Diurnal Variation in Frequency and Severity of Croup in the Emergency Department

We conducted a retrospective cohort study of children with croup presenting to EDs that are part of the Pediatric Health Information System (PHIS). PHIS is an administrative database that contains inpatient, ED, ambulatory surgery, and observation encounter-level data from more than 50 not-for-profit, tertiary care pediatric hospitals in the United States.¹⁴  These hospitals are affiliated with the Children’s Hospital Association (Lenexa, KS). Data quality and reliability are ensured through a joint effort between the Children’s Hospital Association and participating hospitals. Data are deidentified at the time of data submission and are subjected to a number of reliability and validity checks before being included in the database. We included hospitals with complete clinical and billing data throughout the study period and excluded those without complete arrival hour data.

We included children between 3 months and 11 years of age who were treated in the ED between January 1, 2016, and December 31, 2020, and had a discharge diagnosis code for croup (International Classification of Diseases, Tenth Revision [ICD-10] codes J04 and J05). J05 has been shown to have very good sensitivity and positive predictive value for croup,¹⁵  and J04 was included to not miss potential cases.

We excluded patients with conditions or diagnoses that may alter clinical management or outcomes, including those with complex chronic conditions (as defined by Feudtner et al¹⁶ ), those with alternative airway diagnoses (congenital anomaly of larynx or trachea, burns, airway foreign bodies, and retropharyngeal, parapharyngeal, or peritonsillar abscess), as well as patients within the inclusion diagnosis categories who had diagnoses of epiglottitis (J05.1×) or supraglottitis (J04.3×). We also excluded patients with a previous croup encounter in the ED within the previous 30 days because repeat visits were an outcome measure and may have different clinical course.

We extracted the following variables from the PHIS database for each encounter: ED arrival hour, ED arrival date, age, sex, race, ethnicity, and insurance status, encounter diagnoses, racemic epinephrine administration, procedures (laryngoscopy and endotracheal intubation), patient disposition from the ED (discharge versus hospital admission), and all-cause ED return visits and hospitalizations within 3 days. ED arrival time was limited to the hour of arrival. For example, all patients arriving between 1:00 and 1:59 am would be classified as arriving at 1:00 am. We categorized age into <1 year, 1 year, 2 to 5 years, and 6 to 11 years in accordance with the National Institute of Child Health and Development age group recommendations.¹⁷  Although PHIS does not distinguish between ED and inpatient charges, we attempted to isolate racemic epinephrine use in the ED. We did this in hospitalized patients by including only charges on the date of presentation for patients presenting before 6:00 pm and including only charges on the date of presentation and the subsequent date for patients presenting at 6:00 pm or later. This definition has been used previously for PHIS studies in the ED.^18,19  We captured laryngoscopy and endotracheal intubation using ICD-10 procedure codes associated with each encounter.

We calculated total numbers of encounters during the study period across all hospitals at each of the 24 potential arrival hours and each of the 7 days of the week. We plotted the numbers of encounters at each hour to visualize trends. To visualize differences between croup and noncroup ED presentations, we also summarized numbers of encounters by hour and day of the week for all ED encounters at participating hospitals during the study period. We divided days into weekdays (Monday through Friday) and weekends (Saturday and Sunday) and calculated numbers of encounters per day on weekdays versus weekends. To further describe diurnal variation in croup, we also calculated and plotted the proportion of patients requiring hospitalization, racemic epinephrine treatment, and ICU admission at each arrival hour.

To determine differences in children presenting during the day and at night we classified into 2 groups. All encounters between 8:00 am and 7:59 pm were classified as “day” and all encounters arriving between 8:00 pm and 7:59 am were classified as “night.” We described the patient demographics and clinical outcomes for all encounters in the cohort using median and interquartile range for age and frequencies and proportions for all other variables. We tested differences between day and night encounters using χ² tests to compare proportions.

We determined the association between ED arrival time and severe clinical outcomes using multivariable generalized linear mixed-effects models. Our outcome variables were hospitalization, racemic epinephrine use, ICU admission, and all-cause 3-day revisits. For the analysis of 3 day revisits, we only included encounters discharged from the ED at the index encounter. The primary independent variables of interest in the models were arrival time (categorized into binary variable of day or night) and day of week (categorized into binary variable of “weekend” or “weekday”). We included age category, sex, race, ethnicity, and insurance type as fixed effects in the models to address potential confounding with season, year, and hospital as random effects to address potential clustering of outcomes. We calculated odds ratios for each outcome using the results of the full multivariable models. To ensure our categorization of arrival hour into a binary variable did not affect our results, we performed sensitivity analyses in which we performed the same models but with 3 arrival time categories. Statistical analyses and data visualizations were done using R (version 4.0.2; R Foundation for Statistical Computing, Vienna, Austria).

From 2016 through 2020, we identified 233 483 ED encounters with a discharge diagnosis of croup from 38 PHIS hospitals. We excluded 16 859 children with complex chronic conditions, 203 with alternative airway diagnoses, 131 with diagnoses of epiglottitis or supraglottitis, and 9095 with a previous croup encounter within the previous 30 days, leaving 209 949 encounters in our analytic sample. The median age of the cohort was 2.0 years (interquartile range: 1.2-3.6), and 65% were male. A total of 24.6% of children presenting to the ED with croup were treated with racemic epinephrine, 8.1% were admitted to the hospital from the ED, 0.8% required the ICU, and 5.2% had a return visit to the ED or inpatient setting within 3 days (Table 1).

There was considerable diurnal variation in ED presentation for croup. A total of 126 186 children with croup (60.1%) presented to the ED at night (between 8 pm and 8 am) and 83 763 children presented during the day (39.9%). The peak hour for croup presentation was 12:00 am (14 189 encounters) and the nadir was at 2:00 pm (5231 encounters). There was a slight increase in patient arrivals from 7:00 am until 10:00 am. There was an increase in the rate of croup presentations after 8:00 pm (Fig 1). This pattern contrasts with the overall arrival pattern noncroup encounters at the included hospitals, which demonstrated lower numbers during the overnight hours and the highest numbers of arrivals between 10:00 am and 8:00 pm (Fig 1).

In addition to time of day, frequency of croup encounters also varied by day of week; weekend days had 25% more encounters per day than weekdays during the study period. In contrast, noncroup encounters at the same hospitals averaged 2% fewer per day on weekends than weekdays.

We found significant differences between patients presenting during the night versus day in racemic epinephrine use (26.9% vs 21.2%, P < .001), hospital admission (7.4% vs 9.3%, P < .001), ICU admission (0.7% vs 1.0%, P < .001), and all-cause 3-day revisits (4.8% vs 5.9%, P < .001). There were very few patients requiring laryngoscopy or endotracheal intubation during either day or night (Table 1). Rates of severe outcomes by arrival hour are shown in Figure 2.

The results of our multivariable models are shown in Table 2. After adjustment for covariates in our multivariable models, children presenting during overnight hours had increased odds of treatment with racemic epinephrine (odds ratio [OR] 1.33; 95% confidence interval [95% CI], 1.30-1.36), but lower odds of hospitalization (OR 0.76; 95% CI, 0.73-0.78), ICU admission (OR 0.61; 95% CI, 0.56-0.68), and all-cause 3 day ED revisits (OR 0.86; 95% CI, 0.83-0.90), compared with children presenting during the day. Compared with children presenting during weekdays, children presenting on weekends had lower odds of racemic epinephrine use (OR 0.93; 95% CI, 0.0.91-0.95), hospitalization (OR 0.90; 95% CI, 0.87-0.93), and ICU admission (OR 0.87; 95% CI, 0.79-0.97), but no difference in all-cause 3 day ED revisits (OR 0.98; 95% CI, 0.94-1.02).

To ensure that our cutoff times for day and night did not substantially affect our results, we performed a sensitivity analysis in which we fit the same multivariable models but categorized ED arrival hour into 3 groups (8:00 am to 3:59 pm, 4:00 pm to 11:59 pm, and 12:00 am to 7:59 am). These models produced similar results, which are shown in Supplemental Table 3.

In this multicenter observational study of children in the ED, we found considerable diurnal variation in the presentation and outcomes of children with croup. We found that children with croup present most commonly at night and disproportionately on weekends. Although children presenting at night were more likely to be treated with racemic epinephrine, they had reduced odds of hospitalization, ICU admission, and 3 day revisits when compared with children presenting during the day.

To our knowledge, this is the largest and most comprehensive description of the diurnal nature of croup. Indeed, many review articles and practice guides^10,20,21  cite a nonpublished conference abstract (which does not make any mention of croup’s diurnal patterns) to support statements about croup’s diurnal nature.²²  Our finding that 60.1% of croup presentations occur at night is consistent with the limited previous literature on the subject.^12,13  In addition to the differences in overall proportions, the pattern we found of arrivals peaking in the middle of the night, decreasing continuously to the morning with a small peak mid-morning before an afternoon nadir, is very consistent with the pattern found in Canadian EDs.¹²  One potential physiologic mechanism previously postulated for this diurnal variation in croup presentations could be the circadian changes in serum cortisol.²³  Cortisol peaks around 8:00 am and reaches a nadir between 10:00 pm and 2:00 am, almost precisely opposite peak and nadirs of croup ED arrivals in our study.²⁴ 

Our finding of increased croup presentations on weekends is consistent with the finding of increased presentations on Saturdays and Sundays in Canadian EDs.¹²  Because there is likely not a physiologic reason for differences in croup presentations between days of the week, the finding that more presentations occur on weekends may suggest that children present to the ED with croup more frequently when they do not have an available primary care physician. This is also supported by the lower odds of severe outcomes among weekend ED arrivals. Diurnal variation in croup presentation was more pronounced than variation by day of week, however. There were 25% more encounters per day on weekends than weekdays, whereas there were nearly 3 times as many encounters at midnight than in the middle of the afternoon.

Interestingly, although more children presented to the ED with croup at night than during the day and were more likely to receive racemic epinephrine, they had lower odds of hospitalization, ICU admission, and 3 day return ED visits. The reason for this disconnect between increased odds of racemic epinephrine use at night but lower odds of severe clinical outcomes is unclear. One explanation could be that many croup presentations at night and on weekend days may be of lower acuity patients who might otherwise have been seen by primary care during office hours, although this would not explain the increased racemic epinephrine use. Use of racemic epinephrine may be impacted by a combination of factors, including different outdoor air temperatures, serum cortisol levels, and the ability for children to stay calm at night compared with during the day. Another explanation could be that patients with croup presenting to medical care during the day may represent a more severe phenotype of the disease than those with croup presenting suddenly and transiently at night. Additionally, providers may be less likely to treat children with croup-like symptoms during the day with usual treatments such as racemic epinephrine because of more diagnostic uncertainty in children presenting atypically. It is unlikely that decreased use of racemic epinephrine during daytime hours caused the observed increased odds of admission and other severe outcomes because the percentage of admitted patients receiving racemic epinephrine was essentially the same between night and day (76% vs 77%). We speculate that the contrast between greater odds of racemic epinephrine treatment and reduced odds of hospitalization, ICU requirement, or revisits could indicate a more transient or treatable phenotype of respiratory distress in patients who have sudden onset of croup symptoms at night.

Our study has several limitations. PHIS is an administrative database and therefore relies on diagnosis codes and billing data. It does not contain clinical results or physical examination findings, so we are unable to determine severity of illness at presentation. Although it is possible patients may have been misclassified as croup because of our use of ICD-10 codes for identification, the diagnosis codes used have been shown to have high sensitivity and positive predictive value for croup.¹⁵  The data set does not contain outpatient primary care visit or prescription data, so we are unable to determine whether patients were seen by a physician or treated with steroids or racemic epinephrine before or after their ED visits. It is possible children had ED revisits outside of the PHIS hospitals; however, previous studies have demonstrated this would be the minority of patients^25,26  and likely would not affect comparisons between day and night encounters. The PHIS hospitals are tertiary care pediatric centers and have higher admission rates for croup^5,27  than general EDs,⁴  so our results may not be generalizable to community EDs.

Most ED encounters for croup occur at night. In comparison with children presenting at night, children presenting to the ED with croup during the day have greater odds of hospitalization, ICU admission, and return visits after ED discharge, but lower odds of racemic epinephrine treatment. It is possible that croup presenting during the day may represent a more severe phenotype than nighttime croup.