Browse the corpus

Azithromycin for Preschoolers with Wheezing in the Emergency Department. BACKGROUND: Wheezing illnesses are a leading cause of hospitalization for preschool-age children and are frequently treated with antibiotics. Observational studies have shown more frequent isolation of three pathogenic bacteria (Streptococcus pneumoniae, Moraxella catarrhalis, and Haemophilus influenzae) from nasopharyngeal samples from children with recurrent episodes of wheezing than from those without such illnesses. METHODS: In this multicenter trial, we randomly assigned patients 18 to 59 months of age who presented to an emergency department with a moderate-to-severe episode of wheezing to receive azithromycin once daily at a dose of 12 mg per kilogram of body weight or matching placebo for 5 days. The primary outcome was the sum of scores on the Asthma Flare-up Diary for Young Children (ADYC) over 5 days. Primary-outcome scores could range from 5 to 35, with higher scores indicating more severe wheezing-related symptoms. Efficacy was assessed separately in patients who tested positive for pathogenic bacteria (the positive cohort) and in those who tested negative (the negative cohort). Secondary outcomes were length of stay in the emergency department, length of hospital stay, and return emergency department visits or hospitalizations within 72 hours. Bacterial clearance and antimicrobial resistance were measured at follow-up visits 1 to 3 weeks after randomization. RESULTS: Among 840 patients who underwent randomization, 521 tested positive for pathogenic bacteria. The trial was stopped for futility by the data and safety monitoring board after a planned interim analysis. ADYC scores did not differ significantly between the azithromycin and placebo groups in either the positive cohort (median, 9.59 [interquartile range, 7.29 to 12.60] vs. 9.72 [interquartile range, 7.66 to 12.17]; P = 0.70) or the negative cohort (median, 9.30 [interquartile range, 6.97 to 11.62] vs. 9.10 [interquartile range, 7.19 to 11.45]; P = 0.69). In the positive cohort, bacterial clearance was 58.7% in the azithromycin group and 11.4% in the placebo group. Secondary outcomes appeared to be similar in the two groups for both cohorts, as did the development of bacterial resistance and the incidence of adverse events. CONCLUSIONS: Azithromycin did not lead to a greater reduction in the severity of wheezing-related symptoms than placebo in preschool-age children who presented to the emergency department with moderate-to-severe acute wheezing. (Funded by the National Heart, Lung, and Blood Institute and others; AZ-SWED ClinicalTrials.gov number, NCT04669288.).

Lower respiratory illnesses presenting with wheezing are a major cause of morbidity and health care utilization in early life. Hospitalization rates with a diagnosis of asthma are highest in children less than 5 years of age, with approximately 30,000 such admissions in the United States in 2019.1 Most young children who have acute episodes of wheezing have concomitant viral infections, predominantly due to rhinovirus, after infancy.2 Recently, several studies have reported high rates of nasopharyngeal isolation of three pathogenic bacteria (S. pneumoniae, M. catarrhalis, H. influenzae) in preschool children with recurrent episodes of wheezing.3–5 This suggested the possibility that these bacteria may play a role in the pathogenesis of acute lower respiratory illnesses in this age group. Based on this assumption, several randomized, placebo-controlled trials have assessed the effectiveness of azithromycin for the treatment of preschool children with lower respiratory illnesses of different levels of severity, and the results have been mixed. In children with recurrent wheezing, Stokholm et al6 and Bacharier et al7 showed that azithromycin initiated early in the course of illness, was effective in decreasing duration and use of oral corticosteroids when compared with placebo. Conversely, a trial performed in an emergency department (ED) found that azithromycin did not reduce the duration of symptoms in children aged 12–60 months with wheezing.8

thromycin initiated early in the course of illness, was effective in decreasing duration and use of oral corticosteroids when compared with placebo. Conversely, a trial performed in an emergency department (ED) found that azithromycin did not reduce the duration of symptoms in children aged 12–60 months with wheezing.8 Azithromycin has both anti-inflammatory and antibacterial effects,9 and its potential role in preschool wheezing illnesses could be attributable to one or both of these properties. None of the studies described above assessed efficacy in association with nasopharyngeal bacterial colonization, which may help to elucidate this question. Given that up to one-fourth of wheezing episodes among preschoolers are treated with antibiotics,10 we conducted the Azithromycin Therapy in Preschoolers with a Severe Wheezing Episode Diagnosed at the Emergency Department (AZ-SWED) study, a randomized, placebo-controlled, multicenter clinical trial. The objective of the trial was to evaluate whether a 5-day course of azithromycin decreases the severity of illness in young children presenting to the ED with wheezing, separately in those with and without nasopharyngeal colonization with S. pneumoniae, M. catarrhalis, or H. influenzae.

The AZ-SWED trial was conducted at eight pediatric EDs in the U.S. affiliated with the Pediatric Emergency Care Applied Research Network (PECARN). Written permission was obtained from the children’s parents/guardians. The University of Utah functioned as the single institutional review board (sIRB), and all sites relied on this sIRB and approved the trial protocol (see nejm.org for the full protocol). A Data and Safety Monitoring Board (DSMB) was assigned by the National Heart, Lung, and Blood Institute (NHLBI) and this board periodically reviewed enrollment, study conduct, safety, and efficacy. The role of each author is described in the Supplement.

The target population was children aged 18 to 59 months presenting to the ED with moderate/severe episodes of expiratory wheezing (as ascertained by ED care providers). Children with a Pediatric Respiratory Assessment Measurement (PRAM)11 score ≥4 at any time during the ED visit, indicating moderate to severe symptoms, were eligible. PRAM assessments were made as soon as possible after ED arrival by trained care providers (e.g., physicians, nurse practitioners and trained nurses) and were periodically updated The most important exclusion criterion was presentation to the ED with an acute infection requiring systemic antibiotics or receipt of antibiotics for any reason within the prior 2 weeks. Additional exclusion criteria included use of a systemic glucocorticoid for wheezing or suspected foreign body aspiration within the prior 2 weeks, prematurity, known systemic illness other than allergies, known sensitivity or allergy to azithromycin, and active COVID-19 infection. The complete list of exclusion criteria is detailed in the protocol.

Patients were randomized to receive one 12 mg/kg dose per day of oral azithromycin or placebo for 5 days. Masked bottles of azithromycin or placebo, prepared by Sharp Services (Allentown, PA), were matched to be similar in taste, color, and smell. Randomization was performed through a built-in module of the Research Electronic Data Capture (REDCap) System using permuted blocks with random block sizes. Randomization was stratified by clinical center and qualifying PRAM score (4–7 vs. ≥8). Nasopharyngeal samples were obtained prior to the administration of azithromycin or placebo and later tested for respiratory viruses and for H influenzae, M catarrhalis, and S pneumonia (Supplementary Methods). Randomization could not be stratified by the presence of the 3 pathogenic bacteria, as these test results were not available at the time of eligibility assessment. All participants were expected to receive their first dose of study drug orally in the ED, or shortly thereafter on the inpatient unit if admitted, and all were prescribed oral/systemic glucocorticoids along with albuterol treatment in accordance with NHLBI guidelines.

The primary outcome was the sum of wheezing-related symptoms over the 5 days following randomization as reported by caregivers on the Asthma Flare-up Diary for Young Children (ADYC), a previously validated web-based measure that includes 17 questions rated on a 1-to-7 scale. Higher scores indicate more severe symptoms. Daily ADYC scores were calculated as the average score per question and the primary outcome calculated as the sum of the 5 daily scores. The range of possible primary outcome values was 5–35.12 There were three predefined secondary outcomes: length of ED stay, length of hospital stay, and return visits (ED visits or hospital admissions) within 72 hours after randomization. Data on study drug adherence were also collected. A safety-related follow-up survey was completed on Day 8. Prior to randomization, caregivers were asked whether the child would participate in optional follow-up visits. Those who agreed and were subsequently found positive for bacteria were contacted to return to the site for follow-up visits at 5–8 days and again at 14–21 days following the ED enrollment. Nasopharyngeal swabs were collected at these follow-up visits to test for bacterial presence and, when found, antimicrobial resistance. Patients found positive for at least one of these three bacteria were classified as “bacterial”, while those negative for all three were classified as “non-bacterial”. Study outcomes were compared separately in the bacterial and non-bacterial cohorts.

Prior to randomization, caregivers were asked whether the child would participate in optional follow-up visits. Those who agreed and were subsequently found positive for bacteria were contacted to return to the site for follow-up visits at 5–8 days and again at 14–21 days following the ED enrollment. Nasopharyngeal swabs were collected at these follow-up visits to test for bacterial presence and, when found, antimicrobial resistance. Patients found positive for at least one of these three bacteria were classified as “bacterial”, while those negative for all three were classified as “non-bacterial”. Study outcomes were compared separately in the bacterial and non-bacterial cohorts. Bacterial resistance to any of the 3 pathogenic bacteria was a safety outcome assessed in the subgroup of the bacterial cohort that returned for at least one follow-up visit. All adverse events were also reported.

The full statistical analysis plan is available in the Supplemental Information. Briefly, the target sample size was 1,476 participants. Differences with respect to the sum ADYC score were compared between treatment arms using a Van Elteren test, stratified by site and presenting PRAM category. A Bonferroni-adjusted significance level of 0.025 was used for each of the two primary comparisons (i.e., bacterial and non-bacterial cohorts). To account for missing data, we applied multiple imputation. Based on this analysis plan, 10 full datasets were generated with fully conditional expectation, using backward selection to choose sufficiently predictive models for each variable given the others. A complete-case analysis was also performed, limited to patients with no missing data. The secondary outcomes of ED and hospital length of stay were analyzed in the same manner as the primary outcome, whereas return visits were analyzed using a Mantel–Haenszel test, stratified according to site and PRAM category. As there were very few missing data points for secondary outcomes, complete-case analysis was used. Significance thresholds of secondary outcomes were adjusted for multiple comparisons with the use of the Holm procedure.13 Bacterial resistance was analyzed using standard methods (see Supplemental Information).

The full statistical analysis plan is available in the Supplemental Information. Briefly, the target sample size was 1,476 participants. Differences with respect to the sum ADYC score were compared between treatment arms using a Van Elteren test, stratified by site and presenting PRAM category. A Bonferroni-adjusted significance level of 0.025 was used for each of the two primary comparisons (i.e., bacterial and non-bacterial cohorts). To account for missing data, we applied multiple imputation. Based on this analysis plan, 10 full datasets were generated with fully conditional expectation, using backward selection to choose sufficiently predictive models for each variable given the others. A complete-case analysis was also performed, limited to patients with no missing data. The secondary outcomes of ED and hospital length of stay were analyzed in the same manner as the primary outcome, whereas return visits were analyzed using a Mantel–Haenszel test, stratified according to site and PRAM category. As there were very few missing data points for secondary outcomes, complete-case analysis was used. Significance thresholds of secondary outcomes were adjusted for multiple comparisons with the use of the Holm procedure.13 Bacterial resistance was analyzed using standard methods (see Supplemental Information). Significance levels were adjusted for multiple subgroups (sex and race/ethnicity). Consistency of results was also analyzed across sites and presenting PRAM categories. We used IVEware software (University of Michigan) for imputation and SAS software, version 9.4 (SAS Institute), for all other analyses. This report describes the trial and presents results according to the 2010 Consolidated Standards of Reporting Trials (CONSORT) guidelines.14

From study initiation in September 2021 to December 2024, 840 patients were enrolled. In December 2024, as specified in the protocol, the DSMB reviewed the results of a futility analysis, and determined that the conditional power of the trial in both cohorts was less than 16%, even when assuming a true reduction in sum ADYC score (primary outcome) due to azithromycin as much as 1.3 points. Based on this finding, the NHLBI, on the recommendation of the DSMB, stopped the trial for futility.

At trial stoppage, 840 patients had been randomized, with 521 of these positive for one or more of the three pathogenic bacteria (62.0%), 312 negative, and 7 of unknown bacterial status. Participant flow is shown in Figure 1. In each cohort and assigned treatment group, ≥80% of patients completed all daily surveys, and ≥94% completed at least one survey. The demographics and baseline characteristics of randomized children are shown in Table 1. They were generally representative of similarly affected children in the United States (Table S14 in Supplementary Information). Respiratory viruses were isolated from the nasopharynx in 86.1% of participants. The most frequently associated virus was rhinovirus (72.5%), followed by respiratory syncytial virus (9.2%), bocavirus (4.6%), and metapneumovirus (3.7%). The isolation rate of rhinovirus was similar between the positive and negative bacteria groups (see Table S9, Supplemental Information). Adherence to assigned treatment (defined as parental report of the patient receiving at least 4 out of 5 daily doses) was 78.3% in the bacterial cohort and 80.4% in the non-bacterial cohort.

There was no significant difference in the sum of ADYC scores over 5 days between the azithromycin and placebo groups for either the bacterial or the non-bacterial cohorts (Table 2 and Figure 2). Only 11.7% of ADYC questionnaire data points from the bacterial cohort and 11.2% from the non-bacterial cohort were missing and required imputation, with similar missingness across treatment arms. The results of complete-case analysis and per-protocol analysis were similar to the primary analysis (Table S2 and Figure S5 in Supplementary Information).

Approximately half of the randomized participants (417/840) were hospitalized, and rates were similar for the bacterial/azithromycin subgroup (123/257 48%) as for the bacterial/placebo subgroup (138/263,52%), the non-bacterial/azithromycin subgroup (77/158,49%) and the non-bacterial /azithromycin subgroup (78/154,51%). Length of ED and hospital stay appeared to be similar in the azithromycin and placebo groups in both the bacterial and non-bacterial cohorts (Table 2 and Figure S1 in Supplemental Information). Return visits were also appeared to be similar between arms. Prespecified subgroup analyses of the primary and secondary outcomes by sex and race/ethnicity, as well as exploratory subgroup analyses are shown in Supplemental Figures S2–S5.

Of the 521 participants in the bacterial cohort, 444 (85%) agreed to participate in follow-up visits. Of those, 116 (26%) returned for the one-week follow-up visit and 188 (42%) returned for the follow-up visit at 2–3 weeks, resulting in 209 (47%) with at least one follow-up visit to assess bacterial clearance and resistance. Clearance of the three pathogenic bacteria from the upper airway was 58.7% in the azithromycin arm and 11.4% in the placebo arm (Table S4 in Supplementary Information). The effect of azithromycin on ADYC appeared to be similar in those found positive and those found negative for bacteria at one-week follow-up (Table S5 in Supplementary Information). Among patients with at least one follow-up visit, 16 (16.2%) in the azithromycin arm and 19 (17.3%) in the placebo arm were found to have bacterial resistance by at least one of the three pathogenic bacteria at baseline and thus were excluded from the analysis of development of bacterial resistance. Bacterial resistance at follow-up was present in 21.7% of the azithromycin and 22% of the placebo group samples (Table S3 in Supplementary Information).

In the bacterial cohort, three patients in the azithromycin arm and five patients in the placebo arm experienced serious adverse events, consisting of respiratory distress, bronchospasm, increased work of breathing, and RSV infection. In the non-bacterial cohort, three patients in the azithromycin arm and one patient in the placebo arm experienced serious adverse events, consisting of COVID-19 pneumonia, respiratory distress, respiratory failure, septic shock, and increased cough (see Table S6 in Supplementary Information). Non-serious adverse events were also evenly distributed between arms in both cohorts (Table S7). No deaths were reported.

An important finding of our study was that 50% of preschool children presenting to the Emergency Department with wheezing and a PRAM of ≥4 were admitted to the hospital for treatment. This result confirms our initial assumption that, notwithstanding recent advances in our understanding of the etiology and heterogeneity of acute wheezing illnesses,15 they remain a substantial cause of morbidity and healthcare costs, and new approaches are needed to prevent and treat them.

the hospital for treatment. This result confirms our initial assumption that, notwithstanding recent advances in our understanding of the etiology and heterogeneity of acute wheezing illnesses,15 they remain a substantial cause of morbidity and healthcare costs, and new approaches are needed to prevent and treat them. Viral associated wheezing is the most frequent pathogen found in this age group; in our study, 72.5% of participants tested positive for rhinoviruses, while only 13.9% tested negative for all of the viruses we assayed in the nasopharynx. Several studies reported that at least one of three pathogenic bacterial taxa, S. pneumoniae, M. catarrhalis, and H. influenzae, was more likely present in children with a history of recurrent wheezing illnesses than in controls5. This suggested the hypothesis that these bacteria could play a role in the pathogenesis of these illnesses, and therefore, that antibiotics such as azithromycin could decrease their severity10. We show that, in the case of episodes for which ED attention was sought, the severity and clinical outcome of the wheezing episode were not different between children treated for 5 days with azithromycin as compared with those treated with placebo. This was true regardless of whether any of the three bacteria were present or absent. The lack of efficacy met the pre-established conditional power threshold for both arms of the trial and, in accordance with the planned interim analysis, the Data Safety and Monitoring Board recommended early stopping after 840 participants had been randomized.

whether any of the three bacteria were present or absent. The lack of efficacy met the pre-established conditional power threshold for both arms of the trial and, in accordance with the planned interim analysis, the Data Safety and Monitoring Board recommended early stopping after 840 participants had been randomized. Findings in this study are similar to those of Mandhane et al8 who, in a trial conducted in a similar setting but with a sample of 222 preschoolers and no stratification by bacterial colonization, showed no effect of azithromycin on the duration of symptoms in wheezy preschool children treated in the ED. Our results differ from those of two outpatient studies involving children with milder wheezing episodes. Bacharier et al7 compared the frequency of oral corticosteroids use in 607 preschool children who were given azithromycin or placebo at the first signs that preceded the development of a wheezing illness. The azithromycin arm demonstrated a 36% reduced risk of progressing to severe illness relative to placebo. The number needed to treat (NNT) to prevent one severe illness was higher for the first respiratory tract infection (RTI) (NNT = 33) than for the fourth RTI (NNT = 7), suggesting that this approach has justification in children with a history of severe recurrent illnesses. Stokholm et al6 found that azithromycin therapy was associated with a 63.3% shortening of wheezing episode duration relative to placebo. In our RCT of children receiving ED treatment, we found that severity decreased substantially after the first day, becoming quite mild by the fifth day (Figure 2b), with similar trajectories between treatment groups.

omycin therapy was associated with a 63.3% shortening of wheezing episode duration relative to placebo. In our RCT of children receiving ED treatment, we found that severity decreased substantially after the first day, becoming quite mild by the fifth day (Figure 2b), with similar trajectories between treatment groups. One potential explanation for the discrepancy in the results between settings (i.e., emergency department versus outpatient care) is the severity of the illness, with azithromycin being able to have an effect when the respiratory infection has not yet caused severe lower airway obstruction. In support of this is the fact that in the study by Bacharier et al., azithromycin was given before any clinical signs of airway obstruction were present, and in the study by Stokholm et al., the effect of azithromycin was stronger if the drug was taken early in the course of the illness. Azithromycin is known to have anti-inflammatory effects in addition to its defining antibacterial effects.16 Our large RCT did not demonstrate benefit of azithromycin in children with the 3 pathogenic bacteria, despite substantial bacterial clearance from the nasopharynx in the azithromycin group. Based on that observation, we speculate that azithromycin may have been more effective in the outpatient setting than in our more severe ED cohort due to the drug’s preventive anti-inflammatory effects, and should not be used to treat children with established respiratory distress.

asopharynx in the azithromycin group. Based on that observation, we speculate that azithromycin may have been more effective in the outpatient setting than in our more severe ED cohort due to the drug’s preventive anti-inflammatory effects, and should not be used to treat children with established respiratory distress. Our trial has some limitations. While the decision to stop the trial early was justified, given the interim results, trail stoppage limited its sample size, decreased the precision of estimated differences, and greatly reduced power. It should be noted, however, that the results show that a moderate or large true effect is implausible. Due to sample size constraints and the preliminary evidence supporting azithromycin, we did not test the effectiveness of other antibiotics, but as noted, the differences in bacterial clearance between arms do not support a role for antibacterials in these episodes. In summary, compared to placebo, azithromycin did not reduce wheezing-related symptoms in preschool children presenting to the emergency department with moderate/severe acute wheezing.