Predictors of Inpatient Racemic Epinephrine Use in Patients Admitted With Croup Free

We conducted a retrospective cohort analysis at a freestanding tertiary care children’s hospital, in which there is not currently a clinical pathway for croup. We identified patients ages 2 months to <7 years who received RE in the ED and were admitted with a primary diagnosis of croup (International Classification of Diseases, 10th Revision: J05.0 and J04.2) and discharged between January 1, 2016, and December 31, 2019. All patients identified by International Classification of Diseases, 10th Revision diagnosis codes were reviewed for eligibility. This study was approved by the university’s institutional review board.

Potential study subjects were excluded if they had underlying diagnoses that could alter disease course (congenital airway anomalies, history of tracheostomy or other tracheal surgery, or other complex medical diagnoses such as Down syndrome, congenital heart disease, or cystic fibrosis), if they had a noncroup primary diagnosis, if they were initially admitted to the PICU, if they were transferred from another hospital’s inpatient service, if they did not receive RE in the ED, or if the medical record was missing data critical to the study such as ED RE administration times or vital signs (Fig 1).

For each patient, demographic data were collected including age, sex, race, ethnicity, and insurance type. First recorded ED triage vital signs including heart rate, respiratory rate, temperature, oxygen saturation, and weight were also collected. Tachycardia for age and tachypnea for age were determined by using standardized normative values for age¹⁸  to be consistent with the methodology used in previous studies.¹⁴  Fever was defined as a temperature of 38°C or above; abnormal O₂ saturation was defined as <95%.

ED course data included number of RE doses, time between RE dose administrations (doses 1–2, doses 2–3), steroids given (dexamethasone, any other steroid, no steroid), and imaging studies (chest radiograph, soft tissue neck radiograph). The documented RE administration times in the medical record were used to calculate the time between RE doses. For some patients, the first RE dose was given at an outside facility or en route to the ED via emergency medical services.

The primary outcome was whether RE was administered after admission. In our hospital, once a decision to admit is made, the charge nurse is notified, and a bed assignment is made. The admitting team is then called and assumes care of the patient even if they are still physically in the ED. Because clinical activity of RE typically lasts 120 minutes with a peak activity seen at 60 minutes,⁶  we hypothesized that patients who have a shorter time before needing subsequent ED RE administrations are more likely to require inpatient RE administration. Secondary outcomes including LOS, observation time, and 7-day croup revisits and readmissions were analyzed to assess the impact of inpatient RE use on clinical outcomes. LOS was defined as the time from admission order to departure from the hospital. Observation time was defined as the time between the last administered RE in any setting and departure from the hospital. Revisit was defined as any return to the ED for croup within 7 days of discharge; readmission was defined as admission in observation or inpatient status with a diagnosis of croup within 7 days of discharge.

Descriptive statistics including median and interquartile range (IQR) were calculated for continuous variables, and comparisons across the 2 groups (no inpatient RE and inpatient RE) were made using the Mann-Whitney U test. For categorical variables, comparisons between groups were made using the χ² test or the Fisher’s exact test for variables with <10 observations. P values <.05 were considered statistically significant. In addition to the univariate comparisons made for primary and secondary outcomes, a multivariate logistic regression was used in a model similar to Asmundsson et al¹⁴  to determine the adjusted risk of inpatient RE utilization on the basis of demographics and ED parameters by using odds ratios (OR) with 95% confidence intervals. Variables include age, sex, weight, tachycardia for age, tachypnea for age, temperature, O₂ saturation, number of RE doses in the ED, steroids given in the ED, chest and/or soft tissue neck radiograph, and time between RE first and second RE doses. Time between first and second ED RE administrations was included in this model with an effect modification to allow independent analysis of patients who did not receive a second RE treatment. Analysis was completed by using IBM SPSS Statistics for Windows (Version 26.0. Armonk, NY: IBM Corp).

Overall, 386 patient records were identified from January 1, 2016, to December 31, 2019. A total of 148 patients were excluded (Fig 1) with the most common exclusion criterion being insufficient records (n = 29). There were 238 patient encounters selected for our cohort, and 59 (24.8%) of them received inpatient RE. Of these patients, 31 (52.5%) received only 1 dose, 8 (13.6%) received 2 doses, 8 (13.6%) received 3 doses, and 12 (20.3%) received 4 or more doses while hospitalized. Among patients who did receive inpatient RE, age at admission was significantly younger at 13.1 months (P = .045). No other demographic or historical factors were associated with inpatient RE use (Table 1).

In the ED, the number of patients who presented with tachypnea for age was significantly higher for those who received inpatient RE at 49.2% compared to 33.5% (P = .03) in patients who did not receive inpatient RE (Table 2). There were no significant differences between the groups in tachycardia for age, temperature ≥ 38°C, O₂ saturation < 95%, and median weight. Most patients received 2 doses of RE in the ED (Table 2) regardless of group (65.4% and 64.4%; P = .99). The median time between RE doses 1 and 2 did not differ (P = .71) between the no inpatient RE group (1.9 hours; IQR 1.3–2.5) and the inpatient RE group (1.85 hours; IQR 1.2–2.5). There was also no significant difference between groups for time between RE doses 2 and 3. The median time to subsequent RE treatment after admission was 7.8 hours (IQR 2.7–17.4).

A multivariate logistic regression model (Table 3) was developed to calculate the OR for specific characteristics for predicting inpatient RE use. In our adjusted analysis, tachypnea for age in ED triage (OR 2.61; 95% confidence interval = 1.24–5.52) increased the risk of requiring additional RE in the inpatient setting. The time between RE doses 1 and 2 was not associated with an increased likelihood of inpatient RE, nor were the number of administered RE doses in the ED.

LOS was significantly longer (P < .001) in patients who received inpatient RE at 38 hours (IQR 27.1–55.6) compared to 16.7 hours (IQR 12.5–23.2) in patients who did not (Table 4). Overall, 7-day revisit rates and readmission rates were 3.8% and 4.2%, respectively. There was no difference in observation time or 7-day croup revisits and readmissions between groups.

Our study has described the demographics and management of a cohort of patients admitted with croup with specific focus on RE including time between ED RE treatments. Out of 239 admissions for croup, only one-quarter (24.8%) received subsequent RE treatments after admission, whereas 52.4% received just a single dose. These findings are similar to Asmundsson et al,¹⁴  and the percentage of patients receiving inpatient RE are decreased from both Rudinsky et al¹³  and Narayanan and Funkhouser.¹⁷  Neither the number of ED RE doses nor the time between these doses were predictive of subsequent inpatient RE. This indicates that the number of ED RE doses may not be ideal as an admit criteria. However, age and tachypnea for age may be helpful factors in triaging a patient’s overall need for admission. Furthermore, when patients received subsequent inpatient RE, they had significantly prolonged LOS compared to those who did not, despite no difference in revisit and readmission rates between groups.

Multidose RE use is often a primary factor in determining need for admission as used in treatment algorithms¹⁹  and clinical pathways,^11,12  but our findings suggest that scrutinizing current admit criteria may be warranted. Although most of our patients were admitted after 2 or more RE treatments, 23.5% were admitted after just a single treatment. This is <33.5% reported by Asmundsson et al¹⁴  and >10.7% presented by Bagwell et al in a study using an administrative database.¹⁶  This variation may be because of a lack of local clinical pathways or related to individual provider comfort level while indicating that additional factors other than just multidose RE use may contribute to admission decisions. To further evaluate the utility of RE use as a decision tool for admission, we hypothesized that patients who received a subsequent dose of RE more quickly in the ED were more likely to require inpatient RE. However, the median time between RE doses 1 and 2 did not differ between groups, nor was it a predictor of inpatient RE use through our multivariate logistic regression model. RE use has been associated with improved croup scores at 30 minutes posttreatment but no significant effect at 2 hours posttreatment.⁵  Corticosteroids begin to show symptom improvement at 1 to 2 hours with an effect lasting up to 24 hours.⁴  Thus, corticosteroid effects become evident around the time that RE activity is waning. Our findings suggest that patients receive a second dose at around 2 hours after initial treatment, likely for rebound or refractory stridor. This time period coincides with both waning RE effect and initial corticosteroid effect. Other factors contributing to timing between ED RE doses may include general ED workflow that is affected by time of day and patient volume RE factors alone including number of doses and timing of doses may not be satisfactory for determining a patient’s need for admission. Other parameters such as vitals or patient demographics may be useful for determining which patients may need inpatient RE.

Through univariate statistical analysis, we identified age and tachypnea for age as factors indicating need for inpatient RE. In a similar study, Asmundsson et al found that age-defined tachypnea, age-defined tachycardia, and receiving an ED radiograph were indicative of inpatient RE use.¹⁴  Through our multivariate logistic regression model, we found that children presenting with tachypnea for age had increased odds of receiving inpatient RE. Through a similar logistic regression model, Asmundsson et al identified tachypnea and receiving an ED radiograph as characteristics associated with inpatient RE use.¹⁴  Age has not been previously identified as indicative of inpatient RE in previous studies,^13,14  but Tyler et al identified it as a significant factor for rebound symptoms in a cohort of sicker patients admitted to ICU settings.²⁰  Tachypnea at initial vital sign check is a nonspecific finding, but it may reasonably indicate severe disease and a need for further RE treatment. Our analysis not identifying ED radiographs as significant might be indicative of practice variation or a lack of clinical pathway at our institution. Based on our findings, clinicians could consider paying particular attention to age and tachypnea when determining a patient’s need for admission within the broader context of evaluating respiratory status.

Analysis of LOS, observation time, and time to subsequent inpatient RE may allow us to consider alternative approaches in caring for croup patients. For patients receiving inpatient RE, median LOS was significantly longer at 38 hours in our study and 27 hours in a similar study.¹⁴  We found that observation times were not different between groups, indicating that patients were not observed longer solely because of receiving inpatient RE. Instead, other factors influencing observation time and LOS may include clinician or family comfort level with discharge, lack of a standardized croup pathway, availability of reliable follow-up, or the time of day that the patient reaches stability. We also had relatively low readmission rates of 4.2% with 0.4% being readmitted to the PICU. These findings are similar to previous studies showing a 3% 30-day readmit rate,²¹  a 4% 7-day readmit rate,¹⁷  and a 4% 48-hour readmit rate.¹³  Outpatient and ED studies have demonstrated that 2 to 4 hours of observation after RE are sufficient.^6–10  Treatment algorithms presented in the literature¹⁹  and hospital pathways^11,12  have used observation period as an indicator for discharge, while using a benchmark of 2 RE treatments as an indicator of need for admission. This approach seems to be costly overall with patients admitted after multidose epinephrine having mean adjusted billed charges of $23 460.10 and mean adjusted estimated cost of $8654.20 compared to costs for patients discharged from the ED of $4537.10 and $1568.00, respectively.¹⁶  In our cohort, the median time to first inpatient RE after the ED was 7.8 hours, which is outside of the recommended 2- to 4-hour observation period.^6–10  Instead, observing a period of 6 to 8 hours in the ED before admission may be more reasonable. However, this could lead to a higher readmission rate overall. Interestingly, Bagwell et al identified that of patients who were discharged from the ED after multidose RE, only 0.8% returned to the ER for reevaluation with 0.3% being admitted to the hospital on ED revisit. This provides evidence that observing longer in the ED without reflexively admitting for 2 RE treatments could not only still be safe but also more cost effective overall by preventing unnecessary admissions.

Some limitations in this study are inherent to its design. This was a retrospective and single-center study and may not generalize to other hospitals, especially if clinical decision-making is different. Additionally, without a clinical pathway for croup management at our institution, decisions such as RE administration or observation period after RE were more subject to clinical judgment and likely varied among providers. Timing between ED RE doses may have been falsely prolonged, because some patients who were initially seen at an outlying ED or urgent care facility were included if administration times were available, but transportation time was not accounted for. Overall LOS and even observation time were likely affected by time of day of admission, and this was not accounted for. We also were unable to control for other reasons for delayed discharge such as lack of transportation. Additionally, some details may have been missed during records review. We did not have a way to account for patients who may have already been on steroids before ED presentation, which may have altered the patient’s clinical presentation. As we sought to characterize patients admitted for croup, we did not analyze the subset of patients who are discharged from the ED with croup, which may have allowed additional analysis. Finally, it is important to note that our power to detect differences in outcomes was limited. Thus, developing large multicenter studies including children that are not admitted would be a critical next step for providing appropriate safety data to inform higher-value clinical pathways.

Only about one-quarter of patients admitted for croup received inpatient RE, receiving it at a median of 7.8 hours after their last ED dose. Patients were more likely to receive inpatient RE based on younger age and tachypnea for age, but timing between ED RE doses did not differ between groups. LOS was significantly longer in hospitalized patients receiving RE, but observation times after the last RE treatment did not differ between groups. Revisit and readmission rates were also low. Our study suggests that the frequently used admission criteria of multidose ED RE treatments could be reevaluated, instead allowing for longer observation periods before decisions to admit. Age and tachypnea for age may be helpful factors in triaging such patients. Further prospective studies are needed focusing on providing appropriate and value-added care for patients with croup.

We thank Drs Sandra Arnold and Kenice Ferguson-Paul for their support of this project through their participation in our scholarly oversight committee and Dr Stephen Pishko in his role as fellowship program director. This project was supported by the Children’s Foundation Research Institute at Le Bonheur Children’s Hospital and the Children’s Foundation of Memphis.