Twelve studies were included. These comprised 6 observational studies that evaluated the association between risk of twinning and FA supplementation (n=1,491,728, possibly an overestimate since 2 studies reported data from the Swedish Medical Birth Registry and covered overlapping years) and 6 retrospective studies that evaluated the association with FA fortification of food (the maximum possible n was 7,807,364; the review stated there was some overlap among studies).
Five of the 6 studies either stratified or adjusted for maternal age. Two studies used adequate methods to account for ART. One study reported an underreporting of ART that led to classification bias. Two studies reported whether twins were of the same or different sexes.
Methodological problems included variation in the use of FA among different populations (range: 0.6 to 53%), the use of FA assessed retrospectively with the potential for bias, and small sample size. The reviewers stated that 3 of the 6 studies had major methodological flaws.
The reviewers stated that the 'best risk estimate' of the risk of twinning came from 1 study (n=176,042). It reported that FA supplementation (0.2 or 0.4 mg/day) was associated with no statistically significant increase in the risk of overall twinning (adjusted odds ratio, OR 1.02, 95% confidence interval, CI: 0.85, 1.24). There was no significant association between FA and same sex twins, classified as monozygotic (adjusted OR 0.70, 95% CI: 0.35, 1.40), or unlike-sex twins, classified as dizygotic (adjusted OR 1.26, 95% CI: 0.91, 1.73). The analyses were adjusted for age, parity and underreporting of FA and in vitro fertilisation use.
Fortification of foods with FA.
Studies conducted in the USA used different definitions for FA exposure during the optional fortification period (1996 to 1998). There was overlap in the populations among these studies. The reviewers stated that none of the studies was able to ascertain the comparability of FA exposure in women delivering twins compared with women delivering singletons.
One study reported that stratification by maternal age had no effect. In 2 studies, the analysis was adjusted for maternal age. Four studies did not account for ART. The 2 most recent U.S. studies took account of ART and induction of ovulation. Only 1 study accounted for zygosity.
One U.S. study reported no increase in the risk of twins associated with FA (relative risk 1.00, 95% CI: 0.95, 1.04). Three U.S. studies reported a maximum annual increase in the risk of twinning of between 2.4% and 4.6% across all ages. One U.S. study reported a maximal percentage annual increase in twinning of 2% for women aged over 30 years and no increase for younger women. The other study had methodological flaws.