Azithromycin in labour to reduce maternal and newborn sepsis and associated deaths: the need for a harmonized approach [version 1; peer review: 1 approved]

Maternal and newborn infections are a major contributor to mortality and morbidity globally. Lost-cost, effective and safe interventions are needed to address these. Based on promising findings, azithromycin has been identified as potentially effective antibiotic to reduce maternal and newborn infections in low-and middle-income countries (LMICs). However, robust randomized clinical trials in a range of settings are needed to confirm these findings as well as to understand the implications for antimicrobial resistance. To better understand the impact of azithromycin on maternal and newborn health, at least three clinical trials are being conducted to evaluate azithromycin in LMICs. We describe these trials, the importance of harmonizing study measures and the potential public health impact of azithromycin in LMICs.


Background and rationale
While under-5 childhood mortality has decreased over the last two decades, the proportion of neonatal deaths has increased during this period 1 . Infection during pregnancy and the puerperium account for approximately 15% of newborn deaths 3,4 . Maternal sepsis and infections represent about 10% of the global burden of maternal deaths 2 . In such context, there is a need for innovative, context specific interventions to reduce intrapartum sepsis in countries with the highest burden. However, to be sustainable, these interventions should be amenable to scale up in these settings to reduce the burden of both maternal and neonatal mortality.
Based on several studies, treatment with azithromycin, a lowcost antibiotic with immunomodulatory properties, has been identified as a novel approach to reduce maternal, neonatal and child mortality associated with infections [5][6][7] . Azithromycin can target organisms that may be very frequent pathogens but that historically have not been the target of antimicrobial preventive treatment 8 . Whether there are additional benefits beyond the antimicrobial effects remains to be determined.
During recent years, azithromycin has been delivered at pre-specified intervals through mass drug administration to African children under 5 years of age with the aim to reduce child mortality [9][10][11] . While these trials showed heterogenic results, the most apparent benefit has been observed among young infants [9][10][11] . Azithromycin has also been used in randomized clinical trials during pregnancy to reduce preterm birth 11,12 . One meta-analysis of 14 trials of azithromycin in pregnancy, primarily used in conjunction with other treatments, showed reduced risk of preterm birth and low-birth weight, but not significant reductions in other adverse perinatal outcomes 13 .
More recently, there has been a growing interest in using azithromycin prophylaxis during labour. One multicenter randomized clinical trial (RCT) of 500 mg azithromycin orally added to the standard prophylactic regimen for women who underwent cesarean delivery following labour or membrane rupture in the US showed that maternal infection was reduced by about 50% 7 . Use of this intervention was reported to be highly cost-saving 14 . A single center test-of-concept RCT in The Gambia including all women in labour found in a post hoc analysis that treatment with 2 g of intra-partum azithromycin reduced maternal infections by more than 50% and neonatal infections by more than 25% 16 .
These promising results have been tempered by several concerns of risks. One recurrent concern regarding the prophylactic use of azithromycin is the potential for developing macrolide resistance [17][18][19][20] . This is a particular consideration for widespread use of azithromycin in populations. In the MORDOR study from sub-Saharan Africa, antimicrobial resistance was associated with mass distribution of azithromycin to preschool children twice yearly for two years 21 , though levels returned to baseline in later studies. The individual-randomized trial from The Gambia found that macrolide resistance increases appeared to be transient, at least in populations with low baseline resistance 21 . These studies evaluated resistance through surveillance activities, and it is unclear this translates to resistant infections 22 .
A second important concern is the risk of adverse events, that may be rare, but with scale-up to large populations, could result in significant deaths. Azithromycin was associated with cardiac deaths in an observational study of an elderly population with medical comorbidities but not in a younger, healthy population similar to women of reproductive age 23 . Macrolides administered to neonates have been associated with infantile hypertrophic pyloric stenosis in some observational studies 24 , while were not reported in infants following a single dose of azithromycin to women in labour in randomized trials in The Gambia where daily active case detection was conducted by study nurses followed by weekly home visits for eight weeks 22 . A recent study administering azithromycin during the neonatal period to more than 10,000 neonates did not show an association between the use of azithromycin and infantile hypertrophic pyloric stenosis 23 .
Given the potential public health benefit, balanced with potential risks of population-based interventions in vulnerable populations, a thorough evaluation of the impact of intrapartum azithromycin prophylaxis is needed. Evidence for a new strategy is strengthened when the direction of the findings can be replicated in multiple studies conducted independently in different settings and population. However, when trials of the same intervention result in contradicting findings, policy makers, clinicians and others have a dilemma about how to translate the research findings into policy, delaying the implementation of potentially promising, life-saving interventions.
To explore optimal delivery strategies, dosing frequency, and target population for azithromycin administration to benefit pregnant women and their infants, a variety of approaches have been undertaken. Thus, while evaluating interventions in diverse settings is critical for generalizability, in order to facilitate the interpretation of results, prospective harmonization of data collection and methods is important [24][25][26] .
In the case of intrapartum azithromycin, to date there have been positive results but no definitive RCT from LMICs, where the intervention is most urgently needed. To address this concern, at least three independent groups have launched trials with harmonization of the critical metrics to evaluate the impact of azithromycin given during pregnancy and/or labour on maternal and newborn outcomes in different LMIC settings [27][28][29] . The BMGF-funded Sauver avec l'Azithromycine en Traitant les Femmes Enceintes et les Enfants (SANTE) trial uses a factorial design to randomize pregnant women in Mali to receive antenatal and intrapartum azithromycin or placebo, with their infants randomized to receive azithromycin or placebo at routine 6-and 14-week immunization visits 29 . ThePregnAnZI-2, SANTE and the Azithromycin in Labor Study (A-PLUS) include ancillary studies to address the effect of the intervention on macrolide resistance and the newborn's microbiome. These trials also include secondary objectives to determine additional potential benefits of the intervention in non-severe infection and infant's growth. The protocols for these trials are published, as part of an intention to harmonize methods, where possible, and to define the trial differences so that the findings ultimately can be better interpreted.

Methods
The trials are being conducted in The Gambia and Burkina Faso (the PregnAnZI-2 Trial), Mali (the SANTE Trial) and within the Global Network for Women's and Children's Health Research, a multi-site research network conducting research in 8 sites in 7 LMICs including Democratic Republic of Congo, Kenya, Zambia, Guatemala, Bangladesh, India and Pakistan (the A-PLUS Trial). Table 1 summarizes the population, intervention, comparator, and outcomes (PICO) features of these trials. In addition, Table 2 summarizes the secondary outcomes.

Discussion
There are several important commonalities of these trials that will facilitate meta-analyses to address gaps in the current research. First, the intrapartum intervention, including the dosage, was carefully considered. Given the success with a single oral dose of 2 g in The Gambian trial, an oral azithromycin dose of 2 g was selected for the proposed interventions. The PregnAnZI-2 and the A-PLUS trials administered a single 2 g dose of azithromycin during labour, while SANTE administered a single 2 g dose at an antenatal care visit in the second and/or third trimester and during labour or within 24 hours thereafter. The administration of the azithromycin within 24 hours after delivery is justified by the high concentration of azithromycin arriving to the newborns through the breast milk. All three trials utilized a double-blind, placebo-controlled RCT design, selected as the optimal trial methodology.
Second, the important outcomes were prospectively harmonized. Each trial is examining core outcomes, including mortality. The PregnAnZI-2 and the A-PLUS trials are also assessing maternal and newborn infections, and hospitalization rates. Therefore, an individual participant meta-analysis will be possible as the protocols including outcomes were harmonized as much as possible.
The primary neonatal outcome is intrapartum/neonatal death or sepsis within 28 days after birth. For the A-PLUS trial, the primary maternal outcome is maternal death or sepsis within 28 to 42 days after delivery and PregnAnZI-2 maternal sepsis. The primary aim of the antenatal and intrapartum dosing in SANTE is a composite outcome of stillbirths and infant mortality through 6-12 months. SANTE's co-primary aim will assess the efficacy of azithromycin for the prevention of infant deaths from the time of the first infant dose, administered alongside the first dose of pentavalent vaccine, through 6 to 12 months of age.
In addition, the sample size for each of the trials is substantial, allowing cumulatively, full evaluation of potentially rare adverse events. For example, concerns have been raised about potential adverse effects of intra and antepartum azithromycin on the pulmonary and gastrointestinal systems. Altogether, these trials will provide a comprehensive picture of the potential of short-term adverse events associated with azithromycin given during in labour.
Furthermore, several factors mitigate concerns about antibiotic resistance. The design of each of the trials use a single prophylactic intrapartum dose of antibiotic/placebo rather than multiple prophylactic dose administration. Both have also included women at both high-and low-risk for infection. The prophylactic use of azithromycin, if effective, will reduce the risk of infection and in turn reduce the overall use of antibiotics to treat infection as already shown in other trials 6,14 . To better assess this, all 3 trials are monitoring for resistant infection. In addition, the trials are conducting in-depth investigations that will inform on the impact of azithromycin use on potential changes to the microbiome.
Besides the similarities, there are several important differences in the trials, including secondary outcomes (Table 3) as well as among prospectively planned ancillary studies to assess important research questions. PregnAnZi-2 is assessing the impact of the intervention on growth. One of the main areas is the concern about AMR. To address this, the A-PLUS and SANTE trials have a sub-study to assess AMR of those randomized to azithromycin vs. placebo while the PregnAnZI-2 trial has several secondary endpoints to determine the extent of the impact on macrolide resistance in both mothers and babies.
There have been concerns about the impact of azithromycin on the microbiome. The PregnAnZI-2, A-PLUS and SANTE trials will conduct detailed microbiome assessments for a subset of participants (mothers and newborns). Future studies might correlate differences in microbiome due to antibiotics with longer-term outcomes of the child and mother.
As with other interventions administered during pregnancy, assessing the potential for long-term neurological effects on children is another public health concern. To help address this issue, the PregnAnZI-2 study will be following infants until 36 months after birth to assess neurodevelopmental outcomes. The A-PLUS trial will also evaluate neurodevelopmental outcomes among a subset of children at 24 months of age.

Conclusions
Azithromycin is an inexpensive macrolide that shows promise for improving health in LMICs. To date, numerous clinical trials have assessed the effect of azithromycin on maternal, newborn, and child survival as well as health for a range of conditions and using a range of outcomes. Since the puerperal period is the time with the highest risk of sepsis for women and their newborns, assessing the potential effect of this intervention during the time of highest risk is a necessity. With the public health concerns of increasing antimicrobial resistance globally, it is of special importance that these trials definitively address potential of antimicrobial resistance associated with azithromycin, and various approaches will be studied across the trials.
RCTs provide the strongest evidence needed to weigh the benefits and safety of interventions to improve health outcomes. Given the time and costs involved, prospective harmonization of trial outcomes is important whenever possible to allow for meaningful comparisons across trials and meta-analyses to evaluate outcomes among sub-groups. The three trials on azithromycin described here represent an opportunity to gain substantial, high-quality data across diverse populations to inform the use of azithromycin in pregnancy to reduce sepsis and other adverse pregnancy outcomes.

Data availability
No data are associated with this article.