Iron status and risk factors of iron deficiency among pregnant women in Singapore: a cross-sectional study

Iron deficiency is the most prevalent global nutrient deficiency and the most common cause of anaemia worldwide [1, 2]. Iron deficiency represents a spectrum ranging from iron depletion without anaemia (reduced iron stores with a normal haemoglobin (Hb) concentration) to eventual overt anaemia, where the iron supply is insufficient to maintain a normal Hb concentration [3]. Pregnant women are particularly vulnerable to iron deficiency due to substantial increase of iron requirement during pregnancy to support the expansion of erythrocyte mass and plasma volume, and foetal-placental growth [1, 4]. The World Health Organization (WHO) estimates that at least 30–40% of pregnant women are iron deficient and that nearly half are anaemic [5].

For pregnancy, the European Food Safety Authority [6] and the UK Committee on Medical Aspects of Food Policy [7] recommend no increase in iron intake over that for non-pregnant women. The extra iron requirements during pregnancy are considered to be met through cessation of menstrual losses, increased intestinal absorption and mobilisation of maternal iron stores [8]. However, a large proportion of pre-pregnant women or those of reproductive age have low iron stores [9, 10], predisposing them to an increased risk of iron deficiency when becoming pregnant [8]. In Singapore, a developed country, one in five non-pregnant women of reproductive age were found to be anaemic [11].

Either anaemic or non-anaemic iron deficiency prior to and during pregnancy can have adverse consequences for both the mother and offspring, especially with respect to neonatal iron-deficient condition [5, 12]. It was previously thought that neonate was protected from iron deficiency as the developing fetus could acquire sufficient iron from the mother even when she was iron deficient [13]. However, it is now documented that neonatal iron stores can be compromised when the mother is iron deficient or anaemic [13]. Studies on rhesus macaques suggest that infants born to iron deficient mothers before pregnancy had low iron stores after birth [14]. In humans too, newborns from iron deficient mothers at delivery were found to have low iron stores, indicating there is a limited capacity for the fetus to accumulate iron from mothers with low stores [15]. Identifying risk factors for maternal iron deficiency may therefore be helpful in developing preventive strategies to improve offspring health.

Multiple biomarkers have been used to measure iron status. Of these, plasma (or serum) ferritin is the most clinically applicable in pregnancy and has been proposed as the most sensitive single screening test for iron stores [3]. The plasma ferritin threshold used to define iron deficiency in pregnancy varies [16]. According to the World Health Organization [5], plasma ferritin levels of lower than 15 μg/L indicate iron depletion (loss of iron stores) at all stages of pregnancy [3]. This threshold also confirms the presence of iron deficiency anaemia [17]. The Committee for Standards in Haematology from United Kingdom recommends iron supplementation in pregnancy for plasma ferritin less than 30 μg/L [3], a threshold also widely used in clinical practice, including in Singapore, to guide therapy for iron deficiency in pregnancy [16]. This threshold has higher specificity than the lower threshold or when compared with other pregnancy biomarkers [18].

Although iron deficiency anaemia in pregnancy has been a frequent focus of research, few studies have investigated iron deficiency per se in pregnancy [10], leading to uncertainty about its clinical and public health significance. This is particularly true for women living in relatively affluent settings. In this study in highly developed Singapore, we aimed 1) to examine the proportion of iron deficiency in women during the early third trimester of pregnancy, and 2) to assess risk factors associated with iron deficiency in pregnancy.

Of 1152 recruited pregnant women who conceived naturally, 1090 (94.6%) of them returned for a study visit at 26–28 weeks’ gestation. A total of 977 (89.6%) women had sufficient plasma for analysis of sTfR concentrations; while 985 (90.4%) women had sufficient plasma for analysis of ferritin concentrations and were included in the present analyses (Fig. 1). Compared to excluded women, those included were more likely to have age greater than 35 years (21.9 versus 13.1%; P < 0.001) and attained university education (33.7 versus 22.8%; P = 0.003). They were, however, similar in every other factor assessed, including BMI status, ethnicity, parity, smoking status, iron-containing supplementation and history of anaemia (see Additional file 1: Table S1).

Table 1 shows the demographics and characteristics of women categorized by iron status. Overall (n = 985), 73.8% (n = 727) of pregnant women were iron deficient (ferritin < 30 μg/L). Among these women, the majority had modest iron depletion (ferritin 15 to < 30 μg/L, n = 660; 67.0%), while 6.8% (n = 67) had severe iron depletion (ferritin < 15 μg/L). Compared to women with iron sufficiency (ferritin ≥30 μg/L), those with modest iron depletion were more likely to belong to the Malay (27.6 versus 21.7%; P < 0.001) or Indian (21.8 versus 13.6%; P < 0.001) ethnic groups, to attain university education (36.5 versus 29.5%; P = 0.043), to be multiparous (60.0 versus 51.2%; P = 0.013), less likely to have taken iron-containing supplements during pregnancy (85.5 versus 93.4%; P = 0.001) and had higher plasma sTfR concentration (median 16.7 versus 15.5 nmol/L; P < 0.001). Out of 422 pregnant women with available information on Hb in early pregnancy, 81 (19.2%) had anaemia, 91% of whom had iron deficiency (modest and severe iron depletion) at 26–28 weeks. Among non-anaemic women in early pregnancy (n = 341), 255 (74.8%) were iron deficient at 26–28 weeks. A significant correlation was observed between Hb in early pregnancy and plasma ferritin at 26–28 weeks’ gestation (r = 0.22, p < 0.001).

Table 2 shows the results of the univariable and multivariable ordinal logistic regression analyses. In the multivariable model, which was of good fit, maternal age < 25 years (OR 2.36; 95% CI 1.15, 4.84), Malay (OR 2.05; 95% CI 1.30, 3.24) and Indian (OR 1.98; 95% CI 1.14, 3.44) ethnicities, university qualification (OR 1.64; 95% CI 1.13, 2.38), multiparity (OR 1.73; 95% CI 1.23, 2.44) and lack of iron-containing supplementation during pregnancy (OR 3.37; 95% CI 1.25, 8.53) were associated with a significantly increased odds of modest and severe iron depletion. In the sensitivity analysis based on women with complete data (n = 871), the results remained similar (see Additional file 2: Table S2).

During pregnancy, iron status tends to decline with advancing gestation, which can be due to iron mobilization or to haemodilution that peaks at 24–26 weeks [12]. Therefore, different cut-offs to define iron deficiency across gestation have been suggested [10]. Using stained bone marrow aspirates as the gold standard to evaluate iron status among pregnant women, most of those with an absence of stainable bone marrow haemosiderin iron have plasma ferritin < 30 μg/L in the second and third trimesters of pregnancy [18]. In the present study, at 26–28 weeks of gestation, only a quarter of pregnant women had sufficient iron stores, as defined by a plasma ferritin of at least 30 μg/L. The majority of women (74%) in our developed country was found to be iron deficient (ferritin < 30 μg/L), in which 67% had modest iron depletion (ferritin 15 to < 30 μg/L) and 7% had severe iron depletion (ferritin < 15 μg/L). Women with age < 25 years, Malay or Indian ethnicity, attained university qualification, multiparity and those not using iron-containing supplements during pregnancy were more likely to be iron depleted.

The prevalence of iron deficiency in pregnant women varies worldwide [28‐31], depending on the definition used, study population, gestational age at assessment and assay method. Based on a plasma ferritin threshold of < 30 μg/L, the proportion of iron deficiency among women at 26–28 weeks in this study (74%) was higher than that reported among pregnant women in Portugal at early pregnancy (38%) [31], but lower than that in pregnant women in Scotland at late pregnancy (90%) [29]. This is consistent with the notion that ferritin concentrations decline with the progression of pregnancy [32, 33]. These variations may simply reflect normal pregnancy physiology, rather than differences among populations. Though the proportion of our study women with severe iron depletion (ferritin < 15 μg/L; 7%) was much lower than rates reported among pregnant women in other developed countries [10], a high proportion of our women had modest iron depletion (ferritin 15 to < 30 μg/L). This high number warrants further investigation for subsequent biochemical and clinical consequences among both the mother and offspring.

We found that one in five women reported having been anaemic in the first trimester, similar to the prevalence of anaemia (22.2%) in women of reproductive age in Singapore [11]. The vast majority of these women (91%) were found to be iron deficient at 26–28 weeks despite reported iron-containing supplementation (72%). Since nearly 80% of anaemia in the Singapore pregnant population is due to iron deficiency [26], it is likely that most anaemic women in our study already had iron deficiency anaemia in early pregnancy, and even prior to conception. In view that anaemia is a late manifestation of iron deficiency, our findings are not surprising. Of interest is that almost just as high a proportion (75%) of non-anaemic women in early pregnancy was also found subsequently to be iron deficient at 26–28 weeks. Evaluation of mean corpuscular volume (MCV) and mean corpuscular haemoglobin (MCH) together with ferritin may be required in addition to Hb levels in the screening of women at the beginning of pregnancy to better detect iron deficiency and initiate appropriate iron treatment. Assessment of thalassemia, estimated at 9% in this population [34], should be considered during the screening procedure as the disorder might lead to abnormalities in the levels of red blood cell indices and biomarkers for iron status. Meanwhile, the role of prophylactic iron supplementation before the onset of iron deficiency anaemia remains controversial, given the reported associations of universal iron supplementation with development of gestational diabetes [35] and pregnancy-induced hypertension [36].

Similar to a local Singapore study in non-pregnant women [37], Malay and Indian pregnant women were more likely to have iron deficiency than Chinese pregnant women. Differences in dietary practice and/ or iron absorption ability may contribute to the higher proportions of iron deficiency in Malays and Indians [37]. Among the ethnic groups, Chinese women have been reported to consume more meat, poultry and egg, contributing to a rich source of haem-iron with a higher bioavailability; Malay women consumed fewer fruits and vegetables and had a lower vitamin C intake, which could reduce iron absorption; while Indian women tended to be vegetarian where the absorption of non-haem might be inhibited by phytates present in vegetables and cereals [37, 38]. Younger age may reflect poorer nutrition with reduced dietary iron intake, while multiparity may reflect depleted iron supply with increasing number of pregnancies [28, 32]. The finding of a positive association between education level and iron deficiency is supported by previous reports [27, 39], suggesting that knowledge, understanding or awareness of iron deficiency does not translate into action to increase iron-rich food intake or the use of iron supplements. Despite a high reported use of iron-containing supplements among our pregnant women (88%), a large proportion of them had iron deficiency at 26–28 weeks. The frequency and dosage consumed and compliance with iron-containing supplement intake were not recorded, moreover, which have been reported to influence iron status [10]. Otherwise, the absorption of iron from these iron-containing tablets is probably low due to the absorptive interaction of iron with other divalent metal ions in the tablets (e.g. zinc, manganese, calcium) [40].

We acknowledge several limitations. Variability in plasma volume expansion can affect the interpretation of biomarker levels, including levels of plasma ferritin, particularly late in pregnancy. The best time to detect maternal iron deficiency has been suggested to be in early pregnancy, before the plasma volume is fully expanded. As we did not measure plasma ferritin in early pregnancy, the present findings should therefore be interpreted cautiously. However, the use of a plasma ferritin cut-off of < 30 μg/L to define iron deficiency in the late second trimester is supported by van den Broek and colleagues [18], using bone marrow iron as a reference among pregnant women at mid and late pregnancy. Notwithstanding this, our findings probably include a mix of both physiologic plasma volume expansion and true iron deficiency. Multiple indicators of iron status such as transferrin saturation and hepcidin could be collectively used for better determination of iron status at later stages of pregnancy. Data on iron-containing supplementation was self-reported and therefore subject to recall errors. Data on dietary factors that may affect iron absorption were not available in this study. Measurement of Hb level at 26–28 weeks’ gestation is not part of routine clinical assessment in this setting, so data on maternal anaemia status was not available at the same time point as the plasma ferritin measurement. However, the use of a plasma ferritin cut-off of < 15 μg/L has been documented as an indicator for iron deficiency anaemia [17]. The proportions of women who reported smoking and a history of anaemia were small (≤5%) and insufficient to provide reliable conclusions regarding these groups. Some differences in characteristics (i.e., age and education) were noted between included and excluded women, which might have introduced selection bias. Finally, the lack of data on maternal inflammatory biomarkers (e.g. C-reactive protein and α-1-acid glycoprotein) is another limitation. Though sTfR has previously thought to be less affected by inflammation, a recent study found that sTfR has a weak but consistent relation with α-1-acid glycoprotein concentrations [41]. Ferritin concentrations are widely recognised to increase in the presence of inflammation [16] and pregnancy has been associated with low-grade inflammation [42]; thus the detection rate of iron deficiency in this study may be underestimated with no adjustment for inflammatory markers [43].

In conclusion, a substantial proportion of Asian pregnant women were found to be iron deficient at 26–28 weeks in Singapore, highlighting the potential importance of routine monitoring and screening for iron deficiency at several time points during pregnancy for timely commencement of iron treatment. Concerted efforts, including routine dietary advice (e.g. consuming plenty of iron-rich foods with a higher iron bioavailability, along with items containing vitamin C) and individual iron supplementation prophylaxis before and during pregnancy should be considered in this population for optimal maternal and offspring health. However, more research is needed to determine the appropriate use and dose of iron supplements for differing degrees of iron deficiency in pregnancy. Although the treatment of iron deficiency anaemia is not disputed, the maternal and offspring benefits of iron supplementation in iron deficiency with the absence of anaemia require rigorous evaluation in randomized trials.