Dr KKL Chan, Department of Obstetrics and Gynaecology, Queen Mary Hospital 102, Pokfulam Road, Hong Kong. Email email@example.com
Objective To test the hypothesis that iron supplement from early pregnancy would increase the risk of gestational diabetes mellitus (GDM).
Design Randomised placebo-controlled trial.
Setting A university teaching hospital in Hong Kong.
Population One thousand one hundred sixty-four women with singleton pregnancy at less than 16 weeks of gestation with haemoglobin (Hb) level between 8 and 14 g/dl and no pre-existing diabetes or haemoglobinopathies.
Methods Women were randomly allocated to receive 60 mg of iron supplement daily (n= 565) or placebo (n= 599). Oral glucose tolerance tests (OGTTs) were performed at 28 and 36 weeks. Women were followed up until delivery.
Outcome measures The primary outcome was development of GDM at 28 weeks. The secondary outcomes were 2-hour post-OGTT glucose levels, development of GDM at 36 weeks and delivery and infant outcomes.
Results There was no significant difference in the incidence of GDM in the iron supplement and placebo groups at 28 weeks (OR: 1.04, 95% confidence interval [CI]: 0.7–1.53 at 90% power) or 36 weeks. Maternal Hb and ferritin levels were higher in the iron supplement group at delivery (P < 0.001 and P= 0.003, respectively). Elective caesarean section rate was lower in the iron supplement group (OR: 0.58, 95% CI: 0.37–0.89). Infant birthweight was heavier (P= 0.001), and there were fewer small-for-gestational-age babies in the iron supplement group (OR: 0.46, 95% CI: 0.24–0.85).
Conclusion Iron supplement from early pregnancy does not increase the risk of GDM. It may have benefits in terms of pregnancy outcomes.
The need for prophylactic iron supplement during pregnancy is controversial. Iron requirement increases during pregnancy, and the body’s iron store is often insufficient to meet the demands.1 Iron deficiency anaemia was shown to be associated with a greater risk of neonatal morbidity such as preterm birth.2,3 On the other hand, it was argued that a high maternal haemoglobin (Hb) level from iron supplementation would reduce placental perfusion due to increase in blood viscosity and cause adverse pregnancy outcomes such as low birthweight, preterm births, pre-eclampsia and stillbirths.4–6 This is further complicated by increasing concerns about the association between iron status and type II diabetes mellitus (DM). The frequency of diabetes in women with hereditary haemochromatosis, an inherited iron overload syndrome, provided the initial suggestion that systemic iron overload could contribute to abnormal glucose metabolism.7 There was increasing evidence to show an association between elevated serum ferritin and development of type II DM in those without hereditary haemochromatosis. Cross-sectional studies suggested that raised serum ferritin levels were found in women with noninsulin dependent diabetes as well as in women with gestational diabetes mellitus (GDM), and it was reported to be associated with glycaemic control.8–12 These findings were supported by large-population-based studies where elevated ferritin levels were associated with increased risk of development of type 2 DM as well as GDM.13–15 Furthermore, data from population-based prospective cohort studies have shown that those with higher serum ferritin levels or iron intake were at increased risk of developing diabetes during a 10- to 11-year follow-up period.16,17 Iron accumulation can affect glucose metabolism in a number of ways. Iron promotes formation of hydroxyl radicals that can attack cell membranes and affect insulin synthesis and secretion in the pancreas and interferes with the insulin-extracting capacity of the liver. Peripherally, iron deposition in muscle decreases glucose uptake because of muscle damage. Furthermore, insulin stimulates cellular iron uptake, resulting in a vicious cycle, leading to insulin resistance and diabetes.18,19 Therefore, the observed relationship between increased ferritin and the incidence of DM can be possibly explained by increased insulin resistance as a result of increased iron load. However, apart from being a major iron storage protein, which provides an indirect estimate of body iron stores, ferritin is also an acute phase protein, which is increased in various acute or chronic inflammatory conditions. Inflammatory cytokines have been shown to induce ferritin synthesis.20 GDM is increasingly recognised as part of an inflammatory process, associated with increased serum C-reactive protein and interleukins.21,22 Elevated serum ferritin was found to be related to low-grade inflammation in pregnant women.15 It is uncertain whether higher ferritin levels in the development of diabetes reflect increased iron stores or inflammation. So far, studies on the association between ferritin and development of diabetes were mainly observational. It is difficult to elucidate whether the increase in serum ferritin is the cause or the effect of development of diabetes or GDM.
Since the pregnancy state itself lead to insulin resistance, it is important to establish the safety of iron supplement in pregnancy particularly when there are suggestions that increased iron load would further increase insulin resistance. In order to determine the relationship between ferritin level and development of diabetes, we conducted a randomised placebo-controlled trial to investigate whether iron supplementation from early pregnancy to increase the iron store would increase the risk of development of GDM at 28 weeks of gestation. This would help in the understanding of the causal relationship between increased iron store and diabetes as well as in providing some evidence for the safety of iron supplement in pregnancy in terms of development of GDM.
This study was conducted in a single regional hospital between April 2005 and March 2007. The protocol was approved by the Hong Kong University/Hospital Authority Hong Kong West Cluster’s Institutional Review Board. Eligible participants were women with singleton pregnancy who could understand either Chinese or English. Women who had existing diabetes or haemoglobinopathies or Hb levels lower than 8 g/dL or greater than 14 g/dL or had gestational age greater than 16 weeks at booking would be excluded. Women were recruited when they attended their first visit (the booking visit) for their antenatal care. Study procedures were explained by a designated research assistant and written information was also given. Informed written consent was obtained from each woman who wished to participate. Three millilitres of blood was collected in ethylenediaminetetraacetic acid tube for full blood count (FBC), including Hb, haematocrit, mean corpuscular volume (MCV) and white cell and platelet counts, and 5 ml of blood was taken in a lithium heparin gel tube and sent to the laboratory for later batched assay of serum ferritin concentration. Serum ferritin was measured by the Roche Tina-quant Ferritin assay, a latex-enhanced immunoturbidimetric assay automated on the Roche Modular System (Roche Diagnostics, Indianapolis, IN, USA). The assay was standardised against NIBSC (Potters Bar, UK) reagents for Ferritin (human spleen 80/578). The blood results were reviewed by the research assistant; women with MCV <80 would be considered as possibly having thalassaemia and were therefore excluded from the study. Those with Hb levels lower than 8 g/dL or greater than 14 g/dL were also excluded. Women were then randomised to receiving a 300 mg ferrous sulphate tablet (60 mg of elemental iron) daily or placebo tablet indistinguishable from the study medication, containing a combination of starch and lactose. Block randomisation by a computer was used. Generation of randomisation schedule was performed by a person independent of the recruitment, and the seed from which the randomisation schedule was generated was kept securely by the randomiser. Sealed opaque envelopes containing the randomised treatment allocation was prepared and kept by the research assistant prior to the start of patient recruitment. The participants but not the research assistants were blinded to the group assignment. The participants were not told of the results of the randomisation.
Sixteen weeks’ supply of tablets (112 tablets) was dispensed at this visit by the research assistant. A dietary survey for recording 7 consecutive days diet intake was given out to assess the baseline iron intake and other macronutrients including total caloric, carbohydrate, fat and protein from the women’s usual diet. Women were asked to return the survey by post. The total iron and other macronutrients intake for each day was estimated using the Nutritionists IV Diet Analysis Database (1995), together with a Taiwan nutrient database for Chinese food (www.doh.gov.tw/newdoh/90-org/org-3/database/Welcome.html). The mean daily intake in the 7-day survey was calculated. Both the dietary survey and the methods of estimation of the amount of intake had been validated and were used in previous studies.23,24
All women received standard antenatal care, which included a dating scan, biochemical screening for Down syndrome with the option of amniocentesis for definitive diagnosis and an 18–20 week anomaly scan. Those with low MCV at booking were further investigated for haemoglobinopathies. A 75-g oral glucose tolerance test (OGTT) was arranged shortly after the booking visit for those with risk factors such as advanced maternal age or family history of diabetes, etc., for development of GDM. The results were interpreted according to the World Health Organization criteria.25 Impaired glucose tolerance (OGGT 2-hour value ≥ 7.8< 11.1 mmol/l) and diabetes (OGTT 2-hour value ≥ 11.1 mmol/l) were both considered as GDM. Those who were diagnosed as having GDM at booking were excluded from the study.
A 75-g OGTT was performed for all women between 28–30 weeks. During this visit, blood sample was taken for FBC and iron status. Compliance to the study medication was checked by counting the number of tablets remaining. A woman who took 75% of the given tablets would be considered as having a compliance of 75%. An additional 112 tablets and a second dietary survey were given to each participant. Further iron supplements were discontinued in women who developed GDM and in women who had an Hb level of 14 g/dL or more. If women in the placebo group developed anaemia with Hb level of less than 8g/dL, they would be given supplement as clinically indicated.
A third blood sample for FBC and iron status was collected at 36 weeks. A further OGTT was performed for those who had not been diagnosed as having GDM. The tablets were again counted to check compliance, and a third dietary survey was conducted. No further study medications were given.
At delivery, both maternal and cord blood were taken for FBC and iron status. Cord arterial pH was checked to detect hypoxia. The baby and the placenta were weighed to the nearest 1 g with a digital electronic scale. Delivery details and neonatal outcome were recorded as part of standard care. Gestational age was based on a dating scan performed in early pregnancy. Preterm delivery was defined as delivery before 37 weeks of gestation, and small for gestation age (SGA) was defined as birthweight below the 10th percentile for gestational age.
Power calculation and statistical consideration
The incidence of GDM in the local population ranges from 10 to 15%. To detect an increase in the incidence of GDM from 10 to 15% with a power of 80% using a 0.05 level of significance, a sample size of 530 was required in each group. We conducted analysis on an intention-to-treat basis. All participant randomised were included in the analysis (n= 1164, 565 in the study group and 599 in the control group). Statistical analyses were performed using the software JMP for Windows (Release 6.0.2; Cary, NC, USA). Pearson’s chi-square test was performed to test for difference in proportions between the control and the treatment groups for a categorical variable, and the Student’s t test was used to compare the means of a continuous variable of the two independent groups. A P value <0.05 was considered to be statistically significant.
One thousand four hundred women were recruited. Two hundred thirty-six women were subsequently excluded because they did not meet the inclusion criteria, leaving 1164 women randomised, 565 in the iron supplement group and 599 in the placebo group (Figure 1). The background characteristics for the two groups are shown in Table 1. There were no significant differences in terms of age, parity, body mass index, risk factors for GDM and gestational age at booking between the two groups. There were also no significant differences in the baseline Hb, ferritin and ratio of transferrin to ferritin between the two groups. Three hundred thirty-five (55.9%) and 306 (54.2%) women in the placebo and iron supplement group, respectively, returned their completed baseline dietary survey questionnaire, and there were no significant difference between the baseline carbohydrate, protein, fat and iron intake for both groups.
Table 1. Background characteristics
Control (n= 599)
Study (n= 565)
BMI, body mass index; SD, standard deviation.There were no significant differences between the study and control groups.
At 28 weeks of gestation, 1051 (90.3%) women attended follow up. Compliance of study medication could not be assessed in 74 women because they did not bring back the tablets for counting. In 977 (83.9%) women for whom compliance could be assessed, the overall compliance was 54.4%, and this was not significantly different between the iron supplement and placebo groups. Three hundred eighty-one women returned their dietary surveys, and there were no significant difference in the dietary iron intake between the two groups. Information about any additional vitamin supplements could be obtained in 1019 (97%) women, and 452 (43%) women had taken additional supplements. The proportion of women who had taken additional supplement was not significantly different between the two groups. Seven women were given iron supplement for clinical indications in the control group. FBC and serum ferritin levels were obtained from 1042 (89.5%) and 959 (82.4%) women, respectively. The serum ferritin level and ratio of ferritin to transferrin were significantly higher in the supplement group than in the placebo group (both P values <0.001), but there was no significant difference in the Hb levels between the two groups. OGTTs were performed in 1042 (89.5%) women, 531 in the placebo group and 511 in the supplement group. Overall, 116 (11.1%) women developed GDM, 60 (11.3%) in the placebo group and 56 (11.0%) in the supplement group. The number of women in the supplement group who developed GDM at 28 weeks was not significantly different from that in the placebo group (OR: 1.04, 95% confidence interval [CI]: 0.7–1.53 at 90% power). There was also no significant difference in the 2-hour post-OGTT glucose level between the two groups (Table 2).
Table 2. Iron and gestational diabetic status in study and control groups at 28 weeks of gestation
BMI, body mass index; ns, not significant; SD, standard deviation.
No. women who had taken additional vitamin supplement, n (%)**
237 (45.6%) (n= 520)
215 (43.1%) (n= 499)
At 36 weeks, 694 (59.6%) women attended follow up. Compliance of study medication could be assessed in 473 (40.6%) women. Overall, the compliance was 63.2%, and this was not significantly different between the two groups. Dietary intake could be assessed in 145 (12.6%) women and again, there were no significant difference in the dietary iron intake between the two groups. FBC and serum ferritin levels were obtained in 491 (42.2%) and 490 (42.1%) women, respectively. The serum ferritin level in the placebo group remained the same compared with that obtained at 28 weeks, but for the study group, it was significantly higher than that at 28 weeks (70.6 versus 60.4 pmol/l, P < 0.001). Both ferritin and ferritin to transferrin ratio were significantly higher in the supplement group than in the placebo group (both P values are <0.001), and the Hb level was also significantly higher in the supplement group (P < 0.001). OGTT was performed in 492 (42.3%) women, 248 in the placebo group and 244 in the supplement group. An additional 33 (33/492, 6.7%) women developed GDM between 28 weeks and 36 weeks of gestation, 17 (6.9%) in the placebo group and 16 (6.6%) in the supplement group. The number of women in the supplement group who developed GDM at 36 weeks was not significantly different from that in the placebo group. There was also no significant difference in the 2-hour post-OGTT glucose level between the two groups (Table 3).
Table 3. Iron and gestational diabetic status in study and control groups at 36 weeks of gestation
BMI, body mass index; ns, not significant; SD, standard deviation.
Delivery data were available in 925 (79.4%) women, 468 in the placebo group and 457 in the supplement group (Table 4). Delivery data could not be obtained for the remaining 239 women because they had not delivered at our hospital and delivery information could not be traced. The proportion of women with delivery data did not differ significantly between the two groups. The Hb level was significantly higher in the supplement group than in the placebo group at delivery as well as on day 3 after delivery. The ferritin level was also significantly higher at delivery for the supplement group. There was a significantly fewer women in the supplement group who required elective caesarean section than in the placebo group (P= 0.014, OR: 0.58, 95% CI: 0.37–0.89). Neonatal outcomes were available in 862 (74.1%) infants (Table 5). There was no significant difference in the gestational age at delivery and percentage of preterm delivery between the two groups. However, for term infants, the birthweight was significantly higher in the supplement group than in the placebo group (P= 0.001), and there were significantly fewer SGA infants in the supplement group (OR: 0.46, 95% CI: 0.24–0.85). There were no significant differences in other outcomes including Apgar score at 1 and 5 minutes, arterial cord blood pH, and Hb and ferritin of cord blood. There were also no significant difference in the incidence of neonatal complications including birth trauma, neonatal jaundice, metabolic disorders and sepsis between the two groups. There were no major adverse effects from the study medication throughout the study period.
Table 4. Maternal delivery outcome
BMI, body mass index; IUGR, Intrauterine growth restriction; ns, not significant; SD, standard deviation.
In this randomised study, we have increased iron store by giving iron supplement in a dosage commonly prescribed in pregnancy for a minimum of 12 weeks (from 16 to 28 weeks) in the iron supplement group but not in the placebo group. This was reflected in the significantly higher ferritin levels in the supplement group than in the placebo group, at both 28 and 36 weeks. However, we did not find any increase in the incidence of GDM in the supplement group. This does not support the previous findings from observational studies suggesting a link between higher serum ferritin or iron intake and the development of diabetes. It may be argued that the increase in iron store in this study was not sufficient to cause any noticeable change in glucose metabolism, but the ferritin level in the study group was above that seen in women with GDM in previous reports where an association was noted between high ferritin levels and GDM.9,10,25 It is more likely that the damaging effects of high iron load to the various components of glucose metabolisms require a long-term exposure to a high iron environment. A relatively sudden increase in body iron store with iron supplementation within a few months as in this study may not lead to any adverse effects that can be detected with an OGTT. It is also possible that the lack of association between increased ferritin from iron supplement and development of GDM in this study reflects the situation where the increased ferritin in DM in previous observational studies was due to ferritin being an acute phase reactant and the increased ferritin seen in women with DM was a result of the inflammation associated with the disease. Furthermore, we attempted to detect a large difference in the incidence of GDM (from 10 to 15%) at 80% power. Iron supplement would need to have potent diabetogenic properties to achieve this, and this may be another possible cause of not finding a difference.
With administration of a dietary survey, we could obtain the baseline nutritional status for both groups and show that the intake for the two groups was similar throughout pregnancy. The iron intake in our study population is similar to that reported in a population of healthy women enrolled in the Nurses’ Health Study in the USA.16 We felt that it was unethical not to give iron supplement to women who developed clinically significant iron deficiency anaemia during pregnancy. Therefore, research assistants were not blinded so that women in the placebo group who had clinically relevant anaemia could be identified and iron supplement could be given for clinical reasons. We noted that the compliance to study medication was low but appeared in to be in similar range as other randomised trials on iron supplement in pregnancy.26 It may be argued that the low compliance might be due to the undesirable adverse effects of iron supplement, leading to reluctance to take the medication. However, similar compliance was seen in the control group as well, reflecting the general reluctance to any additional medication in pregnant women in fear of any possible adverse effects on the baby. The high default rates for follow up seen in this study, especially at 36 weeks reflected the local antenatal care system. Many women book at both a public and a private unit for antenatal care and will choose the unit for delivery near term and thus default the antenatal care at the other unit. This should not have a major impact on our results since our primary outcome was the development of GDM at 28 weeks and the OGTT could be obtained in about 90% of our participants at 28 weeks.
Although our primary outcome was the development of GDM at 28 weeks, we have continued to observe the pregnancy outcomes for our study population. Previous Cochrane systematic review summarising the evidence for either a beneficial or a harmful effect of iron supplementation on pregnancy outcome was inconclusive. The effect of iron supplementation in nonanaemic women on birthweight, SGA and preterm delivery remained controversial in recent randomised trials investigating the effect of iron supplementation and pregnancy outcome.26,27 Our findings of higher infant birthweight in iron supplement group agreed with Siega-Riz et al.26 where infants from iron supplement group was 108 g heavier but differed from the findings from Ziaei et al.27 where no significant difference in birthweight was found. A small difference in birthweight, although statistically significant, may not have much clinical implications. More importantly, we found that the proportion of SGA infants was lower in the supplement group. This is in contrast with the findings from Ziaei et al. where the number SGA was higher in the supplement group. The study population in Ziaei’s study had a baseline Hb of 14 g/dl, which differed from Siega-Riz’s and our population where the baseline Hb was 12.4 and 12.6 g/dl, respectively. The exact biological mechanisms for iron’s effects on fetal growth are still unclear. It had been proposed that iron deficiency anaemia may cause chronic hypoxia, which would lead to oxidative stress of the placenta and would increase release of stress response hormones such as corticotrophin-releasing hormone that may affect fetal growth.28 Although most of our women were not anaemic or iron deficient, the ideal level of maternal Hb or iron store for fetal growth is still unknown. It is possible that normal Hb or iron store for a nonpregnant woman would not be sufficient for optimal fetal growth. On the other hand, further iron supplement in women with high Hb levels might produce adverse effects due to reduction of placental perfusion as a result of haemoconcentration. Siega-Riz found a lower incidence of preterm delivery, although this is not confirmed by our findings. We also found a significantly lower proportion of women in the supplemented group requiring a caesarean section, although the difference mainly was confined to elective caesarean sections. It is difficult to elucidate the reason for the difference as we did not find any significant difference in the indications for elective caesarean sections between the two groups. Siega-Riz et al. did not report findings on mode of delivery, while Ziaei did not find any significant difference in caesarean section rate for obstetrics reasons.
In conclusion, our study demonstrated that iron supplement in pregnancy did not increase the risk of development of GDM, despite the previous association between iron load and development of DM in observational studies. In order to further investigate the relationship between iron load and development of DM, long-term iron supplement in the general population would be needed. However, unlike in pregnancy where iron supplementation is common and may have beneficial effects, long-term iron supplement in the nonanaemic, nonpregnant population is not a generally accepted practice, and randomised trials to investigate the possibility of development of DM with long-term iron supplement may not be ethical. Overall, iron supplementation did not cause any adverse pregnancy outcome, in terms of both maternal complications, particularly GDM, and infant outcome. Conversely, iron supplement in nonanaemic women with Hb <14 g/dl may have benefits in terms of lower incidence of SGA and lower caesarean section rate.
Contribution to authorship
K.K.L.C., B.C.P.C. and T.T.L. conceived and designed the study. K.K.L.C. and B.C.P.C. supervised the research assistants in data collection and overall smooth running of the project. S.T. contributed in the design of the project and supervised the biochemical analyses involved. K.F.L. performed the statistical analysis. T.T.L. provided advice for various problems encountered during the project. K.K.L.C contributed in the data analysis. K.K.L.C and T.T.L interpreted the data. K.K.L.C. drafted the article with contribution from K.F.L. and S.T. All authors assisted in revising the draft.
Details of ethics approval
This study was approved by the Hong Kong University/Hospital Authority Hong Kong West Cluster Institutional Review Board (Protocol no. UW 04-100 T/422) on 31 March 2004.
This project is funded by a grant from the Research Grant Council, Hong Kong.
We thank the Research Grant Council for supporting this study. We would also like to thank our research assistants Ivy Li, Lesley Lau, Theresa Cheung and Chung Him Wan for their hard work in data collection.
Chan and colleagues are to be congratulated for undertaking a sizeable randomised controlled trial on this important topic. However, when considering their conclusions, it is important to be aware that the strongest evidence in such a study always relates to the primary hypothesis; in this case, that iron supplementation increases the risk of carbohydrate intolerance during pregnancy. Their results support the null hypothesis that it has no effect. When considering their secondary findings, we need to remember that the trial was not designed to look at other endpoints such as birthweight. If it had been, then the trial should have included only women who could potentially benefit from iron supplementation, and the results should have been stratified by initial haemoglobin concentration. Previous studies have shown that iron supplementation increases haemoglobin concentration and ferritin levels, but Cochrane systematic reviews (as Chan et al. point out) have not found any evidence of a measurable improvement in any other outcome in women who are not anaemic at the beginning of pregnancy. Not only would such an improvement be intrinsically unlikely, another study in Hong Kong found that supplementing nonanaemic mothers so that their haemoglobin rose was associated with adverse effects such as an increase in the incidence of preterm birth, low birthweight, and neonatal asphyxia (Lao TT. Hum Reprod 2000;15:1843–8). These adverse effects may occur as a result of a high haemoglobin concentration increasing blood viscosity, thereby reducing placental blood flow. Recent publications in BJOG suggest that routine supplementation of nonanaemic women increases the risk of pre-eclampsia (Smith TG, Robbins PA. BJOG 2007;114:1581–2) and infection (Siassakos D, Manley K. BJOG 2007;114:1308) including malaria (Nweneka CV. BJOG 2007;114:1581). A study of birthweight in more than 115 000 women reported that haemoglobin concentrations above 11 g/dl are associated with significantly smaller babies (Steer P et al.BMJ 1995;310:489–91), and a study in more than 200 000 women reported a significantly higher perinatal mortality, with haemoglobin concentrations above 12 g/dl (Little MP et al.Am J Obstet Gynecol 2005;193:220–6). In 2007, Ziaei et al. in this journal (BJOG 2007;114:684–8) reported that the incidence of small-for-gestational-age babies was 15.4% in supplemented pregnancies versus 10.1% in nonsupplemented pregnancies. It should be noted that in the current study, there was no significant difference in the preterm birth rate, the difference in birthweight at term was less than 100 g, and in the preterm birth group, the babies of nonsupplemented mothers were actually 200 g heavier. It would be inappropriate
to accept the secondary outcomes in the current study at face value. However, they should be welcomed as useful additional data for future meta-analyses, especially if the raw data are available to allow stratification of outcome by initial haemoglobin concentration.
PJ Steer Emeritus Professor, Division of Surgery, Oncology, Reproductive Biology and Anaesthetics, Faculty of Medicine, Imperial College London, Academic Department of Obstetrics and Gynaecology, Chelsea and Westminster Hospital, London, UK