Background
Low birth weight (LBW; birth weight < 2500 g) and preterm birth (PTB; <37 weeks gestation) are major contributors to infant mortality and chronic childhood morbidity. LBW is associated with >80% of infant deaths [
1]. Understanding factors that moderate or exacerbate the burden of LBW and PTB is, therefore, an important global health priority. Poor nutrition, anaemia and infectious diseases are common among pregnant women globally, particularly in resource-limited settings, and are important contributors to LBW. Iron deficiency (defined by the World Health Organisation [WHO] as ferritin < 15 μg/L [
2]) is a common feature of undernutrition and a major cause of anaemia and it contributes to an estimated global burden of 600,000 perinatal and 100,000 maternal deaths per year [
3]. Many regions with a high burden of undernutrition and anaemia are also endemic for malaria and approximately 125 million women living in malaria-endemic areas become pregnant each year [
4,
5]. Malaria contributes to anaemia by causing haemolysis and dyserythropoiesis and it is also a major cause of LBW and PTB [
6,
7].
Despite the high prevalence of undernutrition and anaemia among pregnant women in resource-limited and malaria-endemic settings, there is limited knowledge on the contribution of iron deficiency and other specific preventable deficiencies to LBW and poor pregnancy outcomes. To reduce the burden of anaemia and poor pregnancy outcomes, the WHO recommends that all pregnant women receive daily iron and folate supplements (30–60 mg of elemental iron plus 400 μg of folic acid), in addition to “measures to prevent, diagnose and treat malaria” in malaria-endemic areas [
8]. However, there are concerns regarding the value of antenatal iron supplementation in malaria-endemic areas due to reports of harmful interactions between iron and malaria [
9]. Moreover, the link between iron deficiency and poor pregnancy outcomes in malaria-endemic and tropical regions, and the value of iron supplementation on improving birth outcomes, is unclear.
Iron deficiency has been associated with reduced high-density
Plasmodium falciparum parasitaemia, associated clinical symptoms of malaria and reduced placental
P. falciparum infection in case–control and cross-sectional studies of pregnant women living in malaria-endemic areas [
10]. Iron supplements given to children in malaria-endemic areas have been associated with increased parasitaemia, risk of clinical malaria and risk of severe illness and death (not specific to malaria) in some controlled trials [
11,
12]. However, clinical trials have shown inconsistent associations between daily iron supplementation and risk of peripheral or placental
P. falciparum infection in Africa [
10,
13‐
15]. One study reported an increased risk of
P. vivax after supplementation in Asia [
10]. There is also no clear evidence for the association between iron supplementation and birth outcomes in pregnant women living in areas where they are at risk of malaria [
10,
13‐
16]. A recent trial in west African pregnant and non-pregnant women found an increased rate of adverse events among those receiving iron supplementation [
17]. These issues are compounded by a lack of understanding of how iron deficiency, anaemia and malaria may interact to influence birth outcomes. This knowledge is essential to the evidence base of iron supplementation programmes in malaria-endemic and resource-limited settings where there is a high burden of infectious diseases and poor nutrition.
The objectives of this study were to determine the association between iron deficiency and birth outcomes, and to quantify how malaria mediated these associations, in a longitudinal study of pregnant women in a malaria-endemic area of Papua New Guinea (PNG), which has the largest population at risk of anaemia, undernutrition malaria and poor birth outcomes in the South-West Pacific region.
Discussion
LBW and PTB are major risk factors for infant morbidity and mortality and identifying factors that contribute to these risks, or protect against them, is crucial for advancing child health globally. Furthermore, concerns regarding potentially harmful malaria–iron interactions have raised questions about the appropriateness of antenatal iron supplementation for anaemia in malaria-endemic populations [
9]. Here, we provide the first evidence of a protective association of maternal iron deficiency on LBW, and to a lesser extent PTB, in a malaria-endemic area where iron deficiency and anaemia are common, and elucidated the potential effect of malaria–iron interactions on birth outcomes. Iron deficiency was associated with substantially reduced risks of LBW, and this beneficial effect appeared largely mediated through mechanisms that are independent of an interaction between iron deficiency and malaria. The apparent protective effect of iron deficiency was greatest in primigravid women, a high-risk group who are more susceptible to, and more severely affected by, a range of infectious diseases [
19].
We demonstrated that the associations between iron deficiency and malaria infection were confounded by the presence of an acute phase response (raised CRP). Iron deficiency was associated with reduced malaria infection when the analysis included all women. However, when women who were potentially misclassified as iron replete (ferritin> 15 but raised CRP levels) were excluded, no protective association was found. Women with malaria are more likely to have raised CRP and therefore, more likely to be misclassified as iron replete. As a result, the analyses would more likely demonstrate an apparent protective effect of iron deficiency unless these effects are considered in analyses. Previously published analyses of cross-sectional and case–control studies of associations between iron deficiency and peripheral and placental
P. falciparum infection of women [
10] have included these potentially misclassified women, which may have biased results towards finding an association between iron deficiency and reduced malaria risk, or increased the magnitude of the protective effect. There is some biological plausibility to protective associations of iron deficiency, as in vitro studies have shown that iron deficiency can limit the development of blood-stage infection by starving the parasite of exoerythrocytic forms of iron [
20‐
23], impair merozoite invasion and propagation [
24], or increase the immune-mediated clearance of infected erythrocytes [
25,
26]. However, a direct causal mechanism between iron deficiency and placental malaria independent of the aforementioned reductions in peripheral infection is yet to be proposed and we found no association between iron deficiency and placental malaria in this study.
A strength of our study is that we used sequential mediation analysis as a valuable approach to estimate the contribution of potential causal relationships. This is the first analysis to quantify the mediation of the effect of iron deficiency on birth outcomes through malaria (both peripheral and placental) and anaemia. This is also the first mediation analysis of the effect of iron deficiency on birth outcomes to account for both exposure-outcome and mediator-outcome confounders, which is a source of bias in traditional mediation analysis [
27]. Importantly, we found that only 12% of the protective effect of iron deficiency on LBW was mediated through malaria or anaemia. Iron deficiency was not significantly associated with anaemia or peripheral or placental malaria, which are key determinants of LBW and PTB in this population [
18]. This may explain the lack of mediation through placental malaria in this cohort study and further studies using sequential mediation analyses are warranted in other populations. After accounting for mediation through peripheral
Plasmodium spp. infection and placental
P. falciparum infection, we found that iron deficiency was still associated directly with a >50% reduction in the odds of LBW. Host iron is essential to a range of other bacteria, parasites and viruses, and iron deficiency may also inhibit the development of these pathogens [
28‐
30]. Therefore, reductions in infectious pathogens due to iron deficiency may explain the association between iron deficiency and improved birth outcomes in our study. In PNG, there is a high burden of disease from bacterial and viral pathogens, including respiratory, gastrointestinal and sexually transmitted infections [
31], that may be influenced by iron status [
30]. In PNG, a higher prevalence of these infections is also found in primigravid compared to multigravid women [
32], which may also explain why larger magnitudes of the effect of iron deficiency on birth outcomes were observed in high risk primigravid women. Other infections may also be important. A recent study of iron supplementation in non-pregnant African women found higher rates of treatment for gastrointestinal infections in women receiving iron [
17]. Further studies are needed to understand the mechanisms by which iron deficiency leads to improved birth weight, and this knowledge would help identify appropriate strategies to reduce LBW while also reducing the burden of iron deficiency and anaemia. Examining the effects of iron deficiency on other infectious pathogens common in pregnancy, together with their interactions with malaria and anaemia, is a priority for future studies.
This study is the first longitudinal study globally to demonstrate a temporal relationship between antenatal iron deficiency and birth outcomes, and together with the sequential mediation analyses, provides significant causal evidence for the associations observed. In PNG, malaria transmission is high and both
P. falciparum and
P. vivax are endemic. The majority of infections (>95%) in our study were
P. falciparum and therefore, our findings may be more broadly applicable to other high
P. falciparum transmission areas, such as sub-Saharan Africa. However further studies are required to determine the consistency of our findings and the effect size across areas of varying prevalence of iron deficiency, malaria and other infections and in areas where the relative contribution of the multiple factors that cause iron deficiency differs. Research that identifies populations or settings where iron deficiency associations with birth weight are strongest would be valuable in guiding public health policy. In our study, we experienced a ~40% loss to follow-up; however, given that the characteristics of the women lost to follow-up did not differ from those of the included women (Additional file
1), the loss is unlikely to have affected the internal validity of the study. While the specific causes of iron deficiency have not been defined, these are likely to include low dietary intake as well as infections such as hookworm and malaria that can impact iron absorption or utilisation.
Iron deficiency was defined according to the WHO standard definition (ferritin < 15 μg/L for non-pregnant women) [
2]. Ferritin is also an acute phase protein that may increase during inflammation independently of iron stores [
33]. Previous cross-sectional and case–control studies of iron deficiency and malaria in pregnancy either did not measure inflammation or reported that a proportion of women had inflammation and their reported associations may have been confounded [
34,
35]. In our study, we performed sensitivity analyses to show that the protective associations observed between iron deficiency and birth weight were not confounded by the presence of inflammation. In the absence of bone marrow biopsies, ferritin remains the gold standard to assess iron deficiency [
36], and although other markers of iron deficiency have been explored, such as the soluble transferrin receptor, they currently lack established definitions for iron deficiency in pregnancy. Any potential bias associated with misclassification of true iron status with the use of ferritin for defining iron deficiency is unlikely to change the conclusions of this study given the magnitude and statistical strength of the observed associations and the biological plausibility.
Acknowledgements
We thank all the study participants, Francesca Baiwog and the staff of the Alexishafen Health Centre under the direction of Sister Valsi Kurian for their enthusiastic cooperation with the study, and the staff of the PNG Institute of Medical Research for assistance with microscopy, data entry and study administration.