Reducing stillbirths: behavioural and nutritional interventions before and during pregnancy

Background The vast majority of global stillbirths occur in low- and middle-income countries, and in many settings, the majority of stillbirths occur antenatally, prior to the onset of labour. Poor nutritional status, lack of antenatal care and a number of behaviours increase women's risk of stillbirth in many resource-poor settings. Interventions to reduce these risks could reduce the resulting burden of stillbirths, but the evidence for the impact of such interventions has not yet been comprehensively evaluated. Methods This second paper of a systematic review of interventions that could plausibly impact stillbirth rates covers 12 different interventions relating to behavioural and socially mediated risk factors, including exposures to harmful practices and substances, antenatal care utilisation and quality, and maternal nutrition before and during pregnancy. The search strategy reviewed indexed medical journals on PubMed and the Cochrane Library. If any eligible randomised controlled trials were identified that were published after the most recent Cochrane review, they were added to generate new meta-analyses. Interventions covered in this paper have a focus on low- and middle-income countries, both because of the large burden of stillbirths and because of the high prevalence of risk factors including maternal malnutrition and harmful environmental exposures. The reviews and studies belonging to these interventions were graded and conclusions derived about the evidence of benefit of these interventions. Results From a programmatic perspective, none of the interventions achieved clear evidence of benefit. Evidence for some socially mediated risk factors were identified, such as exposure to indoor air pollution and birth spacing, but still require the development of appropriate interventions. There is a need for additional studies on culturally appropriate behavioural interventions and clinical trials to increase smoking cessation and reduce exposure to smokeless tobacco. Balanced protein-energy supplementation was associated with reduced stillbirth rates, but larger well-designed trials are required to confirm findings. Peri-conceptional folic acid supplementation significantly reduces neural tube defects, yet no significant associated reductions in stillbirth rates have been documented. Evidence for other nutritional interventions including multiple micronutrient and Vitamin A supplementation is weak, suggesting the need for further research to assess potential of nutritional interventions to reduce stillbirths. Conclusion Antenatal care is widely used in low- and middle-income countries, and provides a natural facility-based contact through which to provide or educate about many of the interventions we reviewed. The impact of broader socially mediated behaviors, such as fertility decision-making, access to antenatal care, and maternal diet and exposures like tobacco and indoor air pollution during pregnancy, are poorly understood, and further research and appropriate interventions are needed to test the association of these behaviours with stillbirth outcomes. For most nutritional interventions, larger randomised controlled trials are needed which report stillbirths disaggregated from composite perinatal mortality. Many antepartum stillbirths are potentially preventable in low- and middle-income countries, particularly through dietary and environmental improvement, and through improving the quality of antenatal care – particularly including diagnosis and management of high-risk pregnancies – that pregnant women receive.


Introduction
The vast majority of the world's 3.2 million annual stillbirths occur in low-and middle-income countries [1]. Stillbirths account for half or more of all perinatal deaths. Globally, two-thirds to three-quarters of stillbirths may occur antenatally, before labour begins. The rest are largely associated with complications and/or poor care during labour and delivery [2,3]. While intrapartum care is associated with reductions in fresh stillbirths (deaths < 12 hours prior to delivery), the quality of care during the antenatal period is associated with the incidence of antenatal stillbirths, which occur prior to the onset of labour [4]. Depending on access to and quality of obstetric care, as well as prevalence of antenatal risk factors, the proportion of intrapartum (fresh) and antenatal (macerated) stillbirths may vary; several studies from low-resource settings in Zambia and Pakistan suggest that intrapartum stillbirths may make up a majority of all stillbirths in some settings. Stillbirths during pregnancy have diverse etiologies, and while some causal pathways remain unknown, a growing body of literature suggests that maternal habits like smoking and other toxic exposures, lack of adequate nutrition to the growing fetus, environmental hazards, genetic abnormalities, and maternal infections and conditions contribute to the causation of stillbirths prior to the onset of labour [5].
Rates of stillbirth during pregnancy closely track the use and quality of maternal health care services, being generally higher in economically poorer communities with poor access and/or low utilisation of periconceptual and antenatal care (ANC) services, compared with economically well-off populations with good access and high utilisation of quality ANC services, including monitoring and treatment of risk factors that arise during pregnancy [2]. Hence stillbirth rates, particularly rates of stillbirths that occur before labour begins, can be considered a proxy for access to and quality of reproductive health and ANC services [6].
Interventions delivered during pregnancy, particularly those that alter maternal behaviours to improve maternal health status, minimise fetal exposure to insult, and/or improve utilization and quality of ANC could plausibly prevent antenatal stillbirths, but their impact has not been systematically reviewed and reported. To this end, this paper examines the evidence for non-clinical interventions with a biologically plausible impact on stillbirth incidence that can be delivered via reproductive health or ANC services before or during pregnancy, particularly those that target socially mediated risk factors for stillbirth. Clinical interventions delivered during pregnancy to prevent or manage maternal conditions or infections, another large category of antenatal risk factors for stillbirth, are reviewed in the third paper in this series [7]. Because the burden of stillbirths is highest in low-and middle-income countries, this paper focuses on interventions deliverable in these countries, particularly at the community level, as many stillbirths in these settings occur without any contact with the formal health system, or from care being sought too late to prevent adverse outcomes. Although interventions included in this review are implemented before or during pregnancy, and most interventions impact antepartum stillbirths, some of the interventions may primarily impact intrapartum stillbirths.

Methods
The literature search methodology and data extraction and analysis process used for this paper are summarised in paper 1 of the series [5]. In brief, we reviewed all of the available evidence for the impact of biologically plausible behavioural and nutrition interventions during pregnancy on stillbirth and perinatal mortality, including systematic reviews, peer-reviewed articles in medical journals, and grey literature. Searches included studies published after 1980 including only human subjects, and involved both general searches (e.g., "stillbirth", "fetal death", OR "perinatal mortality") and specific searches for interventions reporting stillbirths as outcome (e.g., iron AND supple-mentation AND pregnancy). Interventions implementable before and during pregnancy (prior to the onset of labour) that were assessed in this paper to determine their impact on stillbirth and perinatal mortality are shown in Table 1.
A total of 130 papers (15 systematic reviews and metaanalyses, and 115 individual studies) met the study criteria and were included in this review.

Prevention of female genital mutilation (FGM) and management of pregnant women with FGM Background
Female genital mutilation (FGM) is comprised of all procedures that involve partial or total removal of the external female genitalia or other injury to the female genital organs whether for cultural or other non-therapeutic reasons [8]. An estimated 140 million girls and women, primarily in African countries and less commonly in Asia and the Middle East, have undergone FGM; an estimated 2 million girls and women undergo the procedure each year [9]. Due to increased migration from Africa, Asia, and the Middle East, FGM has also been increasingly documented in the US, Europe and Australia [10]. Types of FGM include excision of the prepuce and/or partial or complete clitoridectomy (Type I), clitoridectomy accompanied by excision of part or all of the labia minora (Type II), excision of all of the external genitalia with infibulation (surgical narrowing) of the vaginal opening (Type III), or any pricking, piercing, stretching, or cauterisation of the clitoris and/or labia; or vaginal cutting, scraping, or introduction of corrosive substances into the vagina (Type IV). Most of these procedures are irreversible [8,11] and can cause a host of adverse immediate consequences including infection, haemorrhage, and pain, as well as long-term consequences including pelvic infection leading to sterility, scarring, difficulty urinating, increased vulnerability to HIV infection, and childbirth complications [12]. Particularly for women with Type III FGM, there is a risk of obstructed labour and/or severe perineal tears unless defibulation -surgical cutting of the vaginal opening -is performed. Very few studies, however, have examined the association of prior FGM with adverse obstetric outcomes such as stillbirths and perinatal deaths.

Literature-based evidence
We identified 6 relevant observational studies ( Table 2). A large, multi-centre prospective study in 6 African countries by the WHO study group [8] monitored outcomes among women (N = 28,393) attending obstetric centres for singleton delivery. Women were physically examined before delivery to classify them by the World Health Organization (WHO) system described above and then monitored for pregnancy outcome. Compared to women without FGM, the study found a statistically significant increased adjusted relative risk of perinatal mortality for women who had had FGM Type II [Relative risk (RR) = 1.32, 95% Confidence interval (CI): 1.08-1.62] and Type III (RR = 1.55, 95% CI: 1.12-2.16). A retrospective study by Oduro et al. [13] of hospital delivery data in Ghana (N = 5071) found approximately double the rates of stillbirths in mothers with FGM compared to those without (6%, 89/ 1466, vs. 3%, 123/3605), but did not conduct significance testing. In three southwest Nigerian hospitals among women seeking family planning or ANC, Larsen et al. [14] (N = 1851) found that women with FGM had significantly higher risks of tearing (OR = 1.89, 95% CI: 1.04-3.41; P = 0.036) and stillbirths (OR = 1.44, 95% CI: 1.00-2.06), although there was no significant difference between women with Type I and Type II FGM.   ]. An association of FGM with pre-labour fetal deaths has not been documented elsewhere, and the physiological pathway is unclear.
One pathway by which FGM may cause perinatal mortality is by obstructing or prolonging labour. Comparing spontaneous, term, vertex, singleton deliveries among women with FGM (N = 1225) to those without FGM (N = 256) in urban hospitals in Ethiopia, Hakim [17] found significant delays in the second stage of labour, and lower one-and ten-minute mean Apgar scores among women with FGM (p < 0.05), but perinatal mortality rates were not significantly different.

Conclusion
The limited studies examining the impact of FGM on stillbirth and perinatal mortality are of mixed quality and reported mixed results. Overall, the evidence suggests that FGM is associated with increased risk of perinatal mortality or morbidity. Studies from high-income countries likely failed to find consistent associations with perinatal mortality because of the near-universal prevalence of facility-based births and the high quality of obstetric care in those countries [15,16]. The only study from a lowincome country that found no evidence of an impact on perinatal mortality [17] did not grade women based on the degree of FGM, and did not examine stillbirths separately, though it did lend credence to obstructed labour as a causal pathway for stillbirth due to FGM. The largest and most rigorous observational study [8], reported a significant increase in risk of perinatal mortality for women with FGM type 2 and 3, but still failed to report stillbirth and early neonatal mortality rates separately. Only two studies assessed the association of FGM on stillbirth rates specifically [13,14]; both found heightened stillbirth rates among women who had had FGM, but did not control for other variables. As women who have had FGM are likely fundamentally to be different from women who have not had FGM in many contexts, even controlling for all known confounders might not effectively control for all differences between groups. Comparing studies in highversus low-income countries, it appears that the quality of obstetric care, rather than the simple provision of ANC, may be a critical factor in whether FGM is associated with adverse perinatal outcomes. However, all of the studies reflect data from facility-based births. There remains a need for rigorous studies that explore the relationships between FGM, obstructed/prolonged labour, and stillbirth rates (rather than perinatal mortality), particularly for births outside of health facilities.

Birth spacing
Background Short inter-pregnancy intervals (IPI) have been identified as a risk factor for poor pregnancy outcomes, particularly infant mortality, in low-and middle-income countries [18]. Excessively long IPIs (generally exceeding 6 years) are also associated with increased risk of adverse pregnancy outcome. An IPI of 18-23 months was recommended as ideal with respect to the risk of perinatal outcomes by one recent study from the US [19]. Optimally-spaced births have economic, social and demographic significance, and could potentially reduce fetal and maternal morbidity and mortality. Maternal nutritional depletion, competition theory (the concept that pregnancy represents a struggle between mother and fetus for scarce resources) and behavioural risk factors [20] have been used to explain the relationship between short IPIs and adverse perinatal outcomes [21]. For example, while there is evidence that some women seek to avoid pregnancy to recuperate after an adverse pregnancy outcome [22], a number of demographic studies indicate that a large proportion of women seek to become pregnant very quickly after a lost pregnancy, even if the underlying cause of the first adverse outcome remains unresolved [23].

Literature-based evidence
The literature search identified no Cochrane reviews on the subject of birth spacing that reported perinatal mortality outcomes, but 8 observational studies were identified (Table 3). DaVanzo et al. in Bangladesh [24] considered the type of pregnancy outcomes before and after an IPI, as well as the duration of the IPI, controlling for socio-economic and demographic covariates. Of the IPIs that began with a live birth, those <6 months in duration were associated with a 7.5-fold increase in the odds of an induced abortion (95% CI: 6.0-9.4), a 3.3-fold increase in the odds of a miscarriage (95% CI: 2.8-3.9), and a 1.6-fold increase in the odds of a stillbirth (95% CI: 1.2-2.1) compared with 27-to 50-month IPIs. IPIs of 6-14 months were associated with increased odds of induced abortion (2.0, 95% CI 1.5-2.6). IPIs ≥ 75 months were associated with increased odds of all three adverse outcomes, but were not as risky as very short intervals. Women were likely to experience the same adverse outcome sequentially [LOE: 2+]. In the US, Kallan et al. [26] used a logistic regression model to examine the magnitude and shape of the IPI effect on three pregnancy outcomes: pre-term low birth weight, intrauterine growth retardation with low birth weight, and fetal loss. Data were analysed from 104 pregnancies extracted from the US National Survey of Family Growth in 1988. Controlling for demographic variables and reproductive characteristics, short and long IPIs were associated with increased risk of both fetal loss and intrauterine growth retardation with low birth weight [LOE: Examining prolonged birth spacing in an urban hospital in Nigeria, Orgi et al. [27] conducted a case-control study to determine the reasons for prolonged birth spacing and Assessed the impact of the spacing of pregnancies on outcome.
Women who experience a wastage at any given pregnancy number are not only more likely to have another pregnancy, but they do so over a short time interval than those whose last pregnancy resulted in a live birth. Except for terminations, wastage is highest among women who closely space their pregnancy.
Kamau and Mati 1988 [122] Kenya (Nairobi). Kenyatta National Hospital. Cross sectional survey. Women (N = 615) delivered during the months of June and July 1985, who had at least one birth interval to report (N = 2407 pregnancies and 1792 birth intervals).
Assessed the impact of birth intervals on pregnancy outcome.
SBR and first week death rates: the lowest rates (1.9% and 3.2% respectively) were observed when the preceding birth interval was 25-36 months. PMR: 5.2% for this interval. Birth intervals that were 25-36 months long were associated with the most favorable pregnancy outcome. Poor pregnancy outcome was followed by very short birth intervals. There are other studies that have assessed the impact of birth intervals, rather than IPIs, on pregnancy outcomes. Among women (N = 415) delivering at a semi-rural hospital in Ethiopia, Abebe et al. [28] conducted an interview study to assess determinants and consequences of birth intervals. In the sample of 415 women with 2009 pregnancies, the mean birth interval was 22.1 months; 13.2% were <12 months. Almost half (42.3%) of these short birth intervals resulted in miscarriage (32.2%), stillbirth (3.2%), or neonatal mortality (6.9%). Adverse outcomes declined with increased birth interval. Unfortunately, this study assessed intervals between reproductive events rather than inter-conceptional intervals, which may overrepresent the prevalence of poor outcomes associated with short IPI [LOE : 3].

Conclusion
Relatively few studies have evaluated the impact of IPI on stillbirths and perinatal mortality, and all have been observational (overall Grade C evidence). While most observational studies suggest that short IPIs are associated with increased perinatal mortality, few reported stillbirths as a primary or secondary outcome. In several studies, after controlling for possible confounding, the association of IPI with adverse outcome became non-significant, but it is possible that studies were unable to appropriately control for confounding given differences between women with shorter versus longer IPIs. Regardless of impact on birth outcomes, family planning and options for desired birth spacing present a key intervention to improve maternal health and nutrition, birth outcomes and population health in general.

Reducing indoor air pollution Background
Half of the world's population relies on the burning of solid fuels for everyday energy needs. These fuels are typically burned indoors or in partly enclosed cooking areas using poorly vented, inefficient stoves. Because cooking is commonly considered the domain of women, smoke exposure is typically highest for women of childbearing age and their young children. Women usually continue with their cooking duties while pregnant, indirectly exposing the developing fetus to harm that could result in stillbirth or neonatal death. Fuel smoke may lead to impaired fetal growth due to hypoxia and/or oxidative stress from smoke constituents such as carbon monoxide and particulates.

Literature-based evidence
Three recent South Asian studies assessed the risk of stillbirths associated with indoor air pollution from solid fuel use (Table 4). In India, Mavalankar et al. [29] evaluated risk factors in stillbirth cases (N = 451) and normal controls (N = 1465), finding a nonsignificant increase in risk of stillbirth among those exposed to smoke (OR = 1.5, 95% CI: Another Indian study of population-based data from the second Indian National Family Health Survey by Mishra et al. [30] examined the association between household use of biomass fuels (wood, dung, and crop residues), tobacco smoke (both active and passive), and risk of stillbirth. Adjusting for confounders, the study found that women who cooked with biomass fuels were significantly more likely to have experienced a stillbirth than those who cooked with cleaner fuels (

Conclusion
The above studies, while limited in geographic scope and largely observational, provide uncertain level evidence linking exposure to indoor air pollution with increased risk of stillbirth (Grade C evidence). There is a need for rigorous prospective randomised intervention studies in relevant contexts to determine whether exposure to biomass fuels is causal for stillbirth, as the existing studies on the subject are observational and cannot control for a number of sociodemographic differences between women who use cleaner versus less clean fuels. There is also a need to develop and test behavioural and structural means of minimizing exposure of pregnant women to indoor air pollution, particularly exposure to biomass fuels. These interventions could include behaviour change, communication about the risks of indoor air pollution, construction of improved cooking implements or increasing availability and lowering the cost of alternative cooking methods and/or cleaner fuels, one example being the solar-powered cookers utilized in Kenya, Zimbabwe, and Nepal, among other countries [32].

Smoking cessation in pregnancy Background
Smoking is one of few potentially preventable factors associated with a number of poor pregnancy outcomes including low birth weight (LBW), pre-term birth, stillbirth, and neonatal death. Prevalence studies in the 1990s show that 20-33% of pregnant women in developed countries reported smoking [33], and rates have been increasing in many low-and middle-income countries.
Smoking has been associated with a 50% increased risk of intrapartum stillbirth compared to non-smokers (adjusted HR = 1.5; 95% CI: 1.3-1.7). Heavy smoking (10-19 cigarettes per day) raised the risk of intrapartum stillbirth by 70% compared to non-smokers (adjusted HR = 1.7, 95% CI: 1.4-2.0) [34]. The main constituent in tobacco, nicotine, crosses the placenta with fetal levels reaching 15% higher than the maternal levels [35]. Nicotine causes vasoconstriction of the uterine and possibly the umbilical artery, and may also have a direct deleterious effect on the central respiratory control mechanism, leading to fetal hypoxia-ischemia and ultimately, stillbirth.
Interventions for smoking cessation among pregnant women mainly employ behavioural and cognitive counselling [36]. Effective behavioural intervention programs include practical advice in problem solving and skills training and provision of social support [37]. Other strategies include multi-media education campaigns, telephone quit lines, fiscal incentives, and biomarker feedback [38]. The American College of Obstetricians and Gynecologists has proposed a five-step intervention program, called the "5 A's" model (ask about use, advise to quit, assess willingness to quit, assist, and arrange followup) to assist pregnant women quit smoking [39,40].
Women not ready to quit are to be given motivational intervention via "5 R's" (relevance, risk, rewards, roadblocks, and repetition) [41].
Pharmacotherapy can also be potentially used for strongly addicted women [36]. Among pharmacotherapies, nictotine replacement therapy (NRT) during pregnancy has not been approved by the US Food and Drug Administration, as its efficacy has not been studied in controlled clinical trials [36]. Nicotine is considered teratogenic and so the safety of NRT is questionable. The largest study on counselling and use of a nicotine patch (vs. a placebo patch) from Denmark, consisting of 250 pregnant women found no statistically significant difference in smoking cessation rates between the two groups (28% vs. 25% at 4 weeks before delivery, respectively) [42]. Tricyclics, bupropion, and mono-amine oxidase inhibitors are other possible drugs, but only bupropion appears to be effective [38].
There are two small comparative studies from highincome countries looking at the safety of bupropion in pregnancy [43,44], and the larger one [44] found no statistical differences between bupropion and other comparison groups in risk of malformations or miscarriage. There are newer therapies like bromocriptine, varenicline and cytisine, but safety data is limited to animal studies or non-randomised trials [38]. Nicotine vaccine is another novel therapy, but has not yet been tested in phase 3 trials [38].
Smoking cessation efforts targeted towards pregnant women and others in their households could plausibly prevent stillbirth and other adverse pregnancy outcomes; published evidence for impact of smoking cessation is reviewed below.

Literature-based evidence
Our literature search identified one Cochrane review on smoking cessation programmes during pregnancy and two other observational studies ( Table 5 . The metaanalysis found no statistically significant impact of smoking cessation programs on very low birth weight (< 1500 g), stillbirths, or neonatal mortality, but the analyses had limited power to find a significant difference. The strategy of combining rewards for smoking cessation with social support (as in 2 trials), resulted in a significantly greater smoking reduction than other strategies (RR = 0.77, 95% CI: 0.72-0.82). There were no statistically significant reductions in relapse.
Two observational studies explored the impact of pharmacotherapies on smoking cessation in pregnancy. Chun-Fai-Chan et al. [43] conducted a drug safety study that compared bupropion to other anti-depressants and nonteratogenic smoking cessation aids, and found no statistically significant differences in stillbirth rate between the two groups studied. A cohort study by Strandberg-Larsen 2008 [45] using Danish National Birth Cohort data (N = 87,032 singleton pregnancies) assessed the impact of use of nicotine replacement therapy. The stillbirth rate was 4.2 per 1000 among users of NRT with a crude hazard ratio (HR) of 0.75 (95% CI: 0.37-1.15) compared with nonusers. This result must be interpreted with caution as the sample size of NRT users (N = 1927) was small and the data were not originally collected for this study.

Conclusion
The single good quality Cochrane review [33] that examined the impact of smoking cessation programs on stillbirths and perinatal mortality provides strong evidence of reduction in LBW and pre-term birth rates, confirming that smoking cessation can reverse the adverse effects of smoking on perinatal outcomes. However, the study sizes of the component studies were too small to observe any statistically significant reduction in stillbirth incidence. Additionally, all evidence for the impact of smoking is from developed countries; there may be fundamental differences in other risk factors between smokers in highincome vs. low-/middle-income countries that are not yet known and that could influence the effect. The success rate of behavioural interventions, even when culturally appropriate, is very modest in low-/middle-income countries and intervention studies in this regard are urgently needed. The data on pharmacotherapy is even further limited. Our overall evaluation of the evidence for smoking cessation in relation to stillbirths is Grade C. Smoking cessation clearly benefits both mother and fetus, but there is a need for appropriate studies to confirm trends toward reduction in stillbirths and early neonatal deaths. These may need to be undertaken on available large-scale data sets.

Reducing smokeless tobacco exposure during pregnancy Background
Tobacco can be smoked in cigarettes, bidis (thin Indian cigarettes wrapped in a leaf with a thread), cigars, or pipes, or alternatively chewed or sniffed. In South Asia, for example, gutka (crushed betel nut with tobacco and other ingredients and flavourings), betel quid and mishri (pyrolysed and powdered tobacco) are routinely used forms of smokeless tobacco, often as a dentifrice. While women in developed countries smoke cigarettes more than they use smokeless tobacco, use of smokeless tobacco is wide-spread and increasing among women in many low-and middle-income countries [46]. Use of smokeless tobacco has been reported to be as high as 17.1% among pregnant women in India [47]. While the association of cigarette smoking with stillbirth incidence has been well studied, evidence of the impact of smokeless form of tobacco on stillbirths is more limited because of the lower level of attention afforded to smokeless tobacco use.

Literature-based evidence
We identified three studies on the smokeless tobacco exposure among pregnant women and its impact on stillbirths or perinatal mortality ( Table 6). No intervention studies were identified that tested reductions of exposure to smokeless tobacco with reported outcomes on stillbirths or perinatal deaths.
A well-conducted cohort study by Gupta and Subramoney [47]  Two hospital-based studies from India explored the use of smokeless tobacco among pregnant women; one crosssectional study (N = 1388 singleton births) reported a 3fold increased risk of stillbirth among tobacco chewers compared to controls [48]. The other study, a case-control design, reported an increased risk of perinatal death (RR = 1.5, 95% CI: 1.3-1.7) [49].

Conclusion
Based on the above studies, smokeless tobacco use emerges as a clear risk factor for stillbirth; its use is associated with an increased risk of stillbirth similar in magnitude to that associated with maternal cigarette smoking.
Smokeless tobacco use has also been linked to other adverse pregnancy outcomes including growth restriction, pre-term delivery, and changes in placental morphology, which may mediate the higher stillbirth risk observed; nicotine exposure alone may also lead to fetal hypoxiaischaemia. Studies on smokeless tobacco use and its physiological effects in pregnancy have been mainly limited to South Asia; more studies are needed from other countries where its use may be prevalent. Additionally, there is a need for studies that design and test cessation programmes for smokeless tobacco use, to determine whether these programmes can effectively reduce smokeless tobacco use and thus impact perinatal outcomes including stillbirth incidence.

Antenatal care (ANC) packages Background
The main function of ANC is to prevent or identify and treat conditions that may threaten the health of the fetus, newborn and/or the mother and to help a woman approach pregnancy and birth positively. In practice, ANC packages comprise a wide constellation of interventions that a pregnant woman receives from organised health care services, often provided in clinics or through outreach services. Although the World Health Organization (WHO) has proposed standardised content and visitation schedules, in different countries and different settings [50], the component interventions that comprise an ANC package may vary widely.
In low-and middle-income country communities, where it may be difficult for women to access facility-based care and where births often occur at home, ANC may be provided through primary health care or outreach clinics using various nursing cadres and physicians, and occasionally by female community health workers who make household visits. At a basic level, ANC components may include taking the woman's medical and obstetric history and general health assessment, measuring weight gain and fundal height, administration of two doses of tetanus toxoid immunisation, counseling on birth preparedness and postnatal care, and distributing vitamin supplements (especially iron and folate). In malarious areas, ANC may also include malaria chemoprophylaxis, intermittent preventive treatment, and bednet distribution. Where technically and economically feasible, ANC may also include screening for maternal infections and conditions such as pregnancy-induced hypertension, gestational diabetes, and STDs including HIV and syphilis.
This variability of ANC based on local population characteristics and capacity of the health system complicates measurement, because the particular interventions delivered -and potential synergies between these interventions -may have an impact on stillbirths as well as neonatal outcomes. Other variables that may have an impact on stillbirth rates include the quality of care as well as the frequency and timing of its delivery. Relatively few studies of ANC have specifically considered stillbirth as an outcome.

Literature-based evidence
The literature review identified 3 Cochrane reviews comprised of 14 RCTs; one large WHO meta-analysis; and 24 other studies ( Table 7, 8,9,10), evaluating many different facets of ANC, including comparisons of the timing and frequency of ANC visits, the type of provider, and the impact of ANC packages and specific component interventions on perinatal outcomes.

Impact of ANC on mortality
Several studies from high-, middle-and low-income countries have found significant impact of facility-based ANC on perinatal outcomes. Some studies examined the impact of not having ANC on perinatal outcomes. Southwick et al [51] found that among women with inadequately treated current syphilis, those without ANC were more likely to have a stillborn infant than those with ANC (OR = 9.5; 95% CI: 4.0-23.

Frequency of ANC visits
The number of visits and type of provider may impact perinatal outcomes. A systematic review of 7 RCTs (N = 57,418 women) conducted by WHO [57] (Additional file 2) evaluated the effectiveness of different models of ANC and found that reduced numbers of visits were as effective as standard models of ANC in terms of impact on LBW and PMR. Five randomised trials (two individual-ran-  [58].

Type of providers
Whether ANC can be more effectively provided by practitioners other than doctors has important cost and coverage implications. ANC can be managed effectively by general practitioners or midwives, rather than obstetricians, without negatively affecting maternal and birth outcomes [ [69] assessed the impact of a rural regional perinatal care program in North Carolina, USA, using a quasi-experimental, controlled, population-based design to identify high-risk pregnancies during antenatal visits and ensure access (including trans-port) to higher-level care for complications. They observed declines in fetal, neonatal, and birth-weight-specific mortality rates in both pilot and control regions, especially for 1501-2500 g infants, though these changes were not statistically significant between regions [LOE:

2-].
Hodnett et al. [70] conducted a meta-analysis of intervention studies [N = 11 trials, N = 9507 women] that offered additional support to at-risk pregnancies by either a professional (social worker, midwife, or nurse) or specially trained lay person, and found no significant evidence that these interventions were any more effective than routine ANC (RR = A number of promising interventions focused on upgrading or improving health systems in rural areas among poor populations and brought about documented declines in PMR. In rural township clinics and hospitals in South Africa, Wilkinson [72] found that structural and functional changes in the maternity services throughout the district, using standard protocols for care, and conducting in-service training, effectively and rapidly reduced perinatal mortality by one-third

Improving maternal access to and utilisation of ANC
At the community level, other variables that impact the

Conclusion
The effectiveness of ANC for a host of maternal and newborn outcomes has been reviewed elsewhere [80,81]. While it is recognised that effective and appropriate ANC safeguards maternal health and promotes positive pregnancy outcomes and the overall quality of the evidence is relatively good (Grade B), the evidence that it makes a specific impact on stillbirth outcomes remains weak. This may also be related to the fact that relatively few largescale studies have evaluated stillbirth outcomes, and most of the studies that identify significant impact of an intervention appear to be affected by selection bias, where women who received ANC likely differed from women who did not receive ANC. Many studies were not adjusted for potential confounders, and many confounders cannot be adequately controlled for in vastly different study arms. Many of the studies reviewed, moreover, did not specify component interventions that made up the ANC packages. Assessment of the impact of ANC on stillbirth incidence requires well-designed clinical trials for each component intervention, which have not been performed, nor may be feasible given the work undertaken with the WHO collaborative studies.
Routine ANC differs greatly from country to country and services offered generally do not follow evidence-based criteria for improving maternal or neonatal outcomes, much less prevention of fetal deaths. Issues of delays in seeking and accessing care, and poor quality or availability of care, remain unaddressed in many ANC and obstetric care programs, yet these issues are key to preventing maternal and perinatal deaths. In reviewing clinical data from Mexico, Salinas et al. [82] postulated that 35% of all perinatal deaths could have been prevented if structural (e.g., unavailable or poorly maintained equipment, absent or insufficient health workers) and process deficiencies (e.g., incorrect diagnoses or poorly performed procedures), had been addressed (P < 0.0001).
Based on the results of the WHO ANC randomised trial [59] supported by other studies, models of ANC with a reduced number of antenatal visits can be introduced -in high-, middle-, and low-income countries -into clinical practice without any risk of adverse consequences to the woman or the fetus, provided those visits are appropriately focused on effective interventions and quality of implementation is ensured.

Peri-conceptional folic acid supplementation Background
Women taking folic acid supplementation peri-conceptionally -before pregnancy and during the first two months of pregnancy -are less likely to give birth to babies with neural tube defects (NTDs), which account for a small proportion of stillbirths. The precise mechanism by which folic acid is protective is unclear. Bjorkland et al. [83] hypothesised that folic acid provides the methyl group used for post-translational methylation of arginine and histidine in the regulatory domains of the cytoskeleton, which is required for neural tissue differentiation.

Literature-based evidence
The literature search identified 3 Cochrane reviews assessing 8 RCTs; and 1 other relevant intervention and observational study (  [85], and tested peri-conceptional folic acid against multivitamin supplementation in women who had borne a previous child with an NTD (N = 1817). Folic acid supplementation was associated with a 72% reduced risk of NTDs (RR = 0.28, 95% CI: 0.12-0.71), whereas the multivitamin supplement had no significant protective effect The other meta-analyses and studies we reviewed were designed to assess the impact on perinatal outcomes of antenatal administration of folic acid. The review by Pena-Rosas [86] of 2 RCTs found no associated reduction in PMR (Additional file 7). Few studies reported stillbirth incidence, and those that did, reported only non-significant reductions in SBR [87] (Additional file 8).

Conclusion
Evidence of the protective effect of peri-conceptual folic acid supplementation on NTDs (Grade A evidence) suggests that peri-conceptional and antenatal folic acid supplementation offered to all pregnant women, particularly women with a prior affected pregnancy, and public health measures including iron-folate distribution or fortification of food products, may ensure that women of childbearing age have adequate folate intake and consequently reduce NTDs [88]. Evidence from both Cochrane reviews and individual studies is weak regarding impact of folic acid supplementation on stillbirth incidence, and there is no data on its potential to prevent early miscarriages, rendering our assessment of the evidence for an impact of peri-conceptional folic acid supplementation on stillbirths as merely uncertain. Because the biological pathway by which folate deficiency could result in stillbirths arising from severe NTDs is established, and the causal association of folic acid with NTD prevalence has been demonstrated, there is a need for rigorous RCTs of periconceptional administration of folic acid powered to assess stillbirth outcomes. Peri-conceptional administration would improve the likelihood that folate deficiency is remedied early enough in gestation to act in neural tube formation.

Iron (or iron-folic acid) supplementation
Background Anaemia is defined as the reduction in the normal number of red blood cells and quantity of haemoglobin in the blood, defined by the WHO as haemoglobin <11 g/ dl [89].
Routine iron supplementation to pregnant women during pregnancy has been shown to improve biochemical indicators of iron status and to reduce the risk of low maternal haemoglobin at delivery and at 6 weeks postpartum. While there is little information on the impact of iron on functional outcomes, iron (or iron plus folic acid) supplementation is recommended for pregnant women in areas with endemic iron deficiency [90].

Literature-based evidence
The review of literature identified 2 Cochrane reviews comprised of 3 RCTs, as well as 2 intervention studies (Table 12). In their Cochrane meta-analysis which included just one trial comparing two-thirds of the recommended dose of intravenous iron versus full-dose intravenous iron in iron-deficient women, Reveiz L et al. 2007 [91] reported a non-significant reduction in stillbirth risk (RR = 0.70, 95% CI: 0.25-1.93 [NS]) among the group given the two-thirds dose (Additional file 9). As this small study did not meet the Cochrane quality standards for assessing effectiveness and stillbirths were not a primary outcome, this finding may be artifactual [LOE: 1+]. The most recent Cochrane systematic review of antenatal iron supplementation by Pena-Rosas et al. [86] summarised 40 trials (N = 12706 women) with either daily or weekly therapy (Additional file 10). Few of the included studies reported perinatal outcomes or stillbirths, but two trials in this meta-analysis revealed a non-significant increase in PMR in the iron/folic acid recipients compared to placebo (RR = 2.50, 95% CI: 0.10-59.

88) [LOE: 1+]
Our review also identified another study of iron supplementation in pregnancy. In The Gambia, Menendez et al [92] reported a 33% decrease in SBR after supplementing women with 200 mg oral ferrous sulphate, but furnished no significance statistics.

Conclusion
Given widespread anaemia during pregnancy, routine iron supplementation in doses ranging from 60 mg to 300 mg of iron per day is advised during pregnancy for a host of well-documented maternal benefits (overall Grade B evidence). The impact of maternal iron supplementation on stillbirths has not been reported by many studies, however, and the limited available evidence is mixed, due in part because most studies are underpowered to detect differences. Nevertheless, particularly in areas with high prevalence of iron deficiency anaemia, iron supplementation packaged with folic acid is recommended for periconceptual administration for maternal health, but further research is needed to define its impact on stillbirths.
Despite some evidence of impact on perinatal mortality, our overall assessment of the impact of iron or iron-folic acid supplementation on stillbirths indicates no overt benefit.

Vitamin A/β-carotene supplementation during pregnancy Background
Vitamin A is required for fetal tissue growth and maintenance, as well as maternal metabolism; it plays an important role in cell differentiation and, therefore, embryogenesis very early in pregnancy [93,94]. Clinical and subclinical Vitamin A deficiency is a public health concern in at least 75 countries worldwide, and contributes to impaired immune host response, maternal eye problems including xerophthalmia, night-blindness, as well as decreased haemoglobin levels and anaemia [95][96][97]. Vitamin A deficiency co-exists with iron deficiency in many countries [98], and is linked with an elevated risk of LBW, as well as an increased risk of infections (diarrhea, dysentery, acute respiratory illness) and poor growth in young infants [99]. Given the potential teratogenic effect and toxicity of high-dose vitamin A, a variety of vitamin A interventions (daily supplementation with either lowdose vitamin A or β-carotene) have been evaluated with assessment of safety and impact on maternal, newborn and infant outcomes.

Literature-based evidence
Three Cochrane reviews comprised of 9 RCTs were identified following the literature search (

Conclusion
The existence of a Cochrane review on the subject and several strong RCTs allow us to rate the level of evidence as Grade C [96], but the available evidence suggests no role of vitamin A in preventing stillbirths. Many of the studies in the review suffered from design weaknesses, and comparability was hampered by differing outcome definitions, suggesting that there is still a need for more rigorous evidence on this subject. The largest RCT reporting the impact of vitamin A on stillbirths [101] also indicates no benefit of Vitamin A for fetal or early infant survival, although vitamin A does appear to be of benefit to maternal health and survival, and vitamin A deficiencies are known to co-exist with iron deficiency [98]. Any future studies of vitamin A supplementation must focus on highrisk women, assess the most appropriate vitamin type and dosage, and show that the intervention is both safe for the mother and fetus, and effective.

Multivitamin/multiple micronutrient supplementation during pregnancy Background
Nutritional deficiencies are common during pregnancy, particularly in low-and middle-income countries where the diets of pregnant women are often less nutrient-dense than those of women in high-income countries. Inade-
To assess the efficacy, effectiveness and safety of routine antenatal daily or intermittent iron supplementation with (intervention) or without (control) folic acid during pregnancy on the health of mothers and newborns. quate dietary intake before and during pregnancy may also pose fetal risks, and could plausibly be compensated for with appropriate supplementation of missing nutrients. Multiple deficiencies often co-exist, thus supplementation with multiple nutrients simultaneously may have an increased, possibly synergistic effect on perinatal outcomes. Routine maternal nutrient supplementation in low-and middle-income countries is generally restricted to provision of iron-folate supplements. Change in practice toward supplementation with multiple micronutrients (MMN) has been hindered because of a lack of data about the impact of MMN on fetal and neonatal outcomes.

Literature-based evidence
The review of literature identified 3 Cochrane reviews, which included 17 RCTs, and 7 other quasi-experimental and observational studies (      (controls). We found a non-significant trend toward reduced stillbirths among the intervention group versus the control group (RR = 0.91, 95% CI: 0.80-1.03).

Conclusion
Despite the clinical importance of remedying nutrient deficiencies prevalent in low-income countries during the antenatal period, and several systematic reviews (Grade B quality evidence), the data provide only some evidence of impact of multiple micronutrient supplementation on stillbirth incidence or perinatal mortality. Although the previous review by Rumbold et al. [87] found no evidence of impact of multiple micronutrient supplementation on early or late miscarriage or stillbirth, the new large RCT by Shankar et al. study [104] and our new meta-analyses provide evidence suggestive of a small impact on stillbirths. The Shankar et al study suggests that the potential benefit of multiple micronutrient supplementation is likely to be greatest in the most nutrient deficient populations. There remains insufficient evidence of adverse effects related antenatal multiple-micronutrient supplementation on neonatal mortality, although this possibility requires further investigation [106]. Any future studies of vitamin supplementation should focus on women at high risk of miscarriages and stillbirths, assess the most appropriate vitamin type and dosage, and test for benefit without risk of harm to mother or fetus.

Regular magnesium supplementation in pregnancy to alleviate deficient states
Background Magnesium occurs naturally in many foods, making magnesium deficiency rare in healthy individuals eating varied diets [107]. Dietary intake studies during pregnancy, however, consistently demonstrate that many women, especially those from economically disadvantaged backgrounds, have intakes of magnesium below recommended levels [108]. Magnesium plays a role in regulating body temperature, protein synthesis, and nerve and muscle function. In laboratory rats, magnesium deficiency was associated with higher systolic blood pressure and elevated plasma nitrite, suggesting that magnesium may play a role in controlling blood pressure, though the mechanism is unknown [109]. Magnesium supplementation during pregnancy has been associated with a reduced risk of fetal growth retardation, pre-eclampsia, and increased birth weight [110], but these findings were gathered from non-randomised, non-controlled studies.

Literature-based evidence
A review of the literature identified one Cochrane review assessing the potential benefits of magnesium supplementation during pregnancy on pregnancy and neonatal outcomes (Table 15). Makrides and Crowther [107] reviewed the impact of magnesium given to all pregnant women in RCTs and quasi-RCTs (N = 7 trials, N = 2689 women) (Additional file 17). Meta-analysis revealed that oral magnesium treatment from before the 25th week of gestation was associated with a lower frequency of preterm birth (RR = 0. 73

Conclusion
Our overall assessment of the evidence based on studies was Grade C. Although meta-analysis [107] suggested that oral magnesium treatment resulted in lower incidence of pre-term birth, LBW and SGA babies compared with placebo, the findings were weighted heavily by a trial in Hungary which did not appropriately adjust for cluster randomisation. With the Hungarian trial excluded, dietary magnesium supplementation had no evidence of impact on pregnancy outcomes. Dietary magnesium supplementation of pregnant women cannot be recommended for routine clinical practice because of the poor methodological quality of the current evidence. Further trials are warranted and should be of high quality, including concealment of allocation, appropriate unit of randomisation, selection of placebo, blinding of outcome assessments and minimisation of losses to follow-up. Because there is no evidence of impact of magnesium supplementation on stillbirth incidence, it cannot be recommended as a strategy to prevent stillbirths at this time.

Balanced protein-energy supplementation in pregnancy Background
Balanced protein-energy supplements, by definition, provide less than 25% of their total energy content in the form of protein. A previous observational study [111] reported that both gestational weight gain and energy intake were strongly and positively associated with fetal growth, and possibly associated with a reduced risk of preterm birth. Moreover, these associations were stronger in undernourished women, i.e., those with low pre-pregnancy weight-for-height. It is plausible that the improved fetal growth associated with simple supplementation with balanced-protein energy supplementation could result in reduced rates of stillbirth.

Literature-based evidence
Our literature search identified one Cochrane review, comprised of 6 RCTs, and one quasi-experimental study ( Table 16). The Cochrane review [112] consisted of controlled trials of dietary advice to increase or reduce energy or protein intake, or of actual protein supplementation or restriction during pregnancy. Supplementation did not impact pre-term birth, but significantly reduced risk of stillbirth and neonatal death. The pooled reductions in stillbirth rates (RR = 0.55, 95% CI: 0.31-0.97) and neonatal deaths (RR = 0.62, 95% CI: 0.37-1.05) of balanced protein-energy supplementation in pregnancy were based on four trials and were heavily influenced by a single large trial in The Gambia [113] (Additional file 18). The Gambian study also included micronutrient supplementation with balanced protein-energy supplementation; excluding this single study drastically altered the conclusions of the meta-analysis, leaving no demonstrable impact. High protein supplementation in pregnancy was associated with a non-significant reduction in stillbirth [114] (RR = 0.81, 95% CI: 0.31-2.15); the effects of isocaloric balanced protein-energy supplementation in pregnancy was non-estimable because the study reported no stillbirths in either group [115]. The Cochrane review found one study of nutritional advice during pregnancy, which was associated with a non-significant reduction in stillbirths [116] (RR = 0.37, 95% CI: 0.07-1.90). [LOE: 1+].
In rural Punjab, India, a quasi-experimental trial [117], allocated villages to receive medical care alone, balanced protein-energy supplementation alone, medical care plus nutritional supplementation, or no care (controls). Rates of stillbirth were lower in all intervention villages (P < 0.05 compared to controls), and lowest in villages receiving nutritional supplementation (P < 0.025) [LOE: 2+].

Conclusion
The evidence for impact of balanced protein-energy supplementation while based on a reasonably large number of meta-analyses and studies in representative populations, is heavily weighted by the results of one study (Grade B evidence). However, there are adequate reported events of interest i.e. stillbirths/perinatal mortality for one to draw conclusions of benefit. The body of evidence suggests balanced protein-energy supplementation may be useful to prevent stillbirths in poor populations at risk of food insecurity but further well-designed studies are needed to address the impact on stillbirths.

Summary
Most studies identified in this review were observational; rigorous RCTs were virtually non-existent. Only rarely did studies report information about stillbirth reduction, and rarely did any study report a statistically significant impact on stillbirth incidence. Many of the interventions that are being practised during pregnancy are relatively unsupported by evidence, and potential risk factors for antenatal stillbirth are still being identified (e.g., exposure to indoor air pollution). Because of this lack of data, it is premature to recommend most of the interventions that we reviewed for routine practice.
The nutrition interventions described in this review had good quality Cochrane reviews and meta-analyses. For almost every intervention we considered, even where multiple RCTs and reviews were available, stillbirth and perinatal outcomes were often not reported, or reported only as secondary outcomes, rendering them underpowered to detect differences in stillbirth rates. To illustrate, many studies with peri-conceptional folic acid supplementation evaluated neural tube defects (NTDs) as the main outcome of interest, yet very few reported stillbirths. Rather than evidence of no effect, there is no evidence of effect of folic supplementation; further trials are needed to understand whether folic acid supplementation could bring about reductions in stillbirth, or just NTDs. Larger RCTs of all the nutritional interventions we reviewed are needed. The marginally significant reductions in stillbirths associated with multiple micronutrient supplementation were more pronounced among nutrient-deficient women, suggesting that RCTs in deficient populations are most promising to detect any possible impact on stillbirths.
Social and behavioural factors appear to contribute to the risk of adverse birth outcomes and stillbirths through multiple causal pathways, but these risk factors and their related interventions have attracted the least attention from researchers of all of the interventions we reviewed in this series of papers. Reducing rates of FGM, especially infibulation, could reduce rates of obstructed labour where infibulation and home birth with unskilled attendants are common, but more studies are needed. Short inter-pregnancy intervals (IPIs) have been repeatedly associated with increased stillbirth rates but causation has not been established, and logistic regression has rendered associations non-significant, suggesting that better designed studies to assess inter-conceptional intervals and contraceptive use, along with qualitative research about fertility intentions after adverse outcomes, are needed. Rates of smoking and exposure of women to second-hand smoke have been increasing throughout low-and middleincome countries; reducing exposure to smoke, including combustion particulates and by-products, is of clear benefit for maternal and infant health, but further large studies are needed to determine whether smoking cessation reduces stillbirth rates, particularly in low-and middleincome country settings. Exposure to air pollutants, particularly to smoke from heating and cooking fires, appears strongly associated with a woman's risk of stillbirth, but interventions to reduce exposure that measure birth outcomes have not yet been tested. Innovative options are already available for cleaner fuels and smoke-free heating/ cooking methods that would lend themselves to a community-based intervention trial to test whether these options could reduce stillbirth rates. Similarly, smokeless tobacco use has been associated with a statistically significant higher risk of stillbirths, and this is a subject to be included in the research agenda. Interventions to reduce the use of this form of tobacco and their subsequent impact on perinatal outcomes are lacking. Other social factors and behaviours before and during pregnancy likely increase women's risk of antepartum stillbirths; research is recommended to investigate potential social risk factors and appropriate interventions to mitigate these risks.
Most of the interventions we explored in this paper rely on a minimum of an outreach-based model of ANC. ANC is widely accessed by pregnant women worldwide, even though complete coverage (at least 4 visits) varies, as do components and quality of ANC. The health system contact that an ANC visit provides offers an opportunity for delivery of nutritional interventions (including supplements), behaviour change communication efforts, and clinical care including identification of high-risk pregnancies [7,118]. Although we found no clear association of ANC with reductions in stillbirth, there is evidence from observational studies that having no ANC is associated with higher rates of stillbirth, likely attributable both to the specific components and quality of ANC services and to absolute differences in health status, behaviours, and exposures among women who access ANC compared with women who do not.
In low-and middle-income countries, ANC is an essential platform upon which the prevention of both antenatal and intrapartum stillbirths depends. ANC provides a contact point to diagnose maternal conditions such as hypertensive disorders leading to fetal growth restriction and pre-term birth, as well as maternal infections such as syphilis, which may be associated with up to half of all stillbirths in high-prevalence settings in southern Africa [119]. Improving the capacity of ANC services to effectively identify, treat, and monitor these conditions and diseases [7,118,120] may bring about important reductions in stillbirth rates.
Health program planners and providers of ANC should carefully choose and more routinely document and evaluate the services they provide, and explore ways that ANC could be expanded to include evidence-based interventions that screen for and treat pre-existing and gestational conditions and infections, such as anemia, syphilis and hypertensive disorders, that may lead to stillbirths [7]. Health systems strengthening activities are also needed to provide the necessary referral-level care and community-

Pilot/cohort studies of interventions
• Trials of alternative cooking technologies or cleaner fuels* • FGM (especially infibulation) vs. no FGM in non-facility-based births* • Birth spacing studies, including identification of behavioural/emotional factors after a loss leading to short subsequent IPIs.
Well-designed large RCTs of interventions powered to detect stillbirths • Effective mutritional interventions, particularly balanced protein-energy supplementation and multiple micronutrient supplementation* • ANC packages with clearly defined component interventions • Iron (or iron-folate) supplementation in iron-deficient populations • Peri-conceptional folic acid supplementation • Vitamin A in high-risk groups * Priority areas facility linkages to improve outcomes of high-risk pregnancies [121].

Research gaps (Table 17)
Of the 12 interventions studied in this paper, none showed clear evidence of benefit on stillbirths (Table 18), although several seemed promising. Given widespread undernutrition in low-and middle-income countries, interventions to address maternal macronutrient and micronutrient deficiencies must receive priority, as many have substantial benefits for maternal health and nutrition outcomes. If these are scaled up in health systems and population settings, attempts must be made to document stillbirths as a specific outcome. Some interventions, including birth spacing and peri-conceptual folic acid and iron-folate supplementation, benefit maternal health and prevent perinatal morbidity, but their impact on stillbirths has not yet been conclusively measured. Other interventions like balanced protein-energy supplementation may reduce stillbirth rates, but further large-scale effectiveness trials are needed.
The underlying patho-physiological mechanisms are clear for only some of the causes of antenatal stillbirths; significant gaps in understanding remain for many others, particularly those with socio-cultural, behavioural, and/or psychological components, including substance use/ abuse, elevated exposures to harmful substances shaped by socioeconomic disparities, and how emotional consequences of an adverse pregnancy outcome shape fertility intentions and subsequent outcomes. Many other socially mediated risk factors likely have yet to be identified. A better understanding of these mechanisms could lead to formulation of new interventions, health communication campaigns, and medical treatments to prevent stillbirths.

Conclusion
Complex social behaviours, including maternal diet, exposure to harmful substances, and care-seeking before and during pregnancy, influence maternal health and fetal outcomes. A number of strategies, particularly nutritional supplementation strategies, have been employed to improve maternal and fetal outcomes, but the evidence base for an impact on stillbirths is weak. There is a need for interventions to be developed and tested in community settings for micro-and macronutrient deficiencies, including anemia and folic acid deficiency. Large trials sufficiently powered to detect stillbirths are especially needed to adequately assess the effectiveness of interventions in preventing antenatal stillbirths. Other socially mediated behaviours, including smoking, exposure to indoor air pollutants, the practice of FGM, and shortened inter-pregnancy intervals after an adverse pregnancy outcome, appear to present an increased risk of stillbirth. Effective smoking cessation interventions exist, but inter-ventions are needed to address these other risk factors. Culturally appropriate behavioural interventions alone or in combination with pharmacotherapy, should be done studying impact on stillbirths as well. However, more safety trials are needed for agents like nicotine replacement therapy and bupropion. Because smokeless tobacco use has effect on stillbirths at least as much as maternal smoking, interventions to reduce this form of tobacco use should also be developed, especially in low-income countries. These numerous research gaps warrant studies to identify interventions that alter behaviours and remedy nutritional deficiencies that place women at risk of stillbirth. ANC provides an important platform for distributing these interventions, including opportunities to educate mothers about behavioural changes they can practice while pregnant to safeguard their and their future children's health.