The Effect of Chronic Azithromycin Use in Winter on Health Care Utilization for Children With Bronchopulmonary Dysplasia: A Double Blind Randomized Controlled Study (RCT)

ABSTRACT

Background

Bronchopulmonary dysplasia (BPD), a chronic lung disease in preterm infants, often leads to acute respiratory exacerbations triggered by infections. Our previous mouse study suggested that azithromycin's anti‐inflammatory properties may benefit virus‐induced respiratory illnesses prophylactically.

Methods

In this single‐center, double‐blind RCT, 60 children (6 months to 6 years) with BPD received azithromycin (5 mg/kg three times weekly; n = 30) or placebo (n = 30) for one winter season (October to March). Primary outcome was the total number of days of unscheduled healthcare clinic visits, ER visits, and hospital days. Secondary outcomes comprised clinic visits, ER visits, hospital admissions, hospital days, and adverse events. Standard frequentist and Bayesian analyses were used.

Results

No significant difference in primary outcomes, unscheduled healthcare visits (macrolides 14.9 vs. 4.2 per child‐year; p = 0.1, IRR = 2.1; 95% CI: 0.8−5.4), Bayesian analysis showing 11% probability of benefit; however, high‐risk children (< 2 years, no tracheostomy), rates were lower with the macrolide group (2.7 vs. 4.6 per child‐year; IRR = 0.6; 95% CI: 0.2–1.5), with an 82% probability of benefit. Two outliers in the macrolide group stayed > 40 days in the hospital for non‐medical reasons. For secondary outcomes, we observe a decrease in the intervention group on ER visits (0.5 vs. 1.3 per child‐year; p = 0.047, IRR = 0.4; 95% CI: 0.1−0.9). Hospital days increased (12.1 vs. 1.2 per child‐year; IRR = 9.3; 95% CI: 5.5−16.8).

Conclusions

Macrolide prophylaxis did not lead to a significant reduction in the primary outcome of unscheduled healthcare encounter days, but subgroup analysis suggests a potential benefit in high‐risk children, with 82% probability of benefit.

Trial Registration: NCT02544984.

1. Introduction

Bronchopulmonary dysplasia (BPD) is a chronic lung disease (CLD) affecting preterm infants, characterized by neonatal lung injury and persistent need for respiratory support [1]. Survivors of BPD often endure long‐term pulmonary impairments, including oxygen dependency, airway reactivity, and increased susceptibility to respiratory infections [2, 3, 4]. In early childhood, particularly during the winter viral season, these children are at a high risk for severe respiratory illnesses [3]. Readmissions are common; a high‐risk clinic reported that 37% of hospitalizations in infants with BPD were due to respiratory infections during winter months [5]. This occurs despite standard prophylactic measures, including routine immunizations and respiratory syncytial virus (RSV) monoclonal antibody (palivizumab) for eligible infants [5, 6]. Additional strategies are needed to reduce infection‐related morbidity, especially during peak winter viral circulation.

Azithromycin, a macrolide with antimicrobial and anti‐inflammatory effects, is being explored as a prophylactic agent in CLDs [3]. Macrolides can down‐regulate excessive inflammation and cytokine release triggered by respiratory viruses, mitigating neutrophil‐driven lung injury and damage [7, 8, 9]. These effects have been studied in other airway disorders. For example, long‐term macrolide therapy improves outcomes in diffuse panbronchiolitis in cystic fibrosis patients [10, 11, 12] and has shown benefits in some patients with severe asthma or post‐transplant bronchiolitis obliterans [3, 12]. This rationale suggests azithromycin may benefit children with BPD, whit ongoing airway inflammation and recurrent infections [13, 14].

A multicenter AZTEC trial in the UK, involving 796 infants < 30 weeks' gestation, found that a 10‐day azithromycin course did not improve survival free of moderate/severe BPD (42% vs. 45% event‐free survival in azithromycin vs. placebo) [15]. However, the trial reduced the incidence of retinopathy of prematurity (ROP). A systematic review found azithromycin reduced oxygen supplementation in BPD patients [16]. A recent meta‐analysis in preterm infants found that azithromycin prophylaxis did not significantly reduce overall BPD or death but showed a significant benefit in the Ureaplasma‐positive subgroup [16]. Recent evidence from a large adaptive cluster‐randomized trial showed significant mortality reduction with azithromycin in children 1−59 months (particularly > 11 months), highlighting its potential benefits in vulnerable population [6]. However, the chronic outpatient use of azithromycin after the neonatal period in established BPD and outpatient settings has not been well studied to date. Many BPD survivors continue to have respiratory symptoms and repeated infections beyond infancy [17]. Other factors, such as race/ethnicity and public insurance, are more likely to influence outpatient respiratory outcomes, irrespective of gestational age [18].

Preclinical research supports the biological plausibility that chronic azithromycin therapy can mitigate virus‐induced lung injury, by dampening the immune overreaction. In our basic science study, prophylactic azithromycin blunted the inflammatory response in a murine model of viral bronchiolitis. Mosquera et al. demonstrated that azithromycin administered to BALB/c mice pre‐ and during RSV infection reduced weight loss, neutrophilic airway inflammation, cytokine levels, and prevented mortality (0% compared to 8% in untreated RSV mice) [17, 19]. Another study in a neonatal rat model of BPD (hyperoxia‐induced lung injury) found that the macrolide erythromycin enhanced lung antioxidant defenses (increasing glutathione) and inhibited pro‐inflammatory cytokines (TNF‐α, IL‐1β), reducing lung pathology [20]. These animal data suggest macrolides not only fight infections but also directly protect the developing lung from inflammatory damage.

Together, human and animal evidence supports investigating the potential of chronic azithromycin use during winter can reduce respiratory infections and improve pulmonary outcomes in young children with BPD. Therefore, we designed a clinical trial to evaluate winter‐season prophylactic azithromycin in children aged 6 months to 6 years with BPD, assessing its effectiveness in reducing illness through decreased hospital utilization and its safety profile over several months of use.

2. Methods

2.1. Study Site, Design, and Population

We conducted a single‐site, double‐blind RCT of children receiving primary care (medical home) at either the High‐Risk Children's Clinic (HRCC) or the High‐Risk Infant Clinic (HRIC) at The University of Texas Health Science Center at Houston (UTHealth) at McGovern Medical School in Houston, Texas, USA. During two respiratory virus seasons, defined as October 1 to March 31 of each year (2015−2016 season and 2016−2017 season), we enrolled children who met the following inclusion criteria: (1) age 6 months to 6 years during the respiratory virus season (October 1 to December 31), and (2) diagnosis of CLD secondary to BPD as defined by the American Thoracic Society (ATS) [21]. Children with conditions known to benefit from macrolides or at risk for adverse effects were excluded, including those with cystic fibrosis, bronchiectasis, cardiac arrhythmias, cyanotic heart disease, colitis, short bowel syndrome, renal/hepatic failure, known macrolide allergy, and those taking interacting medications.

2.2. Study Intervention and Procedures

After screening for eligibility, patients were approached during a routine clinic visit in our medical home. If the patient was interested in participating in the study, a baseline electrocardiogram (ECG) was performed to ensure that enrolled patients did not have a prolonged QT interval or other undiagnosed arrhythmia. If the ECG was normal, written informed consent was obtained from the parent or legal guardian of each eligible child.

Once a patient was deemed eligible, he or she was randomized to either azithromycin or placebo using the Research Electronic Data Capture (REDCap) module. The allocation ratio was 1:1 and was stratified by very high‐risk kids (Age > 2 years or has tracheostomy) or high‐risk kids (Age < 2 years and does NOT have a tracheostomy), Less than 2 years of age was a surrogate for Palivizumab use during wintertime. The statistician created the randomization sequence using A/B labels and uploaded it to REDCap, ensuring allocation concealment as the sequence was only revealed after recruitment. Patients were recruited and enrolled into the study on a rolling basis from October 1 through December 31. Once enrolled, all participants completed the intervention phase of the protocol on March 31. Half of the patients received azithromycin at a dose of 5 mg/kg, to be taken once daily on Monday, Wednesday, and Friday. The other half, the control group, received a placebo drug with similar taste, color, texture, and consistency, also to be taken once daily on Monday, Wednesday, and Friday. Both the study drug and the placebo were oil‐fish based to ensure a shelf life of more than 6 months and were flavored with citrus to improve palatability. Planned oscillometer studies and laboratory evaluations were designed to include assessments such as myeloperoxidase (MPO) levels, cytokines, and respiratory virology. Parents were contacted monthly to monitor progress and potential side effects. If adverse reactions occurred, medication was discontinued. Safety monitoring included unblinding procedures for allergic reactions, with a statistician not involved in patient care breaking the blind to determine causation. At study conclusion, a final ECG was obtained.

3. Outcomes

Our primary outcome was the total number of days of unscheduled healthcare‐related encounters for all diagnoses (defined as unscheduled sick visits, urgent care visits, ER visits, and hospital admissions) during the 3–6‐month treatment phase of the study.

The secondary outcomes were the following:

• 1.Individual components of the primary outcome (unplanned medical visits, urgent care visits, ER visits, hospital admissions rate, and hospital days).
• 2.Adverse side effects, including gastrointestinal upset (vomiting/diarrhea) and diaper rash.

3.1. Data Collection, Management and Analysis Plan

All data were stored in the REDCap Baseline data, which included demographics and ECG findings. Monthly phone calls assessed medication adherence and side effects. Data entry was restricted to the research coordinator, with medical records verified at study completion to ensure accurate event capture. Intent‐to‐treat analyses (frequentist and Bayesian) assessed treatment effects (azithromycin vs. placebo) on unscheduled encounters, admissions, and visits using Poisson regression. Models adjusted for covariates and follow‐up time. Bayesian models evaluated moderator effects (age, tracheostomy). Prior distributions were centered on RR 1.0. Results include RRs, CIs, and benefit probabilities.

3.2. Sample Size and Power

Based on data from our clinic outcomes, we expected the placebo group to have 1.6 encounters per child‐year (SD = 1.66). Assuming a two‐sided α of 0.05, a sample size of 92 (46/group) had 80% power to detect a difference of 1 in the encounter rate between the placebo and azithromycin groups (i.e., 1.6 vs. 0.6 encounter rate or 38% reduction). Several outcome measurements from the original protocol were omitted during implementation. The final study enrolled 60 participants (30 per group) rather than the planned 92 participants due to recruitment challenges and stricter eligibility criteria. While core clinical endpoints were maintained, most participants did not complete planned oscillometer studies and laboratory evaluations, including MPO levels, cytokines, and respiratory virology due to technical challenges and resource constraints. The planned cost‐effectiveness analysis was simplified to focus on direct healthcare utilization metrics.

4. Results

Between October 2015 and March 2017, we screened 69 children with BPD for eligibility. Of these, nine were excluded: three were ineligible, and six declined to participate. The remaining 60 eligible participants were randomized in a 1:1 ratio to receive either azithromycin (n = 30) or placebo (n = 30). During the follow‐up period, 28 participants in the azithromycin group (93.3%) and 28 participants in the placebo group (93.3%) completed the full‐time study protocol. All 60 randomized participants were included in the final analysis (Figure 1).

Figure 1.

Flow diagram for patient enrollment, randomization, and follow up.

4.1. Primary Outcome

All demographic factors and strata were balanced between the Macrolide group (n = 30 followed for a total of 11.4 child‐years) and placebo group (n = 30 followed for a total of 11.3 child‐years) (Table 1). Overall, no treatment effect was seen on the primary outcome (total number of days of unscheduled face‐to‐face encounters: clinic, ER visits and hospital days): 168 days (14.9 per child‐year) in Macrolide group vs 48 days (4.2 per child‐year) in placebo group; IRR = 2.1 (0.8, 5.4) (Table 2). Bayesian analysis with a neutral prior showed similar results with an estimated treatment effect of IRR = 1.5 (0.8, 3.1) and a low 11% probability of benefit from Macrolides treatment in terms of primary outcome; however, when stratified by clinical risk, a notable difference emerged. Among post hoc defined strata of very high‐risk patients (Age > 2 years or has tracheostomy at enrollment), there was a significant effect of harm from the treatment with macrolides; IRR = 4.6 (1.1, 20.2) compared to placebo, with 8% probability of benefit, in terms of primary outcome. In contrast, among high‐risk patients (with no tracheostomy and ≤ 2 years of age at enrollment), there was a trend toward benefit with macrolide therapy with IRR = 0.6 (0.2, 1.5). While this did not reach statistical significance by frequentist analysis (p = 0.237), Bayesian analysis revealed a promising 82% probability of benefit in this subgroup.

Table 1.

Baseline characteristics by treatment group.

Type of treatment. No (%)
	Placebo (n = 30)	Macrolide (n = 30)	Total (n = 60)
Age ranges
6−24 month	19 (63.3)	19 (63.3)	38 (63.3)
24−72 month	11 (36.7)	11 (36.7)	22 (36.7)
Sex
Female	16 (53.3)	11 (36.7)	27 (45)
Male	14 (46.7)	19 (63.3)	33 (55)
Race
Asian	1 (3.3)	1 (3.3)	2 (3.3)
Black/African American	9 (30)	10 (33.3)	19 (31.7)
Caucasian	1 (3.3)	2 (6.7)	3 (5)
Unknown/not reported	19 (63.3)	17 (56.7)	36 (60)
Gestational age
< 28 weak	22 (73.3)	20 (66.7)	42 (70)
28−31 weak	6 (20)	4 (13.3)	10 (16.7)
32−36 weak	2 (6.7)	6 (20)	8 (13.3)
Tracheostomy
Yes	5 (16.7)	5 (16.7)	10 (16.7)
No	25 (83.3)	25 (83.3)	50 (83.3)
Synagis injection
Yes	14 (46.7)	16 (53.3)	30 (50)
No	16 (53.3)	14 (46.7)	30 (50)

Table 2.

Primary outcome by treatment group.

Outcome measure	Macrolide intervention (n = 30)a		Placebo control (n = 30)b		Rate ratio (95% Cl)c	p value	Bayesian rate ratio (95% CI)	Bayesian probability of reduction (%)
	No.	Rate/1 child‐year	No.	Rate/1 child‐year
Total unscheduled face to face encountersd (Clinic visit + ER + Hospital admission days)	168	14.9	48	4.2	2.1 (0.8, 5.4)	0.114	1.5 (0.8, 3.1)	11%
Very high‐risk group age > 2 years or has tracheostomy (n = 27)	152	28.7	20	3.8	4.6 (1.1, 20.2)	0.032	NA	8%
High risk group: age < 2 years and does not have tracheostomy (n = 33)	16	2.7	28	4.6	0.6 (0.2, 1.5)	0.237	NA	82%

Note: Admission days where estimates are from Poisson model due to lack of convergence with negative binomial.

^a There were 11.4 child‐years;

^b There were 11.3 child‐years;

^c Incident rate ratios and p values from negative binomial model adjusting for palivizumab and tracheostomy, except for outcome Hospital;

^d Includes unscheduled respiratory sick visits, other sick visits, ER visits, and admission days.

There were 2 children with long hospital admissions (> 40 days) in the Macrolide group (one with tracheostomy treated with palivizumab, one without tracheostomy not treated with palivizumab) that greatly influenced the encounter days in the Macrolide group (non‐medical reasons). Sensitivity analysis excluding these 2 children resulted in a potentially favorable treatment effect with IRR 0.6 (0.3, 1.3). Bayesian sensitivity analysis excluding these outliers showed an IRR = 0.7 (0.4, 1.3) with a substantially higher 86% probability of benefit, suggesting the overall results were considerably impacted by these two cases.

4.2. Secondary Outcomes

Types of unscheduled days varied between groups, with a significant benefit observed in emergency room utilization. The number of ER visits was lower in the Macrolide group (6 visits; 0.5 per child‐year) compared to the placebo group (15 visits, 1.3 per child‐year); IRR = 0.4 (0.1, 0.9) (Table 3). However, hospital admissions days were higher in the Macrolide group (136 days; 12.1 per child‐year) compared to the placebo group (14 days, 1.2 per child‐year); IRR = 9.3 (5.5, 16.8). There was no difference in rate of outpatient unscheduled respiratory visits, unscheduled other visits, number of hospital admissions, or adverse events between groups (Table 3).

Table 3.

Secondary outcome by treatment group.

Outcome measure	Macrolide intervention (n = 30)a		Placebo control (n = 30)b		Rate ratio (95% Cl)c	p value
	No.	Rate/1 child‐year	No.	Rate/1 child‐year
ER Visits	6	0.5	15	1.3	0.4 (0.1, 0.9)	0.047
Hospital admissions	5	0.4	6	0.5	0.8 (0.2, 2.8)	0.675
Hospital admission days	136	12.1	14	1.2	9.3 (5.5, 16.8)	< 0.001
Adverse events	2	0.2	4	0.4	0.5 (0.1, 2.9)	0.435
Unscheduled clinic visits due to respiratory symptoms	21	1.9	16	1.4	1.3 (0.6, 2.7)	0.473
Unscheduled clinic visits due to other symptoms	5	0.4	3	0.3	1.6 (0.4, 7.9)	0.505

Note: Admission days where estimates are from Poisson model due to lack of convergence with negative binomial.

^a There were 11.4 child‐years;

^b There were 11.3 child‐years;

^c Incident rate ratios and p values from negative binomial model adjusting for palivizumab and tracheostomy, except for outcome Hospital.

Safety outcomes were favorable for azithromycin therapy, with fewer adverse events observed in the treatment group (2 events; 0.2 per child‐year) compared to placebo (4 events; 0.4 per child‐year); IRR = 0.5 (0.1, 2.9), though this difference did not reach statistical significance (p = 0.435). The types of adverse events were primarily mild gastrointestinal symptoms that were resolved without intervention. No serious adverse events attributable to the study medication were reported, and no patients discontinued treatment due to side effects.

5. Discussion

5.1. Summary and Interpretation of Results

This RCT study aimed to evaluate the anti‐inflammatory effects of prophylactic azithromycin on healthcare utilization for children with BPD during the winter respiratory season. This study was based on our previous mouse study, where prophylactic macrolides decreased RSV infection severity in mice. Despite the theoretical benefits, our findings showed no significant reduction in the primary outcome of unscheduled healthcare‐related encounters with azithromycin. The IRR for the primary outcome was 2.1 (0.8, 5.4), indicating no substantial benefit, and Bayesian analysis indicated a low probability of benefit (11%). Although the study was originally powered to detect a difference in encounter rates with a sample size of 92, real‐world constraints limited enrollment to 60 participants. This reduced sample size introduces uncertainty in interpreting the results. Notably, exploratory subanalyses suggested that azithromycin may reduce emergency room visits and provide potential benefit among lower‐risk patients (those without tracheostomy and younger than 2 years of age). These findings should be interpreted with caution given the limited statistical power but provide important context for future research.

Several factors could explain these findings:

• 1.Influence of outliers: Two azithromycin‐treated children had prolonged hospital admissions (> 40 days), which heavily affected the results. Importantly, these extended stays were not attributable to medical severity. One infant remained hospitalized while awaiting foster placement after CPS intervention, and another stayed longer due to challenges in coordinating home care for tracheostomy and mechanical ventilation despite being clinically stable. Excluding these socially driven outliers in sensitivity analyses reversed the trend, suggesting a favorable effect (IRR = 0.6, 95% CI: 0.3–1.3) with a higher probability of benefit (86% by Bayesian analysis). In contrast to traditional frequentist analyses, which provide a p‐value reflecting how compatible the data are with a null hypothesis, Bayesian modeling allows us to directly estimate the probability that an intervention is beneficial. For example, a posterior probability of benefit of 86% means that, given the observed data and model assumptions, there is an 86% chance that azithromycin reduced healthcare encounters compared with placebo. This interpretation is more intuitive for clinicians, as it quantifies the likelihood of benefit rather than relying solely on statistical significance testing.
• 2.Heterogeneity of the Study Population: The study included a diverse group of children with varying degrees of BPD severity and different comorbidities [22]. This heterogeneity might have diluted the potential benefits of azithromycin. Furthermore, age may have played a role in the observed outcomes; older children (particularly those over 2 years of age) often experience a natural reduction in disease burden, as BPD tends to become less aggressive or even resolve with time. This could contribute to the lack of significant differences observed.
• 3.Adherence to Treatment: Although adherence was monitored, variations in adherence may have influenced outcomes. Future studies should adopt stronger adherence monitoring methods.
• 4.Small Sample Size or Lower Power: The initial sample size calculation indicated 96 patients were needed to detect statistically significant changes; however, we could recruit only 60 patients.

Our trial yielded complex results that require careful interpretation. Although the primary outcome did not meet the hypothesized benefit, important findings emerged, advancing our understanding of macrolide prophylaxis in this vulnerable population. Notably, the azithromycin group significantly reduced emergency room visits (IRR = 0.4, 95% CI: 0.1−0.9), representing an important clinical benefit given the burden and stress of ER visits for both patients and families. Chronic azithromycin may reduce respiratory illness severity, shown by fewer ER visits, without increasing clinic visits or hospital admissions. The azithromycin group had more hospital admission days, driven by two outliers with extended stays (> 40 days) for non‐medical reasons. When the two outliers were removed, there was no difference between the two groups, and chronic macrolides reduced ER visits without increasing clinic visits, hospital admission rates, or hospital days. Further investigation is warranted to understand the mechanisms underlying this observation.

5.2. Subgroup Analyses and Patient Selection

Subgroup analyses showed key differences in treatment response. Children with tracheostomies and those not receiving palivizumab had significantly longer hospital stays in the azithromycin group (the two outliers were part of this subgroup). Very high‐risk patients (> 2 years of age or have tracheostomy) showed a significant adverse effect from azithromycin treatment (IRR = 4.6, 95% CI: 1.1–20.2). However, our findings suggest that lower‐risk patients (no tracheostomy and ≤ 2 years of age) may potentially benefit from azithromycin prophylaxis (IRR = 0.6, 95% CI: 0.2−1.5), with a promising 82% probability of benefit according to Bayesian analysis. This aligns with Bacharier et al.'s findings, which demonstrated that early azithromycin use during respiratory tract infections in preschool children with recurrent severe illnesses reduced the risk of progression to severe lower respiratory tract illness compared to placebo [23]. These findings underscore the importance for careful patient selection and timing for intervention when considering prophylactic azithromycin.

5.3. Long‐Term Safety Considerations

The safety profile of long‐term azithromycin use presents several important considerations [23, 24]. Recent research has indicated that even short courses of macrolides can significantly affect both gut and airway microbiota [25]. Encouragingly, in our study, only approximately 10% of children receiving azithromycin experienced mild gastrointestinal symptoms, mainly loose stools, similar to the GI symptoms observed in the placebo group. These effects were generally transient and manageable, with no instances of severe antibiotic‐associated colitis reported. In terms of other systemic effects, no cases of ototoxicity or cardiac complications were observed in the study [25]. The absence of such complications may be attributed to the study protocol, which excluded children with pre‐existing sensorineural hearing issues and included thorough cardiac screening to reduce potential risks.

5.4. Clinical Implications

The results of this study suggest that winter‐long azithromycin prophylaxis should not be routinely implemented for all children with BPD. However, the significant reduction in emergency room visits and potential benefits seen in certain subgroups indicate that targeted use might be appropriate for specific patient populations.

The AMAZES trial demonstrated that the addition of azithromycin (500 mg three times weekly for 48 weeks) to standard treatment in adults with persistent uncontrolled asthma significantly reduced exacerbations and improved quality of life compared to placebo, with benefits observed in both eosinophilic and non‐eosinophilic asthma phenotypes [26]. Wong et al. reviewed the immunomodulatory mechanisms of macrolides in asthma, highlighting their effects on neutrophilic inflammation, viral infections, and airway remodeling, while emphasizing the need for better phenotyping to identify which asthma patients are most likely to benefit from macrolide therapy and to balance potential benefits against concerns about antimicrobial resistance [27].

Several strategies could optimize this approach, including limiting prophylaxis to peak viral seasons rather than year‐round use, carefully selecting lower‐risk patients based on risk factors and previous exacerbation patterns, focusing on children with wheezing who have an increased risk for asthma in the future, and conducting regular monitoring for adverse effects and resistance development.

5.5. Limitations

This study has several limitations that should be considered. It was conducted on a single site, which may limit the generalizability of the findings. Additionally, the relatively small sample size may have reduced the power to detect significant differences in some outcomes. The study included a diverse group of children with varying degrees of BPD severity and different comorbidities. This heterogeneity might have diluted the potential benefits of azithromycin. For example, patients older than 2 years of age might not have been included in the study, or patients with tracheostomy and home ventilator. Variations in adherence to the study medication could also have influenced the results. In addition, we did not have access to healthcare utilization data for some participants in the period before enrollment (October through December). It is therefore possible that some infants experienced encounters before study entry, which could in theory influence their subsequent risk profile during follow‐up. This limitation should be considered when interpreting the results.

5.6. Future Directions

Future research should focus on:

• Larger, multi‐center trials to validate these findings and better identify responsive subgroups. Essilfie et al. demonstrated that macrolide therapy can suppress inflammation in both steroid‐sensitive and steroid‐insensitive experimental asthma models, suggesting potential application in different BPD phenotypes that should be further explored [28].

• Alternative dosing strategies that might maintain benefits while minimizing resistance. Kanoh and Rubin highlighted the immunomodulatory effects of macrolides at doses below those required for antimicrobial activity, suggesting that optimized low‐dose regimens might provide anti‐inflammatory benefits while reducing resistance concerns [29].

• Biomarker development to better predict treatment response. Slater et al. showed that azithromycin therapy significantly impacts airway microbiota in asthma patients, suggesting that microbiome profiling could potentially identify patients most likely to benefit from macrolide therapy [29, 30].

• Further investigation into the significant reduction in emergency room visits observed in our study, which could represent an important clinical benefit even if other healthcare encounters are not reduced.

• Further investigation of the complex interactions between azithromycin, respiratory pathogens, and host immune responses in the context of BPD is needed to explain our paradoxical findings [29, 30].

6. Conclusion

Based on our findings, routine azithromycin prophylaxis for children with BPD cannot be recommended. However, the significant reduction in emergency room visits and potential benefits in lower‐risk populations are promising findings that warrant further investigation. Future research should identify specific subgroups that may benefit from azithromycin prophylaxis, such as younger children without tracheostomies, while optimizing dosing strategies to maximize benefits and minimize risks. Further research is needed to refine patient selection and treatment protocols before this approach can be considered for widespread implementation.

Author Contributions

Ricardo A. Mosquera: conceptualization, investigation, funding acquisition, writing – original draft, validation, supervision, resources, project administration. Aravind Yadav: writing – original draft, writing – review and editing. Maria Del Mar Romero‐Lopez: investigation, validation. Ivan G. Magana‐Ceballos: writing – review and editing, visualization, supervision, investigation. S. Shahrukh Hashmi: methodology, validation, visualization, writing – review and editing, formal analysis. Wilfredo De Jesus Rojas: investigation, formal analysis, data curation. Maria E. Tellez: investigation, writing – review and editing, formal analysis, data curation. Kaleigh Riggs‐Harpur: investigation, methodology, validation, visualization, formal analysis, data curation. Fatima M. Boricha: data curation, formal analysis, investigation. Tina S. Reddy: investigation, formal analysis, data curation. Janice L. John: conceptualization, writing – review and editing, investigation, supervision. Tomika S. Harris: formal analysis, data curation, supervision. Carlos E. Rodriguez‐Martinez: conceptualization, investigation, visualization, writing – original draft, formal analysis, data curation. Jefferson Buendia: investigation, writing – review and editing, data curation, formal analysis, methodology. Katrina E. McBeth: investigation, supervision, data curation. Cindy K. Jon: investigation, data curation, supervision. James M. Stark: investigation, writing – original draft, validation, visualization, writing – review and editing. Giuseppe N. Colasurdo: conceptualization, investigation, funding acquisition, writing – original draft, visualization, resources, supervision, formal analysis.

Ethics Statement

The protocol was approved by the Institutional Review Board of the University of Texas Health Science Center in Houston (approval number [IRB # HSC‐MS‐14‐0476). The trial was registered with ClinicalTrials.gov (Trial registration number: NCT02544984).

Consent

Prior to enrollment, written informed consent was obtained from the parent or legal guardian of each eligible child.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.