- Top of page
Objective To develop a model of the impact of population-wide periconceptional folate supplementation on neural tube defects and twin births.
Design A hypothetical cohort of 100,000 pregnancies ≥20 weeks, plus terminations of pregnancy after prenatal diagnosis before 20 weeks.
Methods Application of pooled data on the relative risks for neural tube defects and twins following periconceptional folate from meta-analysis of the randomised trials.
Main outcome measures 1. Pregnancies with a neural tube defect (i.e. terminations of pregnancy, perinatal deaths, and surviving infants); 2. twin births (i.e. preterm births, perinatal deaths, postneonatal deaths, birth defects, cerebral palsy); 3. numbers needed to treat.
Results The change in neural tube defects would be 75 fewer terminations (95% CI -47, -90), 30 fewer perinatal deaths (95% CI 18, -35), and 13 fewer surviving infants with a neural tube defect (95% CI –8, -16). The change in twinning would be an additional 572 twin confinements (95% CI –100, +1587), among whom there would be 63 very preterm twin confinements (95% CI –11, +174), 54 perinatal and postneonatal deaths (95% CI –9, +149), 48 surviving twins with a birth defect (95% CI –8, +133), and nine with cerebral palsy (95% CI –2, +26). The numbers needed to treat for the prevention of one pregnancy with a neural tube defect is 847, for the birth of one additional set of twins is 175, for the birth of one additional set of very preterm twins is 1587, and for the birth of an additional twin with any of the following outcomes (perinatal death, postneonatal death, survival with a birth defect, or survival with cerebral palsy) is 901.
Conclusions Monitoring rates of neural tube defects and twinning is essential as supplementation or fortification with folate is implemented.
- Top of page
In a secondary analysis of data collected from a randomised controlled trial of periconceptional vitamin supplementation for the primary prevention of neural tube defects in Hungary, Czeizel et al.1 reported in 1994 that 93 of the 2468 infants (3.8%) born to women in the multivitamin group of women were twins or triplets, compared with 2.7% (64 of 2378) in the trace element group (P=0.03). At the time of publication this finding was often dismissed. As the authors themselves pointed out, an increase in multiple births had not been an expected outcome of the trial and it represented a new hypothesis. There was also some confusion because 6.5% of women in the Hungarian trial had been treated with clomiphene to induce ovulation, although the increase in multiple births was the same 40% increase within the subgroup treated with clomiphene and the subgroup not exposed to clomiphene. The major criticism was that Czeizel's group had overstated the evidence by analysing births: the appropriate comparison is multiple pregnancies (1.9%vs 1.4%, P= 0.14) rather than births.
Recently, a systematic review of the randomised trials of periconceptional supplementation with folate or multivitamins or both2, updated with new information from the Medical Research Council Vitamin Study3, identified a consistent increase in the relative risk of twin pregnancy in the three trials where twinning data were collected, with no evidence of heterogeneity2. The pooled relative risk was 1.40, the 95% confidence interval 0.93, 2.11 (Table 1). As twins have worse health and development outcomes than singletons, any increase in twin births would have adverse consequences for parents and health services.
Table 1. Twinning in the RCTs of periconceptional folate/multivitamin supplementation.
|Trial||Folate||No folate||RR||(95% CI)|
|Czeizel 1994||46/2421||32/2346||1.39||(0.89, 2.18)|
|Kirke 1992||3/186||1/95||1.53||(0.16, 14.53)|
|MRC 1991||7/593||5/602a||1.42||(0.45, 4.44)|
|Combined|| || ||1.40||(0.93, 2.11)|
Thus the aim of this paper was to model the impact on neural tube defects and twins of population-wide, periconceptional folate supplementation in a hypothetical cohort of 100,000 pregnancies reaching at least 20 weeks of gestation and including pregnancies terminated after diagnosis of a neural tube defect prior to 20 weeks.
- Top of page
Table 2 shows the impact of periconceptional folate supplementation on neural tube defects in the hypothetical cohort to be a very large reduction. The pattern of outcomes — the distribution across pregnancy termination, perinatal death, and neonatal survivor — within this overall reduction has been changing rapidly in the past five years, with the increase in prenatal diagnosis of neural tube defects. Thus, Table 2 models the impact using the pattern of outcomes recorded for two recent years. In both of them the major impact is a large absolute reduction in terminations of pregnancy, with a much smaller impact on perinatal deaths, and a small impact on surviving infants with a neural tube defect.
Table 2. Neural tube defects (NTD) in a hypothetical cohort of 100,000 confinements.
| ||Before population supplementation|| ||Change after population supplementationa|
|Outcome distribution||1988-1992|| ||1997|| ||1998|| || ||1997|| || ||1998|| |
| ||n||%||n||%||n||%||n||(95% CI)||NNTb||n||(95% CI)||NNTb|
|Termination||59||37||96||60||102||64||-71||(-44, -84)||1408||-75||(-47, -90)||1333|
|Perinatal death||56||35||36||23||40||25||-27||(-17, -32)||3704||-30||(-18, -35)||3333|
|Surviving infants||45||28||28||18||18||11||-21||(-13, -25)||4762||-13||(-8, -16)||7692|
|Total||160||100||160||100||160||100||-119||(-74, -141)||840||-118||(-73, -141)||847|
The impact of population-wide folate supplementation on twinning, and through additional twin births on perinatal and child outcomes in the hypothetical cohort, is shown in Table 3. Given the baseline rate of twinning (14.3/1000) and the predicted increase with folate supplementation of 40% there would be an additional 5.7 twin gestations/1000 confinements; an additional 572 twin confinements in the hypothetical cohort. The implication of this increase in twin pregnancies is demonstrated by calculation of the additional preterm births, perinatal and infant deaths, and surviving infants with a birth defect or cerebral palsy. The right-hand column of Table 3 shows that the confidence intervals are wide, encompassing both a small reduction in these adverse outcomes and a very large increase. The outcomes in Table 3, unlike those in Table 2 are not mutually exclusive.
Table 3. Twins in a hypothetical cohort of 100,000 confinements.
|Perinatal and child outcome||Before population supplementation||Change after population supplementationa|
|Twin confinements ≥20 wk||1430||+572||(-100, +1587)||175|
|Extremely preterm 20-27 weeks (4.0%)||57||+23||(-4, +618)||4348c|
|Very preterm 28-31 weeks (7.0%)||10||+40||(-7, +111)||2500c|
|Moderately preterm 32-36 weeks (39%)||558||+223||(-39, +619)||448|
|Twin births ≥20 wk||2860||+1144||(-200, +3174)|| |
|Perinatal deaths (4.3%)||123||+49||(-9, +137)||2041d|
|Perinatal + postneonatal deaths (4.7%)||134||+54||(-9, +149)||1852|
|Surviving twin with birth defect (4.4%)||120||+48||(-8, +133)||2083d|
|Surviving twin with cerebral palsy (8.6/1000)||23||+9||(-2, +26)||11,111|
Within the hypothetical cohort the predicted reduction in perinatal deaths due to neural tube defects is approximately 30 and is within the range of predicted increases in perinatal deaths resulting from additional twin births (49, 95% CI -9 to +137)). The estimated reduction in the number of surviving infants with a neural tube defect is approximately 13 and is within the range of the estimated increase in the number of surviving twins with cerebral palsy (9, 95% CI -2 to +26). In addition, there is a predicted increase of 48 (95% CI -8 to +133) surviving twins with a birth defect.
The number of women in the population needing to receive periconceptional folate supplementation in order to prevent the birth or pregnancy termination of a single infant with a neural tube defect, calculated as the reciprocal of the difference in prevalence of neural tube defects at birth (including terminations following prenatal diagnosis before 20 weeks of gestation), is 845.
The number of women in the population needing to receive periconceptional folate supplementation, in order for there to be one additional twin pregnancy reaching 20 weeks of gestation, is 175. Given the variation in perinatal and child health outcome in twin births this number needs to be calculated for a range of different outcomes such as a set of preterm twin births (50% of all twins) where it is 350; a set of very preterm twin births (all very preterm twins are likely to need neonatal intensive care), where it is 1,587; or the composite outcome of perinatal death or cerebral palsy or a birth defect in a surviving twin, where it is 943.
- Top of page
Information on neural tube defects prior to any public health interventions to increase women's folate intake was calculated from the pooled Victorian and Western Australian data for 1988-1992. These states include 35% of all births in Australia. Both have established birth defects registers linked to births and infant deaths, including information on terminations of pregnancy before 20 weeks of gestation following prenatal diagnosis of a malformation. Rates of neural tube defects in the two states have been shown to be similar through the 1980s9. Neither state had a population-based serum screening program for these defects at that time. Promotion of folate supplementation did not begin until late 199210 in Western Australia while fortification of selected foods with added folate and associated publicity did not begin until 1996 in Australia. There is no systematic fortification of food components such as flour in Australia.
Evidence for an impact of folate on twinning does not reach conventional 95% statistical significance, since the confidence interval includes 1.0, but it is evidence for a worrying effect. There are several reasons to consider the findings carefully rather then dismissing them. The first is the consistency of the findings in all three trials, regardless of whether they were trials to prevent first neural tube defects or recurrent neural tube defects, and regardless of the geographic location of the studies2. The increases were also consistent regardless of the actual daily folate doses which were 0.8 mg1, 0.36mg12, and 4.0 mg13 in the three trials.
The second reason is that no more randomised trials of this intervention are likely to be carried out. No new evidence from randomised controlled trials will reduce the uncertainty. One planned trial in China was modified to become a population-wide intervention in two provinces when the results of the MRC Vitamin Study were made known11. Three of the four randomised trials were stopped early, one of them because the evidence on neural tube defect prevention had already crossed the predefined stopping boundary12. Another stopped recruitment because of a falling prevalence of neural tube defects in the formerly very high risk population and a reduction in births within the hospital's catchment area13. Even if both trials had continued to their planned sample size it can be calculated that the pooled data would still be inconclusive in relation to an increase in multiple births as the 95% confidence interval would have been 0.96 to 2.09.
The third reason is collateral evidence. In four out of five datasets taken from two case-control studies and one case series of major birth defects the odds of multivitamin supplementation were 30% to 60% higher in mothers who had multiple births compared with mothers of singletons. Such an increase was not found among women who began supplementation in the first month of pregnancy nor among those who began supplementation in the second or third months14.
Information from the intervention study in China11 will increase the available evidence substantially but the extremely low rate of dizygotic twinning in China may limit the extent to which findings in that study can be extrapolated to other countries with much higher rates.
The systematic review identified a relative increase in conceptions with folate supplementation but the effect was much smaller than the increase in twinning (RR= 1.04, 95% CI 1.01, 1.06), and it was not consistent across the three trials which reported conception rates2. There was also a consistent increase in miscarriages (RR= 1.14, 95% CI 0.98, 1.32)2. An increase in births with multiple congenital anomalies associated with periconceptional vitamin supplements has been reported in a recent Californian case-control study15. These findings raise the possibility that periconceptional supplementation may facilitate early fetal survival.
Thus a plausible mechanism for increased twinning is likely to be increased survival of multiple pregnancies rather than increased ovulation. It is likely that multiple pregnancies have greater micronutrient requirements: folate is used to increase litter size in some farmed animals16. A recent review of evidence about early pregnancy loss concluded that losses of one or more fetuses in a multiple gestation are so much commoner than losses of single conceptions that the pooled effect size in the meta-analysis (a 40% increase in twinning with supplementation) is quite plausible17.
Whether it is folate or another component of multivitamin supplements which is the active agent cannot be answered from the available data, since multivitamins were given with folate to all the women allocated to folate in the Hungarian trial, to half the women who received folate in the Irish trial and to half the women who received folate in the MRC Vitamin Study2.
Impact of Population-wide Supplementation
There are very great difficulties in monitoring and interpreting changes in the proportion of births that are twins. An increase in twinning has been reported from several countries in the 1990s, attributable to an increasing use of fertility treatments, to a shift in the maternal age at birth and to improved ascertainment of births close to registration boundaries18–20. Ascertainment of extremely preterm births differs across countries and those regions and countries which do not register fetal deaths before 28 weeks, or do not register a birth as a twin birth if one of the twins is stillborn, will have greater under-estimation of twins. The use of fertility treatments could increase or decline, as might the choices about the age at which to give birth. The extent of elective fetal reduction in multiple pregnancies remains unknown. Elective termination of twin pregnancies when twins are diagnosed early in pregnancy has been noted recently in some practice settings (L. Amir personal communication). These secular trends make it difficult to monitor an increase in twinning which might be associated with greater use of periconceptional folate supplementation at a population level.
Similarly, monitoring trends in neural tube defects is not straightforward. It is essential for the monitoring system to include terminations of pregnancy which are now more than 60% of all neural tube defect outcomes (Table 2). Many countries have data systems which do not capture pregnancies which result in a termination. A new factor in monitoring is the increased use of first trimester ultrasound screening for nuchal translucency and possible Down's Syndrome. Such screening carries the possibility that isolated neural tube defects, multiple malformations and other trisomies including a neural tube defect may be suspected at the time of first trimester screening with the possibility of early termination. Terminations of pregnancy in the first trimester are in general much more difficult for birth defect monitoring systems to capture.
Implications of an Increased Risk of Twinning for Individual Women
The benefits and (potential) risks of periconceptional folate could be presented to individual women so that they might make an informed choice taking into account their previously known risks of having a baby with a neural tube defect, their attitudes to prenatal diagnosis and termination of pregnancy, and their acceptance of an increased risk of twins. We know that women in assisted conception programs accept much higher risks of multiple birth (one in eight) than those identified in Table 2. One difficulty with this rational decision-making approach is the unpredictability of the perinatal outcome of any individual twin pregnancy, compared with the generally very adverse outcome of a pregnancy in which the fetus has a neural tube defect. Many twins have no significant complications. On the other hand there are other adverse outcomes of multiple pregnancy not included in Table 3. Multiple gestations are associated with a substantial increase in maternal medical and obstetric complications and even with a detectable increase in maternal mortality21. Twins are at increased risk of developmental problems especially those of speech and language22. The major drawbacks of this individualised solution as a public health strategy are the relative lack of success of campaigns to increase individual women's use of periconceptional folate23–25, and the high proportion of pregnancies which are unplanned26.
Public Health Implications
Difficulties with the effective promotion of increased folate use before pregnancy has led to an emphasis on the need for fortification of basic foods (e.g. breakfast cereals), or the addition of folate to basic food components (e.g. flour), or a combination of these strategies, as the only public health measures likely to have a major population impact. Arguments for food fortification have been strengthened by evidence from non-experimental studies that folate may reduce the risk of birth defects other than neural tube defects (e.g. limb reduction defects, clefts and some cardiac malformations) although the findings are much less consistent in relation to these anomalies than for neural tube defects27. Dietary folate may also reduce cardiovascular disease28 and be one of the factors in a high fruit and vegetable diet contributing to lower risks of cancer29, again based on non-experimental studies.
In contrast to those possible wider benefits of populationwide folate supplementation, the health, social and economic costs of multiple births are substantial (e.g. Table 3). Additional adverse outcomes, both common and very rare, relevant to individual decision-making but with social costs and consequences as well, are mentioned in the paragraph above.
Although supplementation with lower than recommended doses would also result in changes in neural tube defects it is possible that there is a threshold for the twinning effect which exceeds that needed for neural tube defect prevention. This is relevant to current policy debates focusing on the recommended level of folate supplementation through food component fortification which would be required to eliminate neural tube defects30. There are suggestions that greater reductions in neural tube defects may be occurring than would be predicted from the current levels of supplementation31. Simultaneous monitoring of neural tube defects and twins in all countries and regions with complete data on both outcomes is essential to see whether the impact on neural tube defects and twins predicted from the randomised trials occurs as supplementation, fortification of foods with folate, or both, is implemented.