Table 1.

Key safety outcome measures for safety pregnancy studies

Outcome	Considerations
Preterm birth (PTB)	PTB is defined as delivery of a live‐born foetus prior to 37 weeks of gestation, complicates 15 million pregnancies each year and is associated with neonatal death and disability [11, 12, 13, 14]. Classification of PTB requires accurate estimation of gestational age. It is common in low‐ and middle‐income countries (LMICs) to estimate gestational age using methods that are subject to error and bias (e.g. last menstrual period, symphysis‐fundal height or newborn examination) [15, 16]. To accurately ascertain PTB, research protocols should include foetal ultrasound for dating, ideally in the first trimester.
Small for gestational age (SGA)	SGA is defined as sex‐specific weight‐for‐age at birth <10th centile [17, 18] and affects more than 20 million births per year in LMICs. SGA may result from intrauterine growth restriction or represent non‐pathologic variation. For this reason, the third centile of weight‐for‐age at birth may represent a preferable definition for research purposes, as it is more specific for pathology and associated with higher morbidity/mortality [19].
Foetal loss	Foetal loss before 20 weeks’ gestation is defined as spontaneous abortion; this endpoint is most relevant to surveillance. Stillbirth, defined as the in utero foetal death of a foetus after 20 weeks’ gestation by CDC [20] and 28 weeks by WHO [21], affects 2.6 million pregnancies annually [22]. Ultrasound dating is helpful to ensure correct classification of foetal loss.
Neonatal mortality	The causes of observed differences in neonatal mortality by antiretroviral regimen in some studies are unclear [23, 24]. Research studies should include neonatal mortality as an outcome, with the cause of death (and for high‐priority agents, power studies to detect at least moderate differences in neonatal mortality).
Congenital anomalies	While congenital anomalies related to medications in pregnancy are a concern, true teratogens are rare [25]. Approximately 3–6% of babies in the United States are born with a serious congenital anomaly [26]. Ascertainment and understating of the rate of congenital anomalies varies widely by country with a dearth of information in Africa [27]. Surveillance studies are necessary for understanding the associations of new agents with congenital anomalies [28]. Where congenital anomalies are reported, we recommend: (1) using a pre‐specified definition of anomaly [29]; (2) conducting systematic and standardized anomaly assessment [30] and (3) including expert adjudication of reported anomalies blinded to treatment.
Other outcomes	Low birthweight (LBW, <2500 g) is an easily ascertained birth outcome commonly used in research [31] but can be problematic because it conflates the pathologically distinct processes of preterm birth and small for gestational age. For that reason, PTB and SGA should ideally be considered separately in lieu of LBW. Several other important maternal and infant outcomes should be studied in the postpartum period (including during breastfeeding). While a thorough discussion of these measures is beyond the scope of this manuscript, we would highlight infant growth (ideally through 1 year of age) as a readily measured additional outcome that should be included as a secondary endpoint in RCTs. Summarizing growth may help to understand the possible longer term effects of study drug on prematurity [32].