Neurodevelopmental Pediatric Follow-Up After the Azithromycin Prevention in Labor Use Study

Intrapartum azithromycin was not associated with neurodevelopmental benefit at 2 years after birth asphyxia in low- and middle-income countries.

Abstract

OBJECTIVE:

To evaluate the association between 2 g of intrapartum azithromycin given to laboring mothers and neurodevelopmental outcomes after birth asphyxia.

METHODS:

This was a neurodevelopmental follow-up study of children born at 34 weeks of gestation or later who had concern for asphyxia and whose mothers were enrolled in A-PLUS (the Azithromycin Prevention in Labor Use Study). Mothers in A-PLUS were randomized to receive a single oral dose of azithromycin (2 g) or placebo during labor. This follow-up to A-PLUS spanned six sites across five countries (India [two sites], Pakistan, Zambia, Democratic Republic of Congo, and Guatemala). Asphyxia was defined as 5-minute Apgar score less than 7 or the need for bag-and-mask ventilation at birth. The primary outcome was the CCS (Cognitive Composite Score) of the BSID-III (Bayley Scales of Infant and Toddler Development, 3rd Edition) at a corrected age of 24±1 months. Secondary outcomes included the LCS (Language Composite Score) and MCS (Motor Composite Score) of the BSID-III at a corrected age of 24±1 months. Masked examiners administered the BSID-III and the ASQ-3 (Ages & Stages Questionnaires, 3^rd Edition). Outcomes were analyzed using a generalized linear model and were adjusted for site, gestational age at delivery, and maternal and child baseline characteristics that differed between treatment groups.

RESULTS:

Of 529 eligible mother–child dyads screened, 403 (197 in the azithromycin arm and 206 in the placebo arm) were enrolled and completed the neurodevelopmental follow-up at a corrected age of 24±1 months. The primary outcome, the CCS of the BSID-III, did not differ significantly between groups (azithromycin: 90.9±11.2 vs placebo: 90.9±11.7; mean difference: 0.29; 95% CI, −1.77 to 2.34). No significant differences were observed between treatment arms in the BSID-III’s LCS and MCS or in the total scores of the ASQ-3’s five domains. Subgroup analysis by region (sub-Saharan Africa vs South Asia) also showed no differences in BSID-III or ASQ-3 scores between the azithromycin and placebo groups.

CONCLUSION:

In this follow-up study, a single oral dose of azithromycin given to laboring mothers who delivered neonates with birth asphyxia did not improve neurodevelopmental outcomes at 2 years of age.

CLINICAL TRIAL REGISTRATION:

ClinicalTrials.gov, NCT03871491.

Birth asphyxia and hypoxic-ischemic encephalopathy (HIE) are major causes of long-term neurodevelopmental impairment, particularly in low- and middle-income countries (LMICs),¹ including among survivors of mild birth asphyxia.² Improved resuscitation and essential newborn care have increased neonatal survival^3,4; however, more children are now at risk for neurodevelopmental impairment, although the majority of the survivors have normal neurodevelopment.⁵ Although therapeutic hypothermia improves HIE outcomes in high-resource areas, its scarcity and linked mortality risks make it problematic for LMICs,⁶ where pharmacologic alternatives also lack evidence.⁷ In these regions, the synergy between birth asphyxia and untreated infections (eg, Ureaplasma) triggers inflammatory cascades that worsen brain injury. This creates an urgent need for dual-action anti-inflammatory and anti-infective neuroprotective strategies.

Azithromycin, a broad-spectrum macrolide (terminal half-life of 68 hours), is safe for use during pregnancy and presents a promising solution. A-PLUS (the Azithromycin Prevention in Labor Use Study), conducted in seven LMICs, showed that a single intrapartum oral dose of 2 g azithromycin significantly reduced maternal sepsis and death.^8,9 Beyond its antimicrobial effects, azithromycin exhibits potent anti-inflammatory properties. Preclinical models of neonatal¹⁰ and adult brain¹¹ injury demonstrate that it reduces injury severity and improves neuronal function. Azithromycin rapidly redistributes from the blood to tissues (including the brain), where it is cleared slowly. Because azithromycin has a long half-life and is transmitted through the placenta and breast milk, we hypothesized that the antimicrobial and anti-inflammatory properties have the potential to reduce brain injury from HIE-induced neurodevelopmental impairment.^10,11

Therefore, the ABC Study (Azithromycin for Brain Protection for Children), investigated the link between 2 g of maternal intrapartum azithromycin and neurodevelopmental outcomes after birth asphyxia. We hypothesized that among survivors of birth asphyxia (34 weeks of gestation or later) born to mothers who participated in A-PLUS, intrapartum azithromycin would increase the BSID-III (Bayley Scales of Infant and Toddler Development, 3rd Edition) CCS (Cognitive Composite Score) at a corrected age of 24±1 months when compared with placebo.

METHODS

This study was conducted in five countries in six Global Network for Women's and Children's Health Research of the Eunice Kennedy Shriver National Institute of Child Health and Human Development sites (India [two sites], Pakistan, the Democratic Republic of Congo, Zambia, and Guatemala) that participated in a masked follow-up study of children whose mothers had participated in A-PLUS, a placebo-controlled randomized trial of intrapartum azithromycin. The results of the primary A-PLUS⁸ and its methods have been published previously.¹² The IRB approved this follow-up study protocol at each site and at its partner institution in the United States.

This follow-up study included living neonates born at 34 weeks of gestation or later who had birth asphyxia and whose mothers were randomized in A-PLUS to receive intrapartum azithromycin (2 g) or placebo. Asphyxia was defined as the failure to initiate and sustain breathing at birth (operationalized as bag-and-mask ventilation or 5-minute Apgar score less than 7). Exclusion criteria included known or suspected genetic or neurologic conditions that affect neurodevelopment. A full list of exclusion criteria is shown in Figure 1. Based on the original A-PLUS randomization, this masked follow-up study was employed to compare neurodevelopmental outcomes at a corrected age of 24±1 months.

Fig. 1. CONSORT (Consolidated Standards of Reporting Trials) diagram. *Participants were eligible if they had a 5-minute Apgar score less than 7 or if they received bag-and-mask ventilation or both. A-PLUS, Azithromycin Prevention in Labor Use Study; ABC, Azithromycin for Brain Protection for Children; GA, gestational age; BSID-III, Bayley Scales of Infant and Toddler Development, 3rd Edition.

Ramani. Intrapartum Azithromycin and Neurodevelopmental Outcomes. Obstet Gynecol 2026.

The primary outcome of this study was the BSID-III CCS at a corrected age of 24±1 months. Secondary outcomes included the LCS (Language Composite Score) and MCS (Motor Composite Score) of the BSID-III at a corrected age of 24±1 months.¹³ Language staff, research personnel, and developmental assessors were masked to the treatment allocation of azithromycin or placebo. The BSID-III assessments were performed by trained and certified examiners in dedicated, private, and distraction-free spaces.

The BSID-III was chosen because the sites already had assessors trained in it as part of the Global Network for Women's and Children's Health Research site capacity building.¹⁴ The neurodevelopment oversight committee held an intensive three day in-person workshop and one virtual training. At least one lead examiner was trained and certified at each site by using a review of a video of their administration with a nonstudy participant by the gold-standard Bayley trainer. Each site’s lead examiner then reviewed all other examiners’ administrations at their respective sites before they submitted a video for certification to the gold-standard Bayley trainer. The submissions continued until all individual examiners at each site reached the required minimum reliability of 85% for administration and scoring in each domain. Corrective feedback was provided on all videos. Sites had two to five certified examiners; the number varied based on the site’s size.

The ASQ-3 (Ages & Stages Questionnaires, 3rd Edition), a validated parent-report screening questionnaire for neurodevelopment, was administered to parents by research staff during the 24±1-months-corrected-age visit. The ASQ-3 has six items for five domains—Communication, Gross Motor, Fine Motor, Problem Solving, and Personal Social—resulting in a 30-item questionnaire. Domain scores are totaled and compared with cut scores to determine risk categories.

In addition to using the BSID-III and ASQ-3 measurements, we also collected information regarding children's learning environment by using the Family Care Indicators questionnaire ¹⁵ and collected information regarding growth by using anthropometric measurements. The Family Care Indicators questionnaire evaluated the environmental enrichment of the children through parental interaction and play, as measured by the number of books read to the children and the variety of play materials and sources used. The following information was collected for all mother–child dyads who participated in A-PLUS: gestational age (calculated following the American College of Obstetrics & Gynecology’s hierarchy of sources)¹⁶ using ultrasound dating when available, followed by last menstrual period and clinical examination; birth weight; height; mode of delivery; and neonatal resuscitation status.

The sample size for this study was selected assuming a difference in BSID-III CCS of 5 points (±0.33 SD) at the corrected age of 24±1 months.¹⁷ To detect such a difference between treatment arms with at least 90% power, a sample size of 190 children per treatment arm (total of 380 infants) was required based on a two-sided, two-sample, equal-variance t test with a type I error rate of 0.05.

Maternal and child characteristics at enrollment and at the 24-month visit were summarized in each treatment arm and by region using frequencies and percentages for categorical variables and means and SDs, or medians and IQRs, for continuous variables. The mean and SD of each outcome at the 24-month visit were also summarized in each treatment arm.

The adjusted mean difference between treatment arms (along with 95% CI and P-value) for each outcome was determined using a generalized linear model. The model adjusted for the fixed effects of the treatment arm, study site, and gestational age, as well as prespecified baseline characteristics that differed significantly between groups (such as parity, maternal infections, maternal conditions, preterm labor, 5-minute Apgar score less than 7, and bag-and-mask ventilation). We also tested for the regional modification of the treatment effect by including a 2-way interaction between the treatment arm and site and performed linear hypothesis tests using contrasts to compare outcomes within sub-Saharan African and South Asian sites separately. Missing data were assumed to be missing at random, and a tipping point analysis confirmed the robustness of the results to this assumption (Appendix 2, available online at http://links.lww.com/AOG/E656). All analyses were performed using a 2-sided significant level of 0.05, and no adjustment was made for multiple comparisons across outcomes or subgroups, because this was not a confirmatory study. All analyses were performed using SAS 9.4.

RESULTS

Of the 703 children from A-PLUS who were eligible for the ABC Study, 191 died before follow-up, without a difference between treatment arms (26.7% for treatment vs 27.6% for placebo). A total of 529 children who had survived birth asphyxia were screened between August 2022 and October 2024 for the ABC Study (Fig. 1). Of these, 450 children were enrolled, 409 completed the 24-month visit, and 403 completed the primary outcome (197 children in the azithromycin arm, and 206 children in the placebo arm).

Maternal and child characteristics at enrollment and 24 months did not differ between treatment arms (Table 1). Of the children who completed the 24-month visit, 56.5% were male and 42.8% had stunting (defined as a length-for-age z-score of less than −2). They had a mean weight z-score of −1.1 (SD=2.0) and a mean head circumference z-score of −0.7 (SD=2.4), had mothers with varying degrees of education (24.2% no formal schooling, 15.9% 1–6 years of schooling, 50.1% 7–12 years of schooling, and 9.8% at least 13 years of schooling). In addition, for the children who completed the 24-month visit, the mean number of books was 0.8 (SD 1.2), the variety of play materials composite score was 2.6 (SD 0.6), the sources of play materials composite score was 5.6 (SD 2.0), and the play activities composite score was 6.0 (SD 2.0); these domains assessed environmental enrichment through parental reading, interaction, and playing with children.

Table 1.

Maternal and Child Characteristics at Enrollment and Follow-Up by Treatment Arm

Appendix 3 (available online at http://links.lww.com/AOG/E656) provides a descriptive table of maternal and child characteristics at enrollment for neonates included in A-PLUS and for children included in the ABC Study. Compared with infants included in A-PLUS, a higher percentage of infants in Latin America were included in the ABC study. Appendix 4 (available online at http://links.lww.com/AOG/E656) provides the same descriptive information broken down by Africa and Asia.

The adjusted mean of the BSID-III CCS did not differ between treatment arms (mean±SE; azithromycin 90.9±11.2 vs placebo 90.9±11.7; with adjusted mean difference of 0.29; 95% CI −1.77 to 2.34; P =.78) (Table 2). Neither the BSID-III LCS (azithromycin 96.6±15.2 vs placebo 96.6±16.5; P =.91) nor MCS (azithromycin 99.0±12.5 vs placebo 98.7±14.4; P =.72) differed between the arms (Table 2). There were also no differences between treatment arms in the adjusted mean of the ASQ-3 and its five domains (Table 2). The growth parameters at the 24-month follow-up did not differ between the treatment arms (Table 1). Further contrast testing indicated that there were no differences in the BSID-III or ASQ-3 scores between treatment arms in the sub-Saharan African and South Asian sites (Table 3).

Table 2.

Bayley Scales of Infant and Toddler Development, 3rd Edition, and Ages & Stages Questionnaires, 3rd Edition, Composite Scores at Follow-Up by Treatment Arm

Table 3.

Bayley Scales of Infant and Toddler Development, 3rd Edition, and Ages & Stages Questionnaires, 3^rd Edition, Composite Scores at Follow-Up by Treatment Arm and Region

DISCUSSION

This study found that neurodevelopmental outcomes at 24 months, as assessed with the BSID-III and ASQ, did not differ between children whose mothers received intrapartum azithromycin and those in the placebo group. Additionally, growth parameters did not differ. These results indicate that intrapartum azithromycin was not associated with improvement in HIE-related outcomes, nor was it associated with adverse effects on the child’s growth or development.

Hypoxic-ischemic encephalopathy is a hypoxic reperfusion injury of the brain that leads to a cascade of molecular events such as energy crisis, lactic acidosis, excitotoxicity, and free radical injury, potentially resulting in long-term neurologic damage.¹⁸ Therapeutic hypothermia, the standard of care for HIE, has limited efficacy and may increase mortality in low-resource settings. This has spurred the exploration of alternative anti-inflammatory strategies. Although previous studies have investigated such agents primarily as adjuncts to therapeutic hypothermia,¹⁹ our study assessed intrapartum azithromycin as a standalone therapy for HIE.

Preclinical models of hypoxic ischemic and traumatic brain injuries demonstrate that azithromycin improves sensorimotor and cognitive function and reduces hippocampal injury²⁰ while reducing oxidative stress and neuroinflammation.²¹ In neonatal rat models of HIE, azithromycin reduced brain injury even when inflammation was exacerbated.²² These neuroprotective effects are primarily attributed to its anti-inflammatory properties, which mitigate the cascading processes induced by HIE.

Despite promising preclinical data, this study found no association between intrapartum azithromycin and improved neurodevelopmental outcomes in children with birth asphyxia. The primary outcome (BSID-III CCS) and all secondary outcomes showed no significant differences between the azithromycin and placebo groups. These findings suggest that azithromycin, as administered in A-PLUS, did not provide neuroprotection for this cohort. Several factors may account for this lack of efficacy. First, the pathophysiology of birth asphyxia involves complex mechanisms such as hypoxia, ischemia, and subsequent reperfusion injury, which may not be adequately addressed by azithromycin alone.²³ Second, the intrapartum timing and maternal dosing may not have been optimal for neuroprotection. Additionally, the amount of azithromycin transferred through the placenta and breast milk might have been insufficient to mitigate the asphyxia-induced inflammatory cascade. Finally, this study used the World Health Organization definition of birth asphyxia, which does not meet the criteria used in clinical trials of HIE. Several factors may explain the discrepancy between our findings and those of preclinical studies. Preclinical studies often achieve success by administering azithromycin postinsult during the secondary injury phase. In contrast, our study provided intrapartum dosing before the onset of asphyxia, which may not adequately target the inflammatory cascade, oxidative stress, and reperfusion injury that unfold after the hypoxic insult.^23,24 Although azithromycin crosses the placenta and enters breast milk with a long half-life, intrapartum dosing may not achieve the therapeutic concentrations in the neonatal brain necessary to suppress asphyxia-induced inflammation. Hence, a single maternal dose administered before the injury cascade begins may be insufficient to target the critical postinjury mechanisms that drive ongoing neuronal damage, compared with direct postnatal administration. In this study population, the risk of neurodevelopmental impairment is often magnified by the synergistic effect of co-occurring conditions, particularly high rates of maternal and neonatal infection. The study's dosing and timing may not have been adequate to address this compounded risk. Additionally, precise timing and dosage control in preclinical models often fail to translate directly into human clinical settings, highlighting the need for further research to optimize delivery and maximize neuroprotective potential.

Growth parameters, including weight, height, and head circumference, did not differ between treatment arms. This lack of difference supports the safety profile of intrapartum azithromycin, a reassuring finding for LMIC settings where the drug is frequently used to reduce the high burden of maternal infections.^8,9 Although the sample size was insufficient to detect rare safety events, the observed profile supports continued use of azithromycin for its primary indications without concern for adverse effects on child growth.

Several limitations of this study should be acknowledged. First, the definition of birth asphyxia was based on clinical World Health Organization criteria²⁵ rather than the biochemical markers (eg, cord pH, base deficit) or specific prenatal histories typically used in high-resource settings. This may have introduced population variability. Second, the retrospective nature of the study precluded categorizing children by HIE severity using neonatal neurologic examinations (eg, Sarnat staging).²⁶ This is important, especially because the study cohort's language and motor scores were near population means, suggesting a sample predominantly composed of mild HIE cases. Although mean cognitive scores were lower (approximately two-thirds of an SD below the population mean), this may reflect the socioeconomic disadvantages within the LMIC cohort, including low maternal education levels, that are known to affect developmental outcomes in higher-income countries.²⁷ Alternatively, these lower scores might indicate that BSID-III norms, derived from U.S. populations, are less culturally or contextually appropriate for LMIC children, despite the tool's validated use in similar research settings.²⁸

Also, we did not measure nutritional intake. Given the diverse cultures and dietary traditions across the study sites, creating a standardized, meaningful dietary log for all children was not feasible, even though nutrition remains a critical factor in neurodevelopment. Although the potential effect of nutrition on the neurodevelopment of our study population cannot be ruled out, the absence of significant differences in growth parameters between the treatment arms is reassuring. Although this study found no neuroprotective benefit from intrapartum azithromycin, postnatal administration remains a theoretical alternative. However, given that maternal dosing failed despite placental and lactational transfer, postnatal treatment may also prove ineffective. Future research could focus on identifying specific subgroups, such as those with concomitant infections or varying HIE severity, who might still benefit from azithromycin.

This study indicates that intrapartum administration of azithromycin is not associated with improved neurodevelopmental outcomes in children with birth asphyxia. Although azithromycin is effective for preventing maternal infections, this study does not support its effectiveness as a neuroprotective agent for children in this context. Importantly, it was not associated with adverse developmental or growth effects, confirming its fetal safety as administered in the trial. Given the substantial burden of neurodevelopmental disabilities in LMICs, identifying effective interventions remains an urgent priority.

Authors' Data Sharing Statement

• Individual deidentified participant data and data dictionaries are available.

• Data that were captured on case report forms are available.

• Data are available indefinitely.

• Other documentation includes the study protocol, case report forms, de-identification process, and data dictionaries.

• Data are available through the NICHD Data and Specimen Hub (DASH) at https://dash.nichd.nih.gov/study/426434. Anyone who wishes to access the data for any purpose must submit a request through NICHD DASH. All restrictions or limitations on data use will be specified in the request.