Maternal Multivitamin Use and Infant Cardiac Defects

Transcription

1 American Journal of Epidemiology Copyright by The Johns Hopkins University School of Hygiene and Public Health All rights reserved Vol., 9 Printed in U.S.A. ORIGINAL CONTRIBUTIONS Occurrence of Congenital Heart Defects in Relation to Maternal Multivitamin Use Lorenzo D. Botto, Joseph Mulinare, and J. David Erickson The purpose of this study was to assess the relation between maternal multivitamin use and risk for cardiac defects in the offspring, using a population-based approach. The Atlanta Birth Defects Case-Control study is a population-based case-control study of infants born between 968 and 98 to mothers residing in metropolitan Atlanta, Georgia. The 98 case infants with nonsyndromic cardiac defects were actively ascertained from multiple sources. The,9 infants without birth defects (control infants) were selected from birth certificates by stratified random sampling. Periconceptional multivitamin use, defined as reported regular use of multivitamins from months before pregnancy through the first months of pregnancy, was contrasted with no use during the same time period. Periconceptional multivitamin use was associated with a reduced risk for nonsyndromic cardiac defects in the offspring (odds ratio (OR) =.76; 9 confidence interval (Cl):.6,.97). The risk reduction was strongest for outflow tract defects (OR =.46; 9 Cl:.4,.86) and ventricular septal defects (OR =.6; 9 Cl:.8,.99). No risk reduction was evident when multivitamin use was begun after the first month of pregnancy. If these associations are causal, the results suggest that approximately one in four major cardiac defects could be prevented by periconceptional multivitamin use. Am J Epidemiol ^ : abnormalities; heart defects, congenital; heart septal defects, ventricular; vitamins Cardiac defects, which affect at least one in newborns, are the most common group of major congenital anomalies (-). Cardiac defects are also a major cause of infant mortality due to congenital anomalies (6), and they impose a considerable burden of personal suffering and societal cost (7). Overall, known risk factors account for only a small fraction of all cardiac defects. For some of these factors, such as rubella, effective primary prevention is available through timely vaccination (8). For others, such as maternal diabetes mellitus, interventions are more complex and the results less certain (9). A potential new avenue for the primary prevention of cardiac defects was suggested by the results of a randomized clinical trial which showed that maternal periconceptional use of supplemental multivitamins containing folic acid could reduce the occurrence of neural tube defects (). In that study, the risks for some other birth defects, including some types of heart defects, appeared to be decreased. Subsequently, Botto et al. () and Shaw et al. () presented data from population- Received for publication October 4,998, and accepted for publication May, 999. Abbreviations: Cl, confidence interval; OR, odds ratio. From the Birth Defects and Genetic Diseases Branch, National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, GA. Reprint requests to Dr. Lorenzo D. Botto, National Center for Environmental Health, Mailstop F-4, Centers for Disease Control and Prevention, 477 Buford Highway NE, Atlanta, GA 4. based case-control studies which showed that children of mothers who reported using multivitamins before and around the time of conception were at 4 percent less risk for a specific subset of cardiac anomalies, the outflow tract defects although a subsequent case-control study of multivitamin use () and a study that evaluated folate intake in the year before conception () did not report similar findings. Cardiac outflow tract defects, however, represent only percent of all cardiac defects (-). To assess the overall impact of multivitamin use on cardiac defects, we extended our analysis of a population-based case-control study to other major groups of cardiac defects. MATERIALS AND METHODS Study design We analyzed data from the Atlanta Birth Defects Case-Control Study. In that study, information was gathered in from parents of infants with birth defects (case infants) and parents of infants who did not have birth defects (control infants). Case infants were ascertained through the Metropolitan Atlanta Congenital Defects Program, a populationbased registry that ascertains cases of birth defects among babies born to mothers residing in the fivecounty metropolitan Atlanta, Georgia, area. The reg- 878

2 Maternal Multivitamin Use and Cardiac Defects 879 istry staff actively ascertains structural birth defects among stillborn and liveborn infants whose defects are diagnosed during the first year of life. Sources of information include vital records, medical records from birthing hospitals, pediatric and specialty wards, and cytogenetic laboratories. All infants with birth defects ascertained through the Metropolitan Atlanta Congenital Defects Program and born between January, 968, and December, 98, were eligible for entrance into the Atlanta Birth Defects Case- Control Study. The control group was a percent stratified random sample of babies without birth defects born during the same period who were frequencymatched to the case group by hospital of birth, calendar quarter of birth, and race. Details on the design and procedures of the Atlanta Birth Defects Case-Control Study have been published elsewhere (4, ). Case classification We selected as case subjects all liveborn and stillborn infants with a cardiac defect (International Classification of Diseases, Ninth Revision, Clinical Modification (6) codes ). On the basis of abstracted medical records, each case subject was assigned an anatomic diagnosis and an overall clinical diagnosis. To ensure that each case subject received a single anatomic diagnosis, we used the hierarchical classification () shown in table. Each case subject was also given one of three mutually exclusive clinical diagnoses: ) isolated, if the cardiac defect was the only major anomaly present; ) multiple, if the cardiac defect was part of a pattern of multiple congenital anomalies of unknown cause; or ) syndromic, if the cardiac defect was part of a genetic or teratogenic syndrome (e.g., trisomy, trisomy, congenital rubella). Syndromic cases were excluded from further analysis. Exposure assessment We obtained information on a variety of exposures, including vitamin use, through telephone interviews. The complete questionnaire used has been published elsewhere (). The classification of multivitamin exposure was based on the mothers' responses to three questions: ) "At any time during the period from months before the pregnancy began through the first months of the pregnancy, did you take any vitamins regularly, that is, at least three times a week?" ) "What type of vitamins did you take? That is, were they multivitamins or were they a single vitamin like vitamin C or vitamin A?" ) "In which months during this period did you take the vitamin(s)?" On the basis of their answers to these questions, we grouped mothers into the following mutually exclusive groups: ) nonusers, if they reported no regular use of multivitamins during the 6-month period from months before pregnancy through the third month of pregnancy; ) periconceptional users, if they reported regular multivitamin use during the entire 6-month period from months before pregnancy through the third month of pregnancy; ) early postconceptional users, if they reported regular multivitamin use from the first month of pregnancy through the third month of pregnancy; 4) late postconceptional users, if they reported regular multivitamin use from the second month of pregnancy through the third month of pregnancy; ) other users, if they reported other or discontinuous patterns of multivitamin use; and 6) unknown/uncodable. Regular multivitamin use was defined as taking a multivitamin at least three times per week. The major comparison groups of interest in this study consisted of mothers who used multivitamins throughout the sensitive period of heart development (group above) and mothers who did not use multivitamins during that same period (group above). Relative risk estimation We examined contingency tables for the main exposure variables and covariates to look for confounding by single factors and for interaction. We then constructed logistic regression models to estimate the aggregate contribution of potential confounders. Potential confounders included maternal education, race, age, smoking, alcohol use, diabetes, parity, hospital of birth, period of birth, chronic diseases, and fever (table ). Because of differences in the proportions of case and control infants by race, diabetes, and period of birth, we present risk estimates adjusted for these three variables. For defect groups with less than cases, we provide unadjusted risk estimates. We used SAS software (SAS Institute, Cary, North Carolina) to estimate relative risks and 9 percent confidence intervals. When a cell was empty (see table 4), we used StatXact software (Cytel Software Corporation, Cambridge, Massachusetts) to calculate exact confidence intervals. RESULTS The overall participation rate in the Atlanta Birth Defects Case-Control Study was 7 percent (,9/ 4,46) for mothers of control infants and 69 percent (4,99/7,) for mothers of infants with birth defects. Of the infants with birth defects,,49 had a cardiac defect. The participation rate was 68 percent among mothers of children with a cardiac defect (7 percent among White mothers, percent among mothers of Am J Epidemiol Vol., 9,

3 88 Botto et al. other races). Of the,49 cases of cardiac defects, 9 (9 percent) were diagnosed as a clinically recognized syndrome or a condition with a known cause (genetic or teratogenic) and were excluded from further analysis. The remaining 98 nonsyndromic cases formed the basis of this report. The distribution of defects by cardiac diagnosis and clinical diagnosis is shown in table. As expected, the most frequent specific diagnoses were septal defects, obstructive defects, and outflow tract defects. When we compared demographic characteristics and pregnancy histories of the mothers of case and control infants, we found that, although they were similar with regard to most factors, they differed with respect to mother's race, history of chronic illness, and period of infant's birth (table ). Specifically, compared with mothers of case infants, mothers of control infants were more likely to be White and to have delivered in the earlier years of the study, and less likely to have had a chronic disorder around the time of conception. The pattern of use of multivitamin supplements among case and control mothers is summarized in table. Of the control mothers, 4. percent reported using multivitamins regularly during the periconceptional period, while 8.9 percent reported no use during the same period. We had only limited information on the specific brand or composition of the multivitamin supplements used by the women in the study, so we were unable to evaluate the effect of specific supplements or their individual components. Overall, risk estimates from the stratified analyses and logistic regression models were quite similar to those of the unadjusted analyses, suggesting that there was little individual confounding or aggregate confounding. Table 4 shows the estimated risk of having a baby with a cardiac defect among periconceptional multivitamin users compared with nonusers. Periconceptional multivitamin use was associated with a 4 percent apparent reduction in risk for all heart defects combined (odds ratio (OR) =.76; 9 percent confidence interval (CI):.6,.97). This reduction was driven mainly by a statistically significant reduction in outflow tract defects and ventricular septal defects (4 percent and 9 percent reductions, respectively). We also compared mothers who reported beginning multivitamin use during the second or third month of pregnancy, presumably after the formation of the cardiac structures in the embryonic heart (late postconceptional users), with mothers who did not take multivitamins. Among these women, multivitamin use was not associated with a decreased risk for TABLE. Classification of cardiac defects in the Atlanta Birth Defects Case-Control Study, Diagnoses L-TGA4: and heterotaxias Corrected (L) transposition Heterotaxia (Ivemark, situs variants) Outflow tract defects Tetralogy of Fallot Double-outlet right ventricle Truncus arteriosus Transposition of the great arteries Atrioventricular septal defect Anomalous pulmonic venous return Ebstein anomaly Right-side obstructive defects (pulmonic stenosis, pulmonary atresia with intact septum) Left-side obstructive defects Hypoplastic left heart Coarctation of aorta Aortic stenosis Septal defects Ventricular septal defects Atrial septal defects Other cardiac defects Isolated defect/ multiple defects', Syndromest Overall diagnosis (no.ji Total Total 98 9,49 * Presence of additional noncardiac congenital anomalies, not part of a syndrome of known cause, t Presence of a syndrome of known cause (genetic, chromosomal, or teratogenic). $ L-TGA, levo-transposition of the great arteries. Am J Epidemiol Vol., 9,

4 Maternal Multivitamin Use and Cardiac Defects 88 TABLE. Characteristics of mothers and children in the Atlanta Birth Defects Case-Control Study, Status Characteristic Cases (n = 98) Controls (n =,9) Maternal race* White Other Maternal age (years) at delivery! > Period of child's birth* January 968-March 97 April 97^July 976 August 976-December 98 Maternal education^ Grammar school High school College Chronic illness, including diabetes mellitus*, Cigarette smoking during pregnancy}: Alcohol use during pregnancy^ , , ,,6 497, , *p<.. t There were three missing responses. t There was one missing response. There were 7 missing responses. H There were missing responses. having a child with a heart defect, either overall (OR =.4; 9 percent CI:.87,.4) or among specific cardiac defect groups, including outflow tract defects (OR =.74; 9 percent CI:.49,.) and ventricular septal defects (OR =.99; 9 percent CI:.7,.9). We found no evidence of increased risk for any cardiac defect group in any vitamin use category. TABLE. Reported multivitamin use among mothers of case and control infants in the Atlanta Birth Defects Case-Control Study, Status Multivitamin use Cases (n = 98) Controls (n =,9) No vitamin use* Periconceptional usef Early postconceptional usef Late postconceptional use Other multivitamin usefl Unknown/uncodable# , * No regular use of multivitamins during the 6-month period from months before pregnancy through the third month of pregnancy. t Regular multivitamin use during the entire 6-month period from months before pregnancy through the third month of pregnancy. X Regular multivitamin use from the first month of pregnancy through the third month of pregnancy. Regular multivitamin use from the second month of pregnancy through the third month of pregnancy. ) Reported other or discontinuous patterns of multivitamin use. # Unknown or uncodable pattern of multivitamin use. Am J Epidemiol Vol., 9,

5 88 Botto et al. TABLE 4. Estimated relative risk for nonsyndromic heart defects associated with periconceptional multivitamin use as compared with no use in the Atlanta Birth Defects Case-Control Study, Diagnoses Use No use Odds ratio* 9 confidence interval Controls All heart defects L-TGAif and heterotaxias Corrected (L) transposition Heterotaxia (Ivemark, situs variants) Outflow tract defects Tetralogy of Fallot Double-outlet right ventricle Truncus arteriosus Transposition of the great arteries Atrioventricular septal defects Anomalous pulmonic venous return Ebstein anomaly Right-side obstructive defects (pulmonic stenosis, pulmonary atresia with intact septum) Left-side obstructive defects Hypoplastic left heart Coarctation of aorta Aortic stenosis Septal defects Ventricular septal defects Atrial septal defects 4 8 6, i.oot ,.97.,.66., 6.8., 7.4.4,.86.,.4., 8.79.,.8.,.98.,.66.,.9.4, 8..8,..4,.4.4,..,.9.4,.87.8,.94.8,.99.4,.79 * Odds ratios were adjusted for infant's period of birth, maternal race, and chronic diseases; odds ratios were unadjusted if there were less than cases, t Reference category. X L-TGA, levo-transposition of the great arteries. DISCUSSION We found that women who reported using multivitamin supplements in the periconceptional period were at significantly lower risk of having babies with congenital heart defects than were women who reported not using multivitamins. The apparent risk reduction was specific for some cardiac defects, but not all, and was strongest for outflow tract and ventricular septal defects. The timing of multivitamin use appeared to be critical, in that the apparent risk reduction was evident for use around the time of conception or early in the first month of pregnancy, but not for use starting during the second or third month of pregnancy. In interpreting these findings, one should take into account the limitations and strengths of the study. Because participation was incomplete, selection bias cannot be ruled out. Participation differed between races, and we adjusted for this factor in the analysis. The potential for differential recall or reporting between mothers of case and control infants is also of concern, particularly because of the long time between pregnancy and interview. Other studies in which exposures reported after the birth of a baby were compared with information obtained from prenatal obstetric records (7) or interviews conducted during the pregnancy (8, 9) showed no major differences in the direction and strength of the detected associations. Although this does not necessarily mean that recall bias does not exist, it suggests that its effect, if present, may be small. Even in the absence of systematic biases, other factors that are associated with multivitamin use, and not multivitamin use in itself, may account for the findings. For example, periconceptional multivitamin users may differ from nonusers in ways other than multivitamin use, such as in the levels of micronutrients they obtain from natural or fortified foods. We did explore the effect of and then control for several maternal and demographic factors on which we had some information. The results did not differ appreciably from those of the unadjusted analysis, which suggests that there was little individual or aggregate confounding. Our ability to examine the effects of specific vitamins was limited by the lack of data on brand or composition. Finally, these results may have arisen by chance alone. A major strength of the study is the fact that it was population-based. Case infants were actively ascertained through multiple sources from a geographically Am J Epidemiol Vol., 9,

6 Maternal Multivitamin Use and Cardiac Defects 88 defined residential population, and control infants were selected through a stratified random sample to accurately represent the population from which the case group had been derived. In addition, the comparability of reports from case and control mothers was enhanced by the use of a structured interview, which covered a wide range of potential exposures, not only vitamin use. We believe that this is the first population-based case-control study to have examined the association between multivitamin use and a wide spectrum of major cardiac defects (figure ). An overall reduced risk for cardiac defects was reported in a randomized clinical trial that was designed to study the effect of maternal periconceptional multivitamin use on the occurrence of neural tube defects (). In that study, the children of women who used a multivitamin supplement during the periconceptional period had a percent reduced risk for all heart defects combined, in comparison with women who used a trace-element supplement. Specifically,.4 percent (/,) of infants whose mothers were given the multivitamin supplement had a heart defect, compared with.9 percent (/,) in the trace-element supplement group, and the decrease appeared to involve mostly defects of the cardiac septa (/, vs. /,). The multivitamin supplement contained folic acid (.8 mg), vitamins A (4,-6, IU), B,, B, B 6, B, C ( mg), D, and E, minerals, and trace elements (copper, manganese, and zinc). The trace-element supplement contained the same three trace elements and a small quantity of vitamin C (7. mg). The small sample size of the clinical trial did not allow definite conclusions for other cardiac defect groups. Our results are consistent with the pattern of risk reduction seen in the clinical trial, although the overall risk reduction appeared to be lower in our study (4 percent) than in the randomized trial ( percent). Two populationbased case-control studies (, ), as well as a previous report from our group (), limited their analysis to one particular group of cardiac defects, the cardiac outflow tract defects, which account for approximately percent of all cardiac defects. One of these studies () found a protective effect associated with periconceptional multivitamin use, while the other (), which examined folic acid consumption before pregnancy, did not. A hospital-based case-control study () did not show a reduced risk of outflow tract and ventricular septal defects associated with multivitamin use. A relation between maternal vitamin consumption and cardiac defect occurrence has also been found in animal studies. In particular, folic acid-deficient rats have been reported to produce offspring with a range of cardiac defects, mostly ventricular septal defects and defects of the outflow tract and great vessels (, ). In addition, increased levels of homocysteine, which may be a consequence of folic acid deficiency or abnormalities of the metabolism of folic acid and vitamin B, were shown to induce cardiac defects and neural tube defects in chick embryos (). This last study, which shows how a teratogen can induce both cardiac and neural tube defects, is particularly intriguing, because two case-control studies, one from our group (, ) and another from the California Birth Defects Monitoring Program (, 4), have reported an association between maternal multivitamin use and a reduction in both neural tube defects and a subset of cardiac defects. Furthermore, in the first study, the risk All Heart Defects Hungary '96 (RCT) Atlanta (CC) Outflow Tract Defects Hungary '96 (RCT) California '9 (CC) Baltimore-Washington '97 (CC) U.S.-Canada '99 (CC) Atlanta (CC) Ventricular Septal Defects Hungary '96 (RCT) U.S.-Canada ^99 (CC) Atlanta (CC) Coarctation of aorta Atlanta (CC) Atrial Septal Defects Atlanta (CC),.. Estimated Relative Risk and 9 Confidence Interval (log scale) FIGURE. Maternal use of multivitamins/folic acid and risk of cardiac defects: summary of findings from reports published in 99-. Legend: Hungary '96, Hungarian randomized clinical trial (RCT) (); Atlanta, Atlanta Birth Defects Case-Control Study, populationbased (present study); California '9, California Birth Defects Monitoring Program, case-control study, population-based (); Baltimore- Washington '97, Baltimore-Washington Cardiac Infant Study, case-control study, population-based (); U.S.-Canada '99, US-Canada casecontrol study (Boston, Philadelphia, Toronto), hospital-based (). "All heart defects," all congenital heart defects combined. Am J Epidemiol Vol., 9,

7 884 Botto et al. reduction for neural tube defects depended on the timing of multivitamin use, in a manner similar to that which we describe for cardiac defects (). These similarities suggest that neural tube defects and some cardiac defects may be influenced by similar genetic or environmental factors. To date, genetic research has focused on neural tube defects, mainly because the protective effect of multivitamins and folic acid is more established and because researchers are uncovering growing evidence that polymorphisms of genes involved in the metabolism of folate and homocysteine may indeed play a role in modifying risk for neural tube defects (-). These findings, collectively, suggest that it may be fruitful to study such genetic variants in relation to cardiac defects, particularly outflow tract and ventricular septal defects. At the same time, further epidemiologic studies designed to confirm and extend our findings are warranted, because of the potential public health impact these findings may have. These studies might include an intervention or a randomized clinical trial for testing whether multivitamin use may be effective for primary prevention of at least some cardiac defects. REFERENCES. Ferencz C, Loffredo CA, Rubin JD, et al, eds. Perspectives in pediatric cardiology. Vol 4. Epidemiology of congenital heart disease the Baltimore-Washington Infant Study Mt. Kisco, NY: Futura Publishing Company, 99.. Pradat P. Epidemiology of major congenital heart defects in Sweden, J Epidemiol Community Health 99;46: -.. Fyler DC. Report of the New England Regional Infant Cardiac Program. Pediatrics 98;6(suppl):7^6. 4. Kidd SA, Lancaster PAL, McCredie RM. The incidence of congenital heart defects in the st year of life. J Paediatr Child Health 99;9: Hoffman HI. Congenital heart disease: incidence and inheritance. Pediatr Clin North Am 99;7: Gillum RF. Epidemiology of congenital heart disease in the United States. Am Heart J 994;7: Waitzman NJ, Scheffler RM, Romano PS. The cost of birth defects: estimates of the value of prevention. Lanham, MD: University Press of America, Inc, Cooper LZ, Preblud SR, Alford CA. Rubella. In: Remington JS, Klein JO, eds. Infectious diseases in the fetus and newborn infant. 4th ed. Philadelphia, PA: WB Saunders Company, 99: Shepard TH. Diabetes. In: Shepard TH. Catalog of teratogenic agents. 8th ed. Baltimore, MD: Johns Hopkins University Press, 99:^.. Czeizel AE. Reduction of urinary tract and cardiovascular defects by periconceptional multivitamin supplementation. Am J Med Genet 996,6: Botto LD, Khoury MJ, Mulinare J, et al. Periconceptional multivitamin use and the occurrence of conotruncal heart defects: results from a population-based, case-control study. Pediatrics 996;98:9-7.. Shaw GM, O'Malley CD, Wasserman CR, et al. Maternal periconceptional use of multivitamins and reduced risk for conotruncal heart defects and limb deficiencies among offspring. Am J Med Genet 99;9:6-4.. Scanlon KS, Ferencz C, Loffredo CA, et al. Preconceptional folate intake and malformations of the cardiac outflow tract. Baltimore-Washington Infant Study Group. Epidemiology 998,9: Erickson JD, Mulinare J, McClain PW, et al. Vietnam veterans' risk for fathering babies with birth defects. JAMA 984; :9-.. Erickson JD, Mulinare J, McClain PW, et al. Vietnam veterans' risk for fathering babies with birth defects. Atlanta, GA: Centers for Disease Control, Commission on Professional and Hospital Activities, Health Care Financing Administration. International classification of diseases, ninth revision, clinical modification. nd ed. Washington, DC: US GPO, 98. (DHHS publication no. (PHS) 8-6). 6. Werler MM, Pober BR, Nelson K, et al. Reporting accuracy among mothers of malformed and non-malformed infants. Am J Epidemiol 989;9: Feldman Y, Koren G, Mattice D, et al. Determinants of recall and recall bias in studying drug and chemical exposure in pregnancy. Teratology 989;4: McKenzie SG, Lippman A. An investigation of reporting bias in a case-control study of pregnancy outcome. Am J Epidemiol 989; 9: Monie IW, Nelson MM. Abnormalities of the pulmonary and other vessels in rat fetuses from maternal pteroylglutamic acid deficiency. Anat Rec 96; 47:97^.. Baird CD, Nelson MM, Monie IW, et al. Congenital cardiovascular anomalies induced by pteroylglutamic acid deficiency during gestation in the rat. Circ Res 94;: Rosenquist TH, Ratashak SA, Selhub J. Homocysteine induces congenital defects of the heart and neural tube: effect of folic acid. Proc Natl Acad Sci U S A 996;9:7-.. Mulinare J, Cordero JF, Erickson JD, et al. Periconceptional use of multivitamins and the occurrence of neural tube defects. JAMA 988;6:4-.. Shaw GM, Schaffer D, Velie EM, et al. Periconceptional vitamin use, dietary folate, and the occurrence of neural tube defects. Epidemiology 99;6: Ou CY, Stevenson RE, Brown VK, et al., Methylenetetrahydrofolate reductase genetic polymorphism as a risk factor for neural tube defects. Am J Med Genet 996;6:6-4.. van der Put NM, Steegers-Theunissen RP, Frosst P, et al. Mutated methylenetetrahydrofolate reductase as a risk factor for spina bifida. Lancet 99;46: van der Put NM, Gabreels F, Stevens EM, et al. A second common mutation in the methylenetetrahydrofolate reductase gene: an additional risk factor for spina bifida? Am J Hum Genet 998;6: Whitehead AS, Gallagher P, Mills JL, et al. A genetic defect in, methylenetetrahydrofolate reductase in neural tube defects. Q J Med 99;88: Ramsbottom D, Scott JM, Molloy A, et al. Are common mutations of cystathionine beta-synthase involved in the aetiology of neural tube defects? Clin Genet 997;: Botto LD, Mastroiacovo P. Exploring gene-gene interactions in the etiology of neural tube defects. Clin Genet 998;: Werler MW, Hayes C, Louik C, et al. Multivitamin use and risk of birth defects. Am J Epidemiol 999; :67-8. Am J Epidemiol Vol., 9,