Adverse birth outcomes and their clinical phenotypes in an urban Zambian cohort

Background: Few cohort studies of pregnancy in sub-Saharan Africa use rigorous gestational age dating and clinical phenotyping. As a result, incidence and risk factors of adverse birth outcomes are inadequately characterized. Methods: The Zambian Preterm Birth Prevention Study (ZAPPS) is a prospective observational cohort established to investigate adverse birth outcomes at a referral hospital in urban Lusaka. This report describes ZAPPS phase I, enrolled August 2015 to September 2017. Women were followed through pregnancy and 42 days postpartum. At delivery, study staff assessed neonatal vital status, birthweight, and sex, and assigned a delivery phenotype. Primary outcomes were: (1) preterm birth (PTB; delivery <37 weeks), (2) small-for-gestational-age (SGA; <10 th percentile weight-for-age at birth), and (3) stillbirth (SB; delivery of an infant without signs of life). Results: ZAPPS phase I enrolled 1450 women with median age 27 years (IQR 23–32). Most participants (68%) were multiparous, of whom 41% reported a prior PTB and 14% reported a prior stillbirth. Twins were present in 3% of pregnancies, 3% of women had short cervix (<25mm), 24% of women were HIV seropositive, and 5% were syphilis seropositive. Of 1216 (84%) retained at delivery, 15% were preterm, 18% small-for-gestational-age, and 4% stillborn. PTB risk was higher with prior PTB (aRR 1.88; 95%CI 1.32–2.68), short cervix (aRR 2.62; 95%CI 1.68–4.09), twins (aRR 5.22; 95%CI 3.67–7.43), and antenatal hypertension (aRR 2.04; 95%CI 1.43–2.91). SGA risk was higher with twins (aRR 2.75; 95%CI 1.81–4.18) and antenatal hypertension (aRR 1.62; 95%CI 1.16–2.26). SB risk was higher with short cervix (aRR 6.42; 95%CI 2.56–16.1). Conclusio ns: This study confirms high rates of PTB, SGA, and SB among pregnant women in Lusaka, Zambia. Accurate gestational age dating and careful ascertainment of delivery data are critical to understanding the scope of adverse birth outcomes in low-resource settings.


Introduction
The often overlapping outcomes of preterm birth (PTB), small for gestational age (SGA), and stillbirth (SB), collectively called 'adverse birth outcomes', are responsible for most perinatal morbidity and mortality worldwide. 1,2 Low-and middleincome countries bear the overwhelming burden of global PTB, SB, and SGA. [3][4][5] However, reliable classification and estimation of adverse birth outcomes in low-resources settings is challenging because of a number of interrelated factors, including (1) uncertain gestational age dating, 6 (2) conflation of fetal growth restriction and PTB into the less useful metric of 'low birthweight', 3,5 (3) inconsistent thresholds for fetal viability, 7 and (4) misclassification of stillbirth and neonatal death. 8,9 In many countries, including Zambia, data sources that adequately address these methodological challenges are lacking to the extent that national estimates of adverse birth outcomes must be modeled. 3,6,10,11 In sub-Saharan Africa, cohort studies in pregnancy rarely use reliable gestational age dating or clinical phenotyping to classify outcomes. Deliberate clinical phenotyping that characterizes the events that incite parturition (i.e., spontaneous vs. provider-initiated), quantifies maternal and fetal co-morbid conditions, and reliably distinguishes the timing of perinatal death is essential for rigorous classification of adverse birth outcomes. 12,13 Accurate estimation of gestational age with fetal ultrasound is also critical. Other dating methods, such as maternal recall of last menstrual period (LMP), 14-17 symphisialfundal height measurement, 18 or newborn physical exam 19,20 introduce error (and in some cases, bias 21 ).
We established a cohort of 1450 pregnant women and their infants at a tertiary care institution in Lusaka, Zambia, with the goal of better understanding the epidemiological factors and biological mechanisms leading to adverse birth outcomes. This report presents the outcomes of the first phase of this cohort.

Methods
The Zambian Preterm Birth Prevention Study (ZAPPS) is an ongoing prospective observational cohort study at the Women and Newborn Hospital of the University Teaching Hospitals (UTH-WNH), the primary referral hospital in Lusaka. Phase 1 of ZAPPS, the subject of this report, recruited and enrolled participants beginning in August 2015 and completed follow-up in June 2018. The sample size for this observational study was initially set at 2000 women, with a target of 250 preterm birth events based on published regional population estimates. 4 For budgetary reasons and because the prematurity rate was higher than initially expected, enrollment was stopped in September 2017 after 1450 women had been enrolled. The ZAPPS protocol was developed to align with the Guidelines for Strengthening The Reporting of Observational Studies in Epidemiology (STROBE). 22 Study population Pregnant women meeting the following criteria were eligible for enrollment in Phase 1 of the ZAPPS cohort: (1) 18 years of age or older; (2) viable intrauterine singleton or twin gestation; (3) presentation to antenatal care prior to 20 weeks of gestation if HIV-uninfected or 24 weeks if HIV-infected; (4) residing within Lusaka with no plans to relocate during the study follow-up period; (4) willing to provide written, informed consent; (5) willing to allow participation of their infant(s) in the study; (6) willing to be contacted and followed up at home if necessary.
The ZAPPS protocol was approved prior to study initiation and is subjected to annual review by the University of Zambia School of Medicine Biomedical Research Ethics Committee (reference number: 016-04-14) and the University of North Carolina School of Medicine Institutional Review Board (study number: 14-2113). The study also received approval from the Zambian Ministry of Health National Health Research Authority. Each participant provided written informed consent before enrollment.

Procedures
Full study procedures are described in detail elsewhere. 23 Community educators identified potential participants at antenatal care clinics of UTH-WNH and five surrounding clinics in Lusaka, assessing basic eligibility criteria such as age and approximate gestational age. Interested volunteers underwent ultrasound examination per standard of care to determine pregnancy location, fetal viability, number of fetuses, and gestational age by standard biometry (Sonosite M-Turbo, Fuji Sonosite, Bothell, WA). Gestational age was calculated at enrollment by crownrump length if <14 gestational weeks or by head circumference and femur length if ≥14 weeks. Fetal biometry structures were each measured twice and then averaged to calculate gestational age using INTERGROWTH-21st equations. 24,25 Pregnancies below the lower threshold for INTERGROWTH-21st equations were dated by the Hadlock formula. 26 Interested women who met preliminary ultrasound eligibility criteria completed an informed consent process in their preferred language of English, Nyanja, or Bemba.
At enrollment, study nurses collected demographic and behavioral information through medical record review and participant interview, and documented a thorough health history including prior pregnancy outcomes. As part of standard antenatal care, participants underwent a physical exam and rapid testing for hemoglobin, urinalysis, syphilis (SD Bioline Syphilis 3.0, Abbott Diagnostics), and HIV (SD Bioline 3.0, Abbott Diagnostics).
After enrollment, participants received routine antenatal care at follow-up visits scheduled at approximately 24 weeks, 32 weeks,

Amendments from Version 1
In this version, we have expanded the discussion to address the reviewers' concerns around definitions of preterm birth and stillbirth used as well as to add global context to our estimates of the frequencies of adverse birth outcomes. We have updated Figure 2 to illustrate the co-occurrence of very preterm birth, very small-for-gestational-age, and stillbirth. We present an additional linear regression analysis of continuous exposures and continuous outcomes and note that maternal height is not associated with either preterm birth or small-for-gestational-age.
Any further responses from the reviewers can be found at the end of the article REVISED and 36 weeks. All participants underwent cervical length measurement in the second trimester (i.e., 14-28 weeks) and fetal growth assessment by biometry in the third trimester. 27,28 Cervical length measurements were performed by sonographers with certification in the Cervical Length Education and Review (CLEAR) program. Study nurses staffed the UTH-WNH labor ward full-time and collected detailed information about the clinical course and perinatal outcomes of participants and their infants, including gestational age at birth, neonatal vital status, birthweight, and sex, and assigned a delivery phenotype. For participants who did not deliver at UTH-WNH or were not captured by study staff during their delivery admission, study staff collected perinatal outcomes either in person or by phone. Cohort retention in this analysis was defined as ascertainment of date of delivery.

Outcomes
Primary outcomes of this analysis were: PTB, defined as birth between 16 0 / 7 and 36 6 / 7 gestational weeks, SGA (newborn weight-for-age <10th percentile by INTERGROWTH-21st norms), 29 and SB (delivery of an infant without signs of life ≥16 0 / 7 weeks). Secondary outcomes included very PTB (birth before 34 0 / 7 weeks), very SGA (newborn weight-for-age <3rd percentile), gestational duration (days), and birthweight centile for gestational age at delivery. Both PTB and very PTB were further characterized as either spontaneous (spontaneous labor or membrane rupture prior to labor) or provider-initiated (induction of labor or pre-labor cesarean). We differentiated antepartum stillbirth (i.e., fetal heart tones absent on admission or, if not assessed, maceration skin changes present at delivery) from intrapartum stillbirth (i.e., fetal heart tones present on admission or, if not assessed, absence of maceration skin changes at delivery).

Statistical analysis
We performed descriptive analyses of baseline characteristics and exposures of the cohort, reporting median and interquartile range (IQR) for continuous variables, and frequency and percent for categorical variables. Differences in baseline characteristics between women retained at delivery and those lost to follow-up were evaluated by univariate tests of association.
We summarized parturition phenotype among our retained participants by preterm versus term and spontaneous versus provider-initiated following a standard rubric. 12 Among spontaneous PTB, we identified primary maternal, fetal, and/or placental conditions present at the time of delivery. Among provider-initiated PTB, we reported the primary indication for delivery as recorded by the provider. Finally, key individual conditions present and phenotypic clusters were used to classify all PTB, spontaneous PTB, and provider-initiated PTB.
We calculated the incidence of adverse birth outcomes: PTB, spontaneous PTB, very PTB, spontaneous very PTB, SGA, very SGA, and SB among all participants retained at delivery. Twin deliveries in which at least one neonate was SGA or stillborn were classified as having met the respective outcome. Crude associations between key exposures and outcomes were analyzed as risk ratios estimated using Poisson regression analyses with a robust variance. 30 Adjusted risk ratios were also estimated using Poisson regression accounting for other key exposures plus maternal age, BMI, estimated gestational age at enrollment, and HIV serostatus at enrollment. We then analyzed the association between continuous exposure variables and outcomes (gestational duration and birthweight centile by gestational age at birth) using linear regression.
Kaplan-Meier curves were plotted for time to delivery for participants with and without a history of prior PTB, short cervix, and twin gestation. We accounted for loss to follow-up by right censoring women at their last study visit (if before delivery) and compared survival between exposure groups by log-rank tests. We also used Cox regression to calculate the hazard of delivery between participants with and without prior PTB, short cervix, and twin gestation, adjusting for maternal age. The proportional-hazards assumption was tested based on Schoenfeld residuals. 31 Because of the inherent converging of survival curves in pregnancy at term, we restricted our models to the preterm period by administratively censoring all participants at 37 gestational weeks. 32,33 All statistical analyses were performed with Stata version 14 (College Station, TX, USA) and SAS version 9.4 (Cary, NC, USA).
Maternal age ≥35, prior PTB, short cervix, twin gestation, antenatal hypertension, and EGA at enrollment <14 weeks were associated with PTB (Table 3). Overall, these associations were stable or strengthened when restricting the outcome to spontaneous PTB and to very PTB (  (Table 6).
Elevated risks of PTB among women with prior PTB, short cervix, and twin gestation were supported by survival analyses, with log-rank tests of association demonstrating significant differences between groups of each variable ( Figure 5; Table 5). In proportional hazards models adjusted for maternal age at enrollment, participants with prior PTB, short cervix, and twin gestation had significantly higher hazards of delivering before 37 gestational weeks compared to parous women with no prior PTB, women with cervical lengths ≥25mm, or with single gestations. Participants with increasing numbers of prior preterm births demonstrated increasing hazard ratios of delivering preterm.

Discussion
We present the primary results of the ZAPPS pregnancy cohort, established to evaluate the risk factors associated with adverse birth outcomes in Lusaka, Zambia. This study was notable for enrollment of pregnant women at early presentation to antenatal care, gestational age determination by early ultrasound, universal cervical length screening, comprehensive and uniform antenatal and postpartum care, and clinical phenotyping of birth outcomes. Our analyses revealed strong risks of prior preterm birth, short mid-trimester cervical length, and twin gestation on incident preterm birth, and these risks were supported by analyses of pregnancy 'survival' to term. We also report increased risks of small-for-gestational-age infants among nulliparous women and women with twin gestation, and of stillbirth among those with short cervix.
The proportion of gravidas who deliver before term varies significantly across individual studies and national estimates in sub-Saharan Africa. The most recent global report estimated a PTB rate of 8% in Europe and 11% in North America, compared to 12% across sub-Saharan Africa and 12% in Zambia specifically, where the rate was based on modeled regional estimates instead of national data. 6 In contrast, a census accounting of 237,219 public sector births over 6 years in Lusaka -where the vast majority of pregnancies are dated by last menstrual period -classified 46% of singleton deliveries as preterm. 37,38 In Zambia, obstetrical ultrasound is rare and the reliance on maternal recall of LMP alone substantially over-estimates preterm birth rates, 17,39 an inaccuracy that worsens with later presentation  to care. 21 We report PTB based on ultrasound gestational age dating and prospectively ascertained delivery outcomes such that our data are likely more accurate than reports that rely on LMP recall or regional models.
Inconsistent global PTB definitions hinder interregional comparisons of rates and risk factors, and we acknowledge that gestational age boundaries are somewhat arbitrary. We chose a lower gestational age limit of 16 weeks because of evidence of similar etiological risk factors between preterm births occurring as early as 16 weeks and those that occur later in pregnancy 12,40 . In addition, we included preterm stillbirths in our definition of PTB since excluding stillbirths that occur in the process of parturition would falsely lower the rate of PTB. In a sensitivity analysis that excluded all stillbirths from PTB outcomes, the PTB incidence was modestly reduced. However, risk estimates calculated in regression models remained stable, which supports evidence that risk factors for live and stillborn PTB demonstrate substantial overlap 40 .
The distinction of preterm parturition as spontaneously occurring versus provider-initiated is important but rarely reported from national surveillance or clinical research data in lowresource settings. Deliveries that are preceded by spontaneous labor or membrane rupture are phenotypically distinct from those that are induced medically or surgically for complications such as preeclampsia, antepartum fetal demise, or other maternal or fetal conditions. 12,41-44 Further classification based on primary conditions present in spontaneous PTB and the primary indications for provider-initiated PTB is based on a standardized rubric proposed to elucidate phenotypic clusters of PTB 12 . An understanding of prevailing phenotypes can direct research, policy, and preventive interventions towards regional and populationspecific needs. 13,45,46 While our cohort is limited by a small number of PTB events (n=181), we were able to classify nearly all (i.e., 97%) as either spontaneous or provider-initiated, and to identify the primary complications and phenotypic characteristics of each. Further granularity and generalizability of PTB classification requires a larger sample size, signaling a need for future high-quality obstetrical research on a greater scale.
As with PTB classification, identifying infants born SGA requires accurate gestational age estimation, which can be at best imprecise, and at worst biased, when based solely on  Defined as 1+ leukocyte esterase and/or + nitrites.

Figure 4. Predicted probability of preterm birth <37 weeks by mid-trimester cervical length.
Among ZAPPS cohort participants with a cervical length measured by ultrasound in the second trimester (n=1081), the probability of preterm birth <37 weeks decreased with increasing cervical length. PTB, preterm birth; RR, relative risk; CI, confidence interval.
LMP. 21 The incidence of SGA in our cohort (18%) was modestly higher than regional estimates of SGA in sub-Saharan Africa of 16%, 5 and compared to a recent estimate in Zambia of 13%, which was modeled from published rates in other neighboring countries because of scarcity of data from Zambia itself. 5 In comparison to the ZAPPS cohort, in which older maternal age, low BMI, nulliparity, twin gestation, and antenatal hypertension predicted either SGA or very SGA, a study among over 19,000 singletons in Tanzania identified younger maternal age, height, and nulliparity as strong risk factors for SGA. 47 The WHO Multi-country Survey on Maternal and Newborn Health found nulliparity and hypertensive disorders to indicate higher risk of preterm SGA and hypertensive disorders, sociodemographic factors, and anemia to predict term SGA. 48 With a much smaller sample size and fewer outcomes compared to these two studies, we were not able to differentiate our outcome by preterm vs. term SGA due to low statistical precision for stratified associations with key risk factors. Due to this low precision, we are not able to discern whether or not SGA outcomes were modified by gestational age at delivery. However, both of these studies relied on reported LMP to estimate gestational age at delivery, which itself may have introduced error. Whether growth restriction is a distinct pathological process before 37 weeks compared to after 37 weeks is unclear. Finally, while the INTERGROWTH-21 st Project 49 intended to define universal fetal growth and newborn weight standards derived from an extensive multi-ethnic sample of women with adequate antenatal care and nutrition, its widespread use over ethnicity-specific or customized standards has been disputed. 43,[50][51][52][53][54][55][56] Despite this, we chose to define SGA in our cohort based on INTERGROWTH-21 st standards since local standards that include all pregnancies affected by undernutrition and/or pregnancy comorbidities tend to identify only the severest 10% of cases by definition.
Stillbirth, a composite outcome comprising antepartum and intrapartum fetal death, is particularly understudied in lowresource settings. Compared to developed regions with a stillbirth rate of 3.4 per 1000 total births, the rate in sub-Saharan Africa is estimated as 28.7 per 1000, while 4% of our cohort delivered stillbirths. 3 The true global burden of stillbirth and its underlying causes are poorly classified due to inconsistent fetal viability limits and imperfect classification of neonatal death versus stillbirth, limited resources for case investigations, under-reporting of home births that result in perinatal death, and inadequate national and regional reporting of identified cases. 3 Indeed, recent global and regional estimates of stillbirth included just 17% of its datapoints from sub-Saharan Africa and south Asia, regions that bear 77% of the global burden. 3 Data from the recent Zambia Demographic and Health Survey reported a rate of stillbirth, defined as fetal death over 7 months' gestation, as 1.3% among 13,563 births reported, with equal rates outside Lusaka province as within. 57 This is similar to estimates from a Global Network study in Zambia, in which 2% of women enrolled delivered stillbirths. 9 The higher proportion of deliveries that resulted in stillbirth in the ZAPPS cohort, at least partly attributable to a broader gestational age range, was reflected in the ZEPRS database, in which 6% of 66,395 deliveries at UTH resulted in stillbirth. 38 However, over half of stillbirths in ZEPRS and 67% in a Global Network study in Zambia were classified as intrapartum, compared to  (<25mm), and (c) twin gestation. Survival curves are presented for participants with increasing numbers of prior preterm birth, those with cervical length <25mm compared to ≥25 mm, and those with twin compared to singleton gestation. The dashed vertical line represents a gestational age of 37 weeks, the threshold for preterm versus term delivery. EGA, estimated gestational age; PTB, preterm birth.  less than 20% in the ZAPPS cohort. These disparities may result from differential classification; it is standard practice outside of our study to classify stillbirths solely by neonatal skin maceration at delivery, particularly in the absence of but even despite the presence of documented fetal heart activity during labor. 9 Indeed, previous studies have demonstrated that reliance on observed skin maceration alone can over-estimate stillbirth proportions attributable to the intrapartum period. 58 This study has several limitations, many of which have been noted previously. 23 First, 16% of participants were lost to follow-up. While this is commensurate with other longitudinal pregnancy cohort studies in the region, 59,60 error may be introduced if outcomes are not missing at random. 61 Women lost to follow-up were younger, more likely to be primigravida and nulliparous, had lower BMIs, and had multiple lower measures of socioeconomic status; many of these characteristics were risk factors for at least one adverse outcome. Further, 250 (21%) of the retained participants either did not deliver at the study hospital or delivered at a time when ZAPPS staff were not present, requiring delivery outcomes to be ascertained by record review and/or participant report (it is worth noting that we found no difference in frequencies of outcomes between deliveries attended by ZAPPS staff versus those that were not; see Underlying data). Second, our data have noted missingness of key antenatal test results at baseline (i.e., hemoglobin, syphilis, and urinalysis) because tests were not routinely repeated nor results recorded in our database if performed at the recruitment clinic before enrollment. Of these test results, only syphilis was associated with an outcome (stillbirth), but we cannot determine with certainty whether missingness introduced bias or simply reduced statistical power. Third, while the ZAPPS study recruits from several surrounding primary clinics, it is based at a tertiary referral hospital and many of our participants were drawn from this higher-risk pool. We note high prevalence of prior PTB, miscarriage, and stillbirth, and high HIV and syphilis seropositivity, which may have resulted from self-selection of high-risk women into a cohort study investigating adverse birth outcomes. It is likely that this resulted in an over-representation of outcomes, but less likely to have also introduced a biased association with identified risk factors.
In summary, the ZAPPS cohort study demonstrates high prevalence of antenatal comorbidities and identifies a number of factors associated with increased risks of preterm birth, small-for-gestational-age infants, and stillbirth. This is the first study of its kind to be conducted in Zambia, and one of the largest on the African continent. An understanding of the true global scope of adverse birth outcomes will require consistent definitions, meticulous ascertainment, and systematic reporting that has eluded those settings where the burden of these outcomes is highest. In the absence of sophisticated registry infrastructure, large pregnancy cohort studies may be able to approximate regional incidence estimates and can provide important data to stratify and direct care for pregnancies at highest risk. Future sub-studies using data and stored biological specimens from the ZAPPS cohort will aim to identify underlying biological mechanisms, causal pathways, and appropriate interventions for the accurate prediction and prevention of adverse birth outcomes in Zambia and worldwide. 1.

2.
In this manuscript, the authors report the rates of adverse birth outcomes such as preterm birth, stillbirth and small for gestational age from a well-characterized cohort of 1450 pregnant women from Zambia. This is a very important article, providing robust data from a low-middle income country like Zambia in the domain of maternal and child health.
The emphasis on accuracy and precision of gestational age estimation which is a cornerstone in preterm birth research is praise-worthy. The data is well represented and the statistical analyses are appropriate.

Comments:
Preterm birth has been defined as birth between 16 0/7 and 36 6/7 gestational weeks in this study.
The definition seems to include stillbirths (the proportion of stillbirth among PTB is reported in paragraph 4 of the results section), which is in variance with the convention of reporting preterm birth among live-born babies. What is the rationale behind this choice? Does such a change in definition influence the preterm birth rate in this study population? If so, how much?
The lower limit of gestational age for defining preterm birth is taken as 16 0/7 weeks in this study. What was the rationale behind considering 16 weeks as the lower cut-off of preterm birth? Further, one of the inclusion criteria is "presentation to antenatal care prior to 20 weeks of gestation if Nearly 50% of the women first present to antenatal HIV-uninfected or 24 weeks if HIV-infected". care after 16 weeks in this population and 25% above 18.3w (median EGA at enrolment is 16.1 (IQR: 13.3-18.3)w). This means there would be some pregnant women in the population who delivered preterm (between 16 & 24w) who couldn't be a part of the cohort because of the delay in seeking antenatal care. Would that create a bias? If so, how much would this influence the study's estimate of the preterm birth rate in the population? Reporting preterm birth rate among the participants enrolled less than 16 weeks of gestation will add further information in this aspect.
The manuscript can be accepted for indexing after the resolution of the comments made above.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Partly
No competing interests were disclosed.

Competing Interests:
Reviewer Expertise: Maternal and child health I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Author Response 14 Nov 2019 , University of North Carolina, School of Medicine, Chapel Hill, USA Joan Price We chose to include preterm stillbirths in our definition of preterm birth and demonstrate the overlapping adverse outcomes of preterm birth, stillbirth, and small for gestational age in Figure 2. We acknowledge significant limitations in differentiating intrapartum stillbirth from preterm birth with immediate neonatal death in our setting. We believe that to exclude stillbirths that occur in the process of parturition from spontaneous preterm birth would falsely lower the rate of spontaneous preterm delivery. As we note in the discussion, the categorization of antepartum vs. intrapartum stillbirth is also imperfect. Of 181 preterm deliveries in our cohort, 48 (27%) were stillborn. Excluding stillbirths from our preterm delivery definition reduced preterm delivery from 15% to 11% and spontaneous preterm delivery from 10% to 8%. Despite this overall reduction, excluding stillbirths from our regression models did not significantly alter our estimates of risk, which supports evidence that risk factors for live and stillborn preterm births demonstrate substantial overlap (Kramer 2012). Finally, our cohort was designed to evaluate the risk factors associated with adverse birth outcomes, and not necessarily to estimate population-level rates of these outcomes. We note in the discussion that our study population likely over-represents high-risk women and therefore overestimates the true population incidence of adverse birth outcomes.
Both lower and upper EGA boundaries for defining preterm birth vary widely in research and national statistics worldwide. Participants at the International Conference on Prematurity and Stillbirth of the Global Alliance to Prevent Prematurity and Stillbirth (GAPPS) have argued for including births occurring 16 weeks onward in the definition of preterm birth, citing studies that show similar etiological risk factors for births occurring as early as 16 weeks and those that occur later in the 2 and 3 trimester (Kramer 2012; Villar 2012). We acknowledge that the risk of preterm birth in any cohort increases with earlier gestational age at presentation for precisely the reason explained by the reviewer. There is a strong relationship between EGA at presentation and EGA at delivery, which therefore produces this effect regardless of relatively arbitrary EGA boundaries used to define outcomes. Indeed, the preterm birth rate among participants who enrolled <16 weeks was 18% in our cohort compared to 12% among those enrolled ≥16 weeks. Because of this, we include EGA at enrollment in all multivariable analyses. We also repeated analyses of preterm delivery outcomes restricting our sample to those participants who presented <20 weeks and again to those who presented <16 weeks to evaluate the potential for bias. These restricted analyses had no effect on our results (with the exception of short cervix as an exposure since it was only performed beyond 16 weeks).
No competing interests were disclosed.
scope (and page limit) of this analysis but may be explored in future analyses.
Thank you for this suggestion. We have added a second Euler diagram to Figure 2 to illustrate the co-occurrence of very PTB, very SGA, and stillbirth. We have added linear regression of continuous exposures and continuous outcomes (Table  6). We have noted in the results section that maternal height was not associated with either PTB or SGA. It was similarly not associated with gestational duration. We have added additional global context to our estimates of the frequencies of adverse birth outcomes in the discussion section. However, we acknowledge significant limitations to direct comparisons of risk estimates due to inconsistent definitions. Furthermore, as there is still no undisputed global standard by which to identify small for gestational age neonates, we have not directly compared the incidence of SGA in our cohort to estimates from high-income countries.
No competing interests were disclosed. Competing Interests: