 11.5%; P < .001) in 2019. Serum chemistries were obtained more frequently at the beginning of the study period: 27.5% (95% CI, 26.1% to 28.3%) in 2005 compared with 11.0% (95% CI, 10.5% to 11.6%; P < .001) in 2019. Urinalysis rates also decreased, but to a lesser extent, from 31.4% (95% CI, 30.3% to 32.6%) in 2005 to 22.3% (95% CI, 21.5% to 23.1%; P < .001) in 2019.

Among first-time presentations for SFSs, 19.2% (95% CI, 18.3% to 20.2%) resulted in admission in 2005. This rate decreased to 5.2% (95% CI, 4.8% to 5.6%) in 2019 (P < .001), although largely plateaued after the guideline publication (Fig 4). Use of intravenous antibiotics declined from 17.7% (95% CI, 16.7% to 18.6%) to 3.3% (95% CI, 2.9% to 3.6%; P < .001) from 2005 to 2019. Three-day revisits increased marginally from 0.5% (95% CI, 0.3% to 0.7%) to 1.4% (95% CI, 1.2% to 1.7%; P < .001). After adjusting for health care inflation, there was an overall reduction in mean inflation-adjusted costs before and after guideline publication. Total mean costs were $1523 (SD $3704) in 2005 and $605 (SD $1638; P < .001) in 2019. Similar to hospital admissions, the decline in costs plateaued after the guideline, as demonstrated in Fig 4.

We observed no significant difference in the rate of delayed diagnosis of bacterial meningitis between the periods before the guideline and after the guideline (preperiod 2 of 49 668 [0.0040%; 95% CI, 0.00049% to 0.015%], postperiod 3 of 92 453 [0.0032%; 95% CI, 0.00066% to 0.0094%]; P = .99).

In this national, multicenter, retrospective cohort of >142 000 ED visits for children presenting with a SFS, we observed a significant decline in rates of LP between 2005 and 2019. Among all age groups, the rate of LP for SFS had a relative decrease of 95% from the beginning to the end of the study period, with the greatest impact noted among the youngest infants, aged 6 to 12 months. We found similar declines in rates of diagnostic laboratory and radiologic testing, intravenous antibiotic administration, hospitalization, and costs. Although we initially hypothesized that the 2011 AAP guideline would be the driving factor in changes in the management of SFSs, we found that rates of diagnostic testing, hospitalization, and costs were downtrending before its publication, and this trajectory continued after guideline publication. Importantly, the decrease in testing was not associated with a concurrent increase in delayed diagnoses of bacterial meningitis. Together, these data reveal that, as recommended by the AAP guideline, children with SFSs have been safely managed without delayed diagnoses of bacterial meningitis with decreased testing and treatment over the fifteen-year study period.

Our findings are consistent with smaller studies published before 2011 in which researchers found declining rates of LP in children presenting to the ED with their first SFS.^18–20  In another 2009 study, researchers found that the previous 1996 AAP practice parameter recommending routine LP performance in infants aged 6 to 12 months presenting with a SFS was not being adhered to strictly.¹⁹  In these studies, researchers found no cases of bacterial meningitis in patients who underwent LP.^18–20  Our study augments the existing literature with its large sample size and, therefore, is better powered to detect the exceedingly rare cases of bacterial meningitis among children presenting with SFSs. Moreover, our study is national and spans a longer period of time, allowing for the assessment of nationwide patterns in testing.

Seizures have been described as a clinical predictor of bacterial meningitis and are incorporated into bacterial meningitis risk-stratification scores.^21,22  This has been the subject of some skepticism in the literature, including in a 20-year study of children diagnosed with meningitis in which all patients had additional symptoms beyond fever and seizures.²³  However, between the publication of the AAP practice parameter in 1996 and the AAP clinical practice guideline in 2011, the landscape of infectious diseases in the pediatric population changed dramatically. Over this period of time, bacterial meningitis has become increasingly rare as a result of widespread use of pneumococcal and Haemophilus influenzae vaccines.²⁴  Toward the beginning of our study period, in 2007, rates of bacterial meningitis were 7 per 100 000 among children between 2 and 23 months of age and 0.56 per 100 000 among children between 2 and 10 years of age.²⁴  The declining rates of bacterial meningitis in the years before the 2011 AAP guideline likely spurred the downtrend in rates of LP even before the guideline publication. Furthermore, academic children’s hospitals have been frequently shown to adopt evidence from the scientific literature more quickly compared with general hospitals.^25,26  The hospitals that contribute to the PHIS database are part of the Children’s Hospital Association and are usually affiliated with academic institutions, which may explain the ongoing decline in LP rates in response to research studies before the guideline publication.

Routine ED management of SFS includes identification of a fever source; however, no source of infection is identified in most patients despite the frequently broad testing performed.²⁷  Hematologic testing, EEG, and neuroimaging have not been recommended since the 1996 AAP practice parameter, given the numerous studies showing no added benefit.^28–30  Therefore, it was surprising that our data found that these studies were still being performed in 10% to 40% of visits in 2005. Nevertheless, the steady decline over the subsequent years and much lower rates at the end of the study demonstrate improved adherence to the current guidelines. The decrease in urinalysis testing requires further study to understand, because identifying a source of fever remains a priority in the evaluation of SFS.

The results of our study must be interpreted in the context of its limitations. First, given our reliance on an administrative database with limited clinical information, we are unable to understand the presenting signs, symptoms, and physical examination findings of children included in our study. A specific challenge was not being able to exclude underimmunized or unimmunized children or those who had been previously treated with antibiotics as this may heighten suspicion for bacterial meningitis.¹⁰  This prohibits an assessment of the appropriateness of diagnostic testing among children included in our study. Furthermore, although we included only the first ED visit to a PHIS hospital for febrile seizure, this may not represent the patient’s first-ever febrile seizure if they previously presented to a non-PHIS facility or had an episode that predated our study. However, the frequency of this should not have changed over time, and therefore, the impact should be minimal. Another limitation of the study and our cohort selection is its identification by using the diagnosis code for SFS, which relies on the accurate entry of this information by the clinician. However, it is unlikely that variations in the use of these codes over time would account for the profound changes in management we observed. As diagnosis codes are not assigned until the end of the visit, we cannot comment on the management of children with fever and seizure who did not ultimately receive a diagnosis of SFS. Although it is possible that patients presenting with fever and seizure may not have received a diagnosis of SFS, and therefore would not be included in our analysis, our results underscore that patients diagnosed with SFS were able to receive this diagnosis with decreased testing and treatment over time. Our data were obtained from academic children’s hospitals and may not reflect practice at nonpediatric hospitals. The vast majority of children are cared for in nonpediatric hospitals, with data suggesting differential management for other conditions between pediatric and nonpediatric centers. Thus, further data are needed to understand the management of children with febrile seizures in the general hospital setting.^31–34  Finally, centers gradually joined PHIS over the study years and were not all contributing data throughout the study period.

As recommended by the AAP guideline, there has been a significant decline in rates of LP performance among children presenting to a US EDs with a SFS. Similar reductions have been seen in rates of diagnostic testing, hospital admission, intravenous antibiotic administration, and costs without a concurrent increase in the rates of delayed diagnosis of bacterial meningitis. Although declines began before the AAP guideline, they continued to decrease asymptotically after its publication. These data suggest that, as recommended by the AAP guideline, children with SFSs can be safely managed with less invasive testing.

AAP

American Academy of Pediatrics

CI

confidence interval

CT

computed tomography

EEG

electroencephalography

ED

emergency department

ICD-9

International Classification of Diseases, 9th Revision

ICD-10

International Classification of Diseases, 10th Revision

IQR

interquartile range

LP

lumbar puncture

PHIS

Pediatric Health Information System

SFS

simple febrile seizure