Dexamethasone Versus Prednisone in Children Hospitalized for Acute Asthma Exacerbations

This was a multisite retrospective cohort study done within a tertiary care children’s hospital system including 3 academic teaching hospitals with >600 licensed beds. All 3 hospitals use the same asthma admissions criteria, management guideline, and order set. This study was approved by the hospital system institutional review board.

Children 3 to 21 years of age hospitalized within the hospital system between January 1, 2013, and December 31, 2017, with a primary discharge diagnosis of asthma exacerbation or status asthmaticus on the basis of International Classification of Diseases, Ninth and 10th Revision, Clinical Modification codes were included in the initial electronic search. Manual chart review was then done to validate the cohort. Only children receiving parenteral or oral monotherapy with DEX or oral monotherapy with PRED during the acute illness (before, during, and after hospitalization) were included. Five attending physicians, 2 resident physicians, and 1 medical student completed the manual chart review; all received thorough in-person training and written instructions detailing data acquisition. The principal investigator also periodically checked charts for accuracy.

Children who received an unspecified type of oral steroid or a combination of DEX and PRED at any point during their acute illness (ie, before or during hospitalization, or at discharge) were excluded from the study. Children with missing information on whether steroids were given before admission were also excluded. Children receiving methylprednisone at any point during their acute illness were excluded even if they were later switched to PRED to avoid any confounding variability in dosing and frequency of the 2 medications.²⁶  Children receiving steroids in the 2 weeks before presentation or those receiving a prolonged steroid course (ie, DEX for >3 days or PRED for >7 days) were also excluded. Other exclusion criteria included initial PICU admission; concurrent diagnosis of bronchiolitis, pneumonia, or croup; use of bilevel positive airway pressure; and administration of supplemental therapies such as antibiotics, oseltamivir, heliox, terbutaline, racemic epinephrine, hypertonic saline, chest physiotherapy, and budesonide as rescue medication in the ED. Children with pulmonary or cardiac comorbidities (ie, chronic lung disease, pulmonary hypertension, congenital heart disease, laryngomalacia, obstructive sleep apnea, or cystic fibrosis), sickle cell disease, Down syndrome, or immunosuppression were also excluded. Hospitalizations with paper chart documentation were excluded because of multiple missing variables. Although these exclusion criteria were developed to avoid confounding variability between the DEX and PRED cohorts, it inevitably narrowed our study population to those with mild-to moderate acute asthma severity.

For children with multiple asthma-related hospital visits within a 10-day period, only the first encounter was captured with subsequent ones counted as ED revisits or hospital readmissions.

Demographic variables included age and weight on admission, sex, race, ethnicity, insurance type, and hospital campus. Historical elements included presence of allergies (food, drug, and/or seasonal), baseline asthma severity per previous chart documentation and/or reported baseline symptom severity, and medications given before hospital arrival (DEX, PRED, albuterol, magnesium, and/or ipratropium).

Steroid-specific variables included type of oral steroid administered (DEX or PRED) and number of doses administered before, during, and after hospitalization. Because admission documentation is reliant on patient or family recall, specific dosing data for oral steroids given before hospitalization was not readily available. For oral steroids given during hospitalization and prescribed on discharge, dosing in milligrams per kilograms per day was recorded. Number of children completing full oral steroid course (ie, DEX for 2 days or PRED for 5 days) before discharge and those requiring oral steroid prescriptions at discharge was also recorded.

Asthma acuity was determined by the initial ED clinical respiratory score (CRS), a validated 12-point respiratory assessment tool with good interobserver agreement.²⁸  We also recorded number of albuterol treatments (intermittent and continuous) given during hospitalization, whether ipratropium and/or magnesium were administered during hospitalization, and maximum amount of supplemental oxygen in liters per minute per kilogram required during hospitalization.

For outcomes analyses, we collected data on hospital LOS, frequency of PICU transfers, and frequency of ED revisits and hospital readmissions in our hospital system within 10 days after discharge.

Descriptive statistics were calculated as counts and percentages, means and SDs, or medians and quartiles for all demographic and clinical variables. The χ², Fisher’s exact, or Students t tests were used, as appropriate, to test for a relationship between each variable of interest and type of oral steroid used (ie, DEX versus PRED). The P value < 0.05 was considered statistically significant. All statistical analyses were done by using SAS v9.4. (SAS Institute, Inc, Cary, SC).

LOS in hours was log transformed for modeling. Because of underlying differences in the population, children who started oral steroids before hospital arrival were modeled separately from those who started oral steroids after hospital arrival. Both unadjusted and adjusted least square means and ratios for hospital LOS were calculated by using generalized linear models estimated by maximum likelihood. Adjusted model for children who started oral steroids after hospital arrival controlled for age, sex, race, ethnicity, insurance, allergies, hospital campus, baseline asthma severity, albuterol administered before hospital arrival, initial ED CRS score, magnesium and ipratropium administered after hospital arrival, maximum supplemental oxygen required in the hospital, and year. Adjusted model for children who started oral steroids before hospital arrival included identical covariates except that albuterol administered before hospital arrival was omitted from the model because nearly all of these children received albuterol in addition to steroid before arrival. Weight was collinear with age and therefore excluded from the model.

PICU transfers during the initial hospitalization and ED revisits and hospital readmissions within 10 days after discharge were all uncommon outcomes. Consequently, unadjusted event rates are reported for all 3 secondary outcomes.

Six variables were missing values for 1% or more of the records. In the primary analysis, missing values were dealt with via listwise deletion. For sensitivity analysis, we ran reduced models that did not control for variables with substantial missing data. We also ran models that explicitly coded missing values for these variables as an “unknown” category. Because CRS is scored on a 0 to 12 scale with data points missing nonrandomly, this variable was collapsed into 3 severity categories plus an unknown category for purposes of the sensitivity analysis.

We completed campus specific sensitivity analysis on the primary outcome for 2 of the 3 campuses included in the overall models. One campus favored DEX so heavily that a campus specific model was not feasible. Similarly to the primary analysis, these models were stratified by initial steroid timing and run for multiple missing data strategies.

A total of 3542 children were initially identified by using our inclusion criteria and the electronic database search system. After a manual chart review, 1410 children (39.8%) were included in the study, with 981 (69.6%) receiving only DEX and 429 (30.4%) receiving only PRED (Fig 1). Most common reasons for exclusion were administration of methylprednisolone (n = 671), combined administration of DEX and PRED (n = 446), and paper chart documentations (n = 359).

Baseline characteristics in the DEX and PRED cohorts were similar for age and weight on admission, sex, and history of allergies (Table 1). Baseline asthma severity was worse in the PRED cohort for children starting oral steroids before hospital arrival but similar in both cohorts for children starting oral steroids after hospital arrival. The DEX cohort comprised a larger proportion of children of African American descent and with government insurance compared with the PRED cohort for children starting steroids after hospital arrival. These differences were not significant among children who started oral steroids before hospital arrival. There was significant intercampus variability in the type of oral steroid used, with physicians on campus B almost exclusively prescribing DEX and physicians on campus C prescribing PRED more often than those at other campuses.

Although CRS scores on initial presentation to the ED were similar for both the DEX and PRED cohorts irrespective of when oral steroids were initiated, several other clinical differences existed between the 2 cohorts (Table 2). For children starting oral steroids after hospital arrival, a larger proportion of the PRED cohort received albuterol (P < .001) and magnesium (P < .001) before presentation. For children starting oral steroids before hospital arrival, a larger proportion of the DEX cohort received ipratropium before presentation (P = .01). For children starting oral steroids after hospital arrival, the DEX cohort more often received magnesium and ipratropium treatments during hospitalization (P < .001) whereas the PRED cohort required higher amounts of supplemental oxygen during hospitalization (P < .001). No such differences were observed for children who started oral steroids before hospital arrival.

Children in the PRED cohort received an average dose of 1.8 mg/kg per day for a median of 2 days while hospitalized irrespective of when oral steroids were initiated. Children in the DEX cohort received an average dose of 0.5 mg/kg per day for a median of 1 day in the hospital if oral steroids were initiated before hospital arrival and 2 days if oral steroids were initiated after hospital arrival. For children who began oral steroids in the hospital, 94% of the PRED cohort were given an oral steroid prescription at discharge compared with only 10.7% of the DEX cohort (P < .001). A similar pattern was seen in children who began oral steroids before hospital arrival, with 95.9% of the PRED cohort and 4% of the DEX cohort receiving a steroid prescription on discharge. Of those discharged with an oral steroid prescription, children in the PRED cohort were prescribed an additional 3 days compared with an additional 1 day in the DEX cohort irrespective of when oral steroids were initiated.

Outcomes differed for children whose oral steroid course began before hospital arrival compared to those whose oral steroid course began after hospital arrival (Table 3). For children who received oral steroids before arrival, hospital LOS did not significantly differ between the 2 cohorts (adjusted mean LOS in hours, 26.72 [95% confidence interval [CI]: 21.62–33.03] for DEX versus 25.20 [95% CI: 20.49–30.99] for PRED, P = .45). However, for children who started oral steroids after hospital arrival, the DEX cohort had a significantly shorter hospital LOS compared with the PRED cohort (adjusted mean LOS in hours, 24.43 [95% CI: 21.13–28.25] for DEX versus 29.38 [95% CI: 24.15–35.73] for PRED, P = .03). There was no correlation between oral steroid dosing and hospital LOS in either cohort.

All secondary outcomes (PICU transfers, ED revisits, and hospital readmissions) were uncommon events in both cohorts (Table 4). Rates of PICU transfers (range: 0.0% to 1.7%) and rates of ED revisits (range: 0.0% to 2.2%) and hospital readmissions (range: 0.0% to 1.0%) within 10 days after discharge were statistically insignificant irrespective of when oral steroids were initiated.

For variables with missing data, all 3 analyses strategies (listwise deletion, reduced model, and adding unknown category) yielded similar results (Supplemental Table 5).

For the primary outcome, campus specific sensitivity analysis with employment of multiple missing data strategies was done for the 2 campuses with majority of the children. For children who did not receive oral steroids before hospital arrival, adjusted mean ratios for DEX versus PRED ranged from 0.78 to 0.91, falling within the CI of the main model. The P values for these estimates ranged from 0.02 to 0.55. Estimates for children who received oral steroids before arrival remained insignificant.

To our knowledge, this is the largest retrospective cohort study to date comparing clinical outcomes between the use of DEX and PRED in children hospitalized with mild-to-moderate acute asthma exacerbations. We found similar mean hospital LOS in children who started DEX or PRED before hospital arrival. However, children who started oral steroids after hospital arrival had significantly shorter mean hospital LOS when given DEX compared with PRED. Rates of PICU transfers during the initial hospitalization and rates of ED revisits and hospital readmissions within 10 days after discharge were all uncommon events. These findings support the use of DEX as a safe and efficacious alternative in treating children hospitalized with mild-to-moderate acute asthma exacerbations.

Although there was no statistically significant difference in hospital LOS for patients starting oral steroids before hospital arrival, there was a statistically significant difference in hospital LOS for patients who started oral steroids after hospital arrival. This 5-hour difference in hospital LOS for the latter group can be clinically significant for certain institutions wanting to improve financial, operational, and clinical outcomes. In previous inpatient observational studies, researchers have also noted significantly shorter hospital LOS in children receiving DEX compared with PRED but have not differentiated the timing of oral steroid initiation.^24,25  Variations in study design and statistical analyses may have contributed to different outcomes. Differences in asthma management and unaccountable external factors such as caregiver availability and transportation delays at discharge may have also affected hospital LOS.

For children who started oral steroids before hospital arrival, baseline asthma severity was more severe in the PRED cohort. Baseline asthma severity was similar between the 2 cohorts when oral steroids were initiated after hospital arrival. Acute asthma severity on presentation, as deemed by the initial ED CRS score, was similar for both cohorts irrespective of when oral steroids were initiated. Despite similar baseline and acute asthma severity in children starting oral steroids after hospital arrival, the DEX cohort more frequently required magnesium and ipratropium treatments during hospitalization, whereas the PRED cohort more often required higher supplemental oxygen. Although it is difficult to determine which cohort required more supplemental therapies during hospitalization, the DEX cohort had a significantly shorter hospital LOS even after controlling for these possible confounding variables. Randomized controlled trials are needed to better differentiate clinical outcomes between the 2 cohorts.

Although previous studies have revealed PRED and DEX to have comparable efficacy in managing acute asthma exacerbations, DEX has been reported to have better tolerability and palatability compared with PRED.^{12,16,18,21,22,29}  Pharmacologic properties of DEX also makes it an appealing alternative to PRED because of its less frequent dosing and shorter duration of therapy, which allows for administration of all, if not most, doses while hospitalized and thus ensures better adherence.^30,31  Several studies have revealed poor prescription fill rates after an ED visit or hospitalization for acute asthma exacerbations.^32–34  Even when prescriptions are filled, caregivers often fail to adhere to the prescribed duration of therapy.³⁵  Caregivers also prefer shorter steroid courses.^12,17,18,36  In our study, a significantly higher percentage of the PRED cohort was discharged from the hospital with an oral steroid prescription compared with the DEX cohort irrespective of when oral steroids were initiated (94% to 95.9% for PRED versus 4% to 10.7% for DEX, P < .001). Put together, these factors make DEX a desirable and less burdensome option for families, allowing for better adherence with completion of oral corticosteroid course while inpatient.^4,6–8 

Our study has several limitations. Retrospective study design renders it susceptible to confounding by indication despite accounting for multiple potential confounding variables such as baseline asthma severity and acute asthma severity on presentation. Although our study included children with moderate persistent and severe persistent baseline asthma severity, the majority of the children were classified as having mild intermittent or mild persistent asthma. Moreover, our study exclusion criteria presumably excluded those with more severe asthma exacerbation at presentation. In our study, we did not collect data on previous inhaled corticosteroid use at home. Although risk for confounding by indication may be insignificant given the overall poor adherence with home inhaled corticosteroid, there may be meaningful differences in certain demographic populations not addressed by this study.^6,24,33,34  Finally, our study was not designed to follow-up with children after discharge to track adherence with home oral steroid prescriptions.

Future studies are needed to further examine the safety and efficacy of DEX in treating children hospitalized for acute asthma exacerbations, particularly those with severe exacerbations. Randomized controlled trials that minimize inherent biases associated with retrospective studies are needed. Additional comparative data would also help minimize variability in oral steroid prescribing practices currently seen among different subspecialists caring for children hospitalized with acute asthma exacerbations.^26,27 

Children hospitalized with mild-to-moderate asthma exacerbations have significantly shorter hospital LOS when starting DEX rather than PRED after hospital arrival. Rates of PICU transfers, ED revisits, and hospital readmissions were uncommon events in both cohorts. Children receiving DEX more frequently complete their oral steroid course before discharge, allowing for better treatment adherence. Future prospective studies are needed to further assess outcomes of DEX versus PRED, especially in children hospitalized with severe asthma exacerbations.

The authors thank our clinical librarians, Emily Lawson and Christine Willis, for assisting us with our literature search and review.