Objective
The purpose of this study was to investigate the relationship between spina bifida and 2 established risk factors (pregestational diabetes mellitus and obesity) in both the presence and absence of the recommended daily folic acid intake in the periconceptional period.
Study Design
Cases of spina bifida (n = 1154) and control subjects (n = 9439) from the Slone Epidemiology Center Birth Defects Study (1976-2011) were included. Information on preexisting diabetes mellitus (collected 1976-2011) and obesity (collected 1993-2011), defined as a body mass index of ≥30 kg/m 2 , was collected through interviews that were conducted within 6 months of delivery. Periconceptional folic acid intake was calculated with both dietary and supplement information. Mothers were classified as consuming more or less than 400 μg/day of folic acid; food folate was included at a 30% discount for its lower bioavailability. Logistic regression models that were adjusted for maternal age, race, education, and study site were used to calculate adjusted odds ratios (aORs) and 95% confidence intervals (CIs) for the joint effects of low folic acid intake coupled with diabetes mellitus or obesity.
Results
Case mothers were more likely to have diabetes mellitus or be obese (0.7% and 19.0%, respectively) than control mothers (0.4% and 10.8%, respectively). The joint effect of diabetes mellitus and lower folic acid intake on spina bifida was larger (aOR, 3.95; 95% CI, 1.56–10.00) than that of diabetes mellitus and higher folic acid intake (aOR, 1.31; 95% CI, 0.17–10.30). Folic acid intake made little difference on the association between obesity and spina bifida.
Conclusion
Our findings suggest that folic acid further attenuates, although does not eliminate, the risk of spina bifida that is associated with diabetes mellitus than the risk with obesity.
Folic acid intake in the periconceptional period reduces the risk of spina bifida in offspring. Spina bifida, a neural tube defect, develops in the earliest part of pregnancy, frequently before recognition of pregnancy. Because of this early developmental timing and the fact that approximately one-half of all pregnancies in the United States are unplanned, the United States Public Health Service, the Institute of Medicine, and the United States Preventive Services Task Force recommend that women of child-bearing age consume at least 400 μg of folic acid daily through supplements and fortified food products. Despite these recommendations and mandatory fortification of enriched cereal grain products beginning in the United States in 1998, a recent study estimated that only 1 in 4 women of child-bearing age consumes the recommended amount of folic acid. Given the established role of folic acid in the reduction of the occurrence of spina bifida, it is possible that risk factors for spina bifida may operate differently in the presence and absence of folic acid.
Preexisting diabetes mellitus and obesity have both been identified as independent risk factors for spina bifida, yet few studies have addressed the impact of folic acid intake on these associations. In animal studies, folic acid supplementation has been shown to decrease the incidence of neural tube defects among diabetic pregnancies. The only epidemiologic study conducted to date observed that diabetic women who used folic acid-containing supplements had a lower risk of spina bifida than diabetic women who did not use supplements. Similarities in the physiologic condition of diabetes mellitus and obesity, which includes abnormalities in glucose metabolism, might suggest that folic acid may also affect the association between obesity and spina bifida. A study of prepregnancy weight and neural tube defects observed that folic acid offered some protection among lighter women (<70 kg), but no such protection among heavier women.
With recent studies suggesting that, in the era of folic acid fortification, achieving higher levels of folic acid intake through supplement use and dietary folate may not reduce the risk of neural tube defects, it is important to understand whether higher levels of folic acid intake alter the effect of other risk factors for spina bifida. Using data from the Slone Epidemiology Center Birth Defects Study, we sought to describe the effect of obesity and diabetes mellitus on the risk of spina bifida in both the presence and absence of the recommended amount of daily folic acid intake.
Materials and Methods
Study population
The Slone Epidemiology Center Birth Defects Study is an on-going case-control study in the United States and Canada that began in 1976. Cases of birth defects are ascertained from birth hospitals, tertiary care centers, or birth defects registries in Massachusetts; Philadelphia, PA; San Diego, CA (since 2001); Toronto, Canada (1976-2005); selected counties in Iowa (1983-1985); and parts of New York State (since 2004). Cases include live births, fetal deaths (since 1990), and elective terminations (since 1990). Control subjects are liveborn infants selected from study hospitals and birth certificates of the catchment area from which cases were collected. Before 1993, control subjects consisted of infants with minor malformations, such as heart murmurs and skin tags. Beginning in 1993, nonmalformed infants were used. The study has been approved by the Boston University Institutional Review Board and is in compliance with the Health Insurance Portability and Accountability Act.
A maternal interview is conducted by a trained nurse within 6 months of delivery among subjects who provided informed consent. The interview includes questions on reproductive history, medication use, and illnesses during pregnancy. Beginning in 1988, data on diet were collected with the use of a food frequency questionnaire (FFQ).
The present study includes cases of spina bifida and control subjects who were ascertained from 1976-2011. Spina bifida cases were excluded in the presence of chromosomal anomalies, Mendelian inherited disorders, recognized syndromes, amniotic bands, body wall defects, or conjoined twins. Cases were reviewed by a clinical geneticist and classified into isolated and multiple defect categories. Cases were considered multiple if another major defect, unrelated to spina bifida, was present.
Diabetes mellitus and obesity
Information on diabetes mellitus, both preexisting and gestational, was collected throughout all years of the study. Preexisting diabetes mellitus was defined as the onset of type 1 or type 2 diabetes mellitus before pregnancy. Gestational diabetes mellitus was defined as the onset of diabetes mellitus after the first lunar month of the index pregnancy. Body mass index (BMI) was calculated based on the mother’s self-reported height and prepregnancy weight. Data on BMI were available from 1993 onwards when information on maternal height was incorporated into the interview. BMI was categorized into 4 groups: underweight (<18.5 kg/m 2 ), normal-weight (18.5-24.9 kg/m 2 ), overweight (25-29.9 kg/m 2 ), and obese ( ≥30 kg/m 2 ).
Folic acid intake assessment
Folic acid intake in the periconceptional period, defined as the month before the last menstrual period through the first lunar month of pregnancy, was calculated by summing average daily folic acid intake from supplements and fortified foods. Natural folate from diet was also included but was discounted by 30% of the women because of its lower bioavailability. Data on dietary intake were collected from an adapted Willett FFQ. The FFQ asked about diet in the 6 months before pregnancy to best capture nutritional intake in the periconceptional period. Dietary values of folic acid and folate were adjusted for total energy intake with the use of the residual method. Subjects were categorized into 2 groups, those who achieved the recommended intake of ≥400 μg/day of folic acid (high intake) and those with an intake of <400 μg/day (low intake). Subjects with extreme caloric intake (<500 or >4000 kcal/day) and incomplete (≥3 missing items) FFQs were excluded from analyses that involved folic acid intake; however, 2 exceptions were made. Women who reported ≥400 μg of daily folic acid from vitamin supplementation remained in the analysis, because regardless of dietary intake they would be included in the ≥400 μg/day category. Second, women who reported no supplementation remained in the analysis because, regardless of dietary intake, it is unlikely they would reach ≥400 μg/day of folic acid. Women with missing information on diet and an intake of <400 μg/day from supplementation were excluded.
Statistical analyses
Logistic regression models were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs). We calculated crude and adjusted ORs (aORs) for the associations between spina bifida and preexisting diabetes mellitus and gestational diabetes mellitus, separately. To assess the joint effect of low folic acid intake on the association between preexisting diabetes mellitus and spina bifida, subjects were categorized into 4 unique groups; (1) no diabetes mellitus and ≥400 μg/day folic acid (referent), (2) diabetes mellitus and ≥400 μg/day of folic acid, (3) no diabetes mellitus and <400 μg/day of folic acid, and (4) diabetes mellitus and <400 μg/day of folic acid. The relative excess risk due to interaction (RERI) and 95% CI between diabetes mellitus and low folic acid intake were calculated. Women with gestational diabetes mellitus were excluded from the analysis of joint effect.
Analyses in which BMI was assessed were restricted to birth years 1993-2011 and excluded women with preexisting diabetes mellitus. We calculated ORs for underweight, overweight, and obese groups. To assess the joint effect of low folic acid intake and obesity, subjects were categorized into 4 unique groups; (1) normal-weight and ≥400 μg/day of folic acid (referent), (2) obese and ≥400 μg/day of folic acid, (3) normal-weight and <400 μg/day of folic acid, and (4) obese and <400 μg/day of folic acid. The RERI and 95% CI were calculated. BMI data were missing for 2.5% of participants, and missing BMI did not appear to be independent of other analytic variables. Therefore, we conducted a sensitivity analysis in which multiple imputation methods were used to impute BMI; the impact of using these imputed values to estimate the association between BMI and spina bifida was assessed. Because of changes in case and control ascertainment over the study period, a sensitivity analysis to address the possibility of selection bias was performed by restricting cases and control subjects to those from the same birth hospitals. Maternal age, education, race, study site, and folic acid intake (<400 μg/day, ≥400 μg/day) were included in regression models to adjust for potential confounding. Joint effect models did not include folic acid intake as a covariate. Analyses were performed using SAS software (version 9.3; SAS Institute Inc, Cary, NC).
Results
Diabetes mellitus
A total of 1154 cases of spina bifida and 9439 control subjects were ascertained from 1976-2011 and were included in the diabetes mellitus analysis. Cases and control subjects differed by race/ethnicity, maternal age, education, and folic acid intake ( Table 1 ). Among case mothers, the prevalence of preexisting diabetes mellitus was 0.69% compared with 0.44% among control mothers. The aOR for preexisting diabetes mellitus and spina bifida was 1.84 (95% CI, 0.80–4.22); the aOR for gestational diabetes mellitus was 1.19 (95% CI, 0.84–1.71; Table 2 ). Spina bifida in the presence of other defects had an aOR of 2.56 (95% CI, 0.59–11.13) for preexisting diabetes mellitus; the aOR for isolated spina bifida was 1.70 (95% CI, 0.68–4.29; data not shown).
| Variable | 1976-2011 a | 1993-2011 b | ||||||
|---|---|---|---|---|---|---|---|---|
| Spina bifida cases | Control subjects c | Spina bifida cases | Control subjects | |||||
| n | % | n | % | n | % | n | % | |
| Total | 1154 | 9439 | 389 | 8062 | ||||
| Maternal race/ethnicity | ||||||||
| White, non-Hispanic | 991 | 85.9 | 7112 | 75.3 | 279 | 71.7 | 5816 | 72.1 |
| Black, non-Hispanic | 58 | 5.0 | 628 | 6.7 | 30 | 7.7 | 580 | 7.2 |
| Other races | 105 | 9.1 | 1693 | 17.9 | 80 | 20.6 | 1660 | 20.6 |
| Hispanic d | 73 | 6.3 | 1095 | 11.6 | 57 | 14.7 | 1079 | 13.4 |
| Missing data | 0 | 0.0 | 6 | 0.1 | 0 | 0.0 | 6 | 0.1 |
| Maternal age at conception, y | ||||||||
| <20 | 92 | 8.0 | 594 | 6.3 | 24 | 6.2 | 535 | 6.6 |
| 20-24 | 263 | 22.8 | 1441 | 15.3 | 57 | 14.7 | 1122 | 13.9 |
| 25-29 | 390 | 33.8 | 2687 | 28.5 | 136 | 35.0 | 2143 | 26.6 |
| 30-34 | 296 | 25.6 | 3125 | 33.1 | 109 | 28.0 | 2767 | 34.3 |
| 35-39 | 94 | 8.1 | 1371 | 14.5 | 53 | 13.6 | 1287 | 16.0 |
| ≥40 | 19 | 1.6 | 200 | 2.1 | 10 | 2.6 | 188 | 2.3 |
| Missing data | 0 | 0.0 | 21 | 0.2 | 0 | 0.0 | 20 | 0.2 |
| Maternal education, y | ||||||||
| <12 | 178 | 15.4 | 854 | 9.0 | 63 | 16.2 | 701 | 8.7 |
| 12 | 418 | 36.2 | 1964 | 20.8 | 104 | 26.7 | 1479 | 18.3 |
| >12 | 557 | 48.3 | 6614 | 70.1 | 222 | 57.1 | 5877 | 72.9 |
| Missing data | 1 | 0.1 | 7 | 0.1 | 0 | 0.0 | 5 | 0.1 |
| Study center | ||||||||
| Boston (1976-2011) | 300 | 26.0 | 4491 | 47.6 | 84 | 21.6 | 4121 | 51.1 |
| Philadelphia (1976-2011) | 417 | 36.1 | 2173 | 23.0 | 137 | 35.2 | 1581 | 19.6 |
| Toronto (1979-2005) | 359 | 31.1 | 1236 | 13.1 | 106 | 27.2 | 960 | 11.9 |
| Iowa (1983-1985) | 15 | 1.3 | 132 | 1.4 | — | — | — | — |
| San Diego (2001-2011) | 33 | 2.9 | 1007 | 10.7 | 32 | 8.2 | 1002 | 12.4 |
| New York (2004-2011) | 30 | 2.6 | 400 | 4.2 | 30 | 7.7 | 398 | 4.9 |
| Daily folic acid intake, μg | ||||||||
| <400 | 832 | 72.1 | 5023 | 53.2 | 225 | 57.8 | 4005 | 49.7 |
| ≥400 | 259 | 22.4 | 3991 | 42.3 | 139 | 35.7 | 3797 | 47.1 |
| Missing data | 63 | 5.5 | 425 | 4.5 | 25 | 6.4 | 260 | 3.2 |
c From 1976-1993 cases with minor malformations were used as control subjects
| Diabetes mellitus | Cases (n = 1154) | Control subjects (n = 9439) | Crude OR (95% CI) | Adjusted OR a (95% CI) | ||
|---|---|---|---|---|---|---|
| n | % | n | % | |||
| Preexisting | 8 | 0.69 | 42 | 0.44 | 1.56 (0.73–3.32) | 1.84 (0.80–4.22) |
| Gestational | 39 | 3.38 | 371 | 3.93 | 0.86 (0.61–1.20) | 1.19 (0.84–1.71) |
| None | 1107 | 95.93 | 9026 | 95.62 | 1.0 (Reference) | 1.0 (Reference) |
a Adjusted for maternal age, education, race/ethnicity, folic acid intake, and study center.
The joint effect of low folic acid intake and preexisting diabetes mellitus resulted in 4-fold increased risk for spina bifida (aOR, 3.95; 95% CI, 1.56–10.00) relative to mothers without diabetes mellitus and with higher daily folic acid intake, which was greater than expected given the individual additive effects of low folic acid intake and preexisting diabetes mellitus (RERI, 1.65; 95% CI, −2.87 to 6.18). In the presence of ≥400 μg/day of folic acid, this association was attenuated (aOR, 1.31; 95% CI, 0.17–10.30), although this estimate was based on 1 exposed case ( Table 3 ).
| Preexisting diabetes mellitus | Folic acid <400 μg | Cases/control subjects | Adjusted OR a (95% CI) |
|---|---|---|---|
| No | No | 242/3827 | 1.0 (Reference) |
| No | Yes | 803/4798 | 1.99 (1.69–2.34) |
| Yes | No | 1/15 | 1.31 (0.17–10.30) |
| Yes | Yes | 7/24 | 3.95 (1.56–10.00) |
a Adjusted for maternal age, education, race/ethnicity, and study center.
Sensitivity analyses restricted to cases and control subjects from the same birth hospitals changed diabetes mellitus results modestly, but the RERI still indicated synergy between diabetes mellitus and low folic acid intake (data not shown).
Obesity
Restrictions to the years in which data on BMI were available and exclusion of women with preexisting diabetes mellitus yielded a total of 389 cases and 8062 control subjects in the BMI analysis. During this time period, cases and control subjects differed by maternal age, education, and folic acid intake ( Table 1 ). Among case mothers, 19.0% were obese compared with 10.8% of control mothers. The aOR for spina bifida among overweight women was 1.24 (95% CI, 0.93–1.63); the aOR for obese women was 1.97 (95% CI, 1.46–2.65; Table 4 ). Overall, 4.37% of cases and 2.44% of control subjects were missing data on BMI; Hispanic mothers were more likely to have missing BMI data (15.8% of cases and 9.4% of control subjects; data not shown). Multiple imputation of BMI did not materially change the results ( Table 4 ).
