Maternal Illness, Including Fever,and Medication Use as Risk Factorsfor Neural Tube DefectsGARY M. SHAW,1* KAREN TODOROFF,1 ELLEN M. VELIE,2 AND EDWARD J. LAMMER3

1March of Dimes Birth Defects Foundation, California Birth Defects Monitoring Program, Emeryville, California 946082National Cancer Institute, Division of Cancer Prevention, Bethesda, Maryland 208923Department of Medical Genetics, Children’s Hospital, Oakland, California 94609

ABSTRACT We investigated if selected mater-nal illnesses or medications used during the periconcep-tional period increased risk of having neural tube defect(NTD)-affected pregnancies. We used a population-based case-control study of fetuses and liveborn infantswith NTDs among 1989–1991 California births. In-person interviews were conducted with mothers of 538(88% of eligible) NTD cases and 539 (88%) nonmal-formed controls, usually within 5 months of delivery. Amaternal fever or febrile illness episode in the firsttrimester was associated with an increased risk forhaving a NTD-affected pregnancy, odds ratio (OR) 51.91 (95% confidence interval, 1.35–2.72) for feverand OR 5 2.02 (1.20–3.43) for febrile illness. Riskestimates were not substantially altered after adjust-ment for maternal age, race/ethnicity, education, vita-min use, and body mass index. Other reported illnesseswere generally not associated with risks of 1.5 orgreater, or were too infrequent to adequately estimaterisk. An OR of 1.5 or greater was observed for maternaluse of guaifenesin, OR 5 2.04 (0.79–5.28), and an ORof 0.5 or less was observed for maternal use ofcalcium-containing medicines, OR 5 0.38 (0.14–1.03).Our findings are consistent with previous reports thatsuggested elevated NTD risks from maternal fever. Wecould not discriminate, however, whether the increasedrisks observed for maternal fever were indicative of acausal relation or due to reporting bias. Our findingssuggest that many of the illnesses common to reproduc-tive-aged women and the medications commonly usedto treat them during pregnancy, except, perhaps, forthose illnesses that are febrile-related, do not appear tosubstantially contribute to the occurrence of NTDs inthe population. Teratology 57:1–7, 1998.r 1998 Wiley-Liss, Inc.

Anencephaly and spina bifida, the phenotypes thatcomprise the vast majority of neural tube defects(NTDs), are common congenital anomalies that substan-tially contribute to childhood morbidity and infantmortality (Elwood et al., ’92). Currently, there is compel-ling evidence that multivitamin/folic acid intake bywomen in the periconceptional period reduces the riskfor having a NTD-affected pregnancy (Elwood et al., ’92;

Medical Research Council, ’91; Czeizel and Dudas, ’92;Shaw et al., ’95). However, even though nutri-tional factors appear to be an important con-tributor to the etiology of a large proportion of NTDsoccurring in the population, the etiologies for a sizableproportion of NTDs remain unknown. Several othermaternal factors have been implicated in the riskfor NTDs, including socioeconomic class, various envi-ronmental factors, genetic factors, and maternal ill-ness or medication use during pregnancy (Elwood et al.,’92).

Studies that have investigated a potential relationbetween maternal illnesses and risk for NTD-affectedpregnancies have primarily focused on risks associatedwith diabetes, epilepsy, influenza, fever, or febrile ill-ness, and have found inconsistent results (Elwood etal., ’92). Likely contributing to these inconsistent re-sults are the following: differing study designs (e.g.,case-control vs. cohort studies); small numbers of par-ticipants in some studies; nonspecifically defined tim-ing of illness relative to neural tube closure; nonspecifi-cally defined maternal illness (e.g., ‘‘acute,’’ ‘‘chronic’’);lumping of all central nervous system anomalies as onestudied case group rather than limiting cases to justNTDs; and, in some studies, failure to consider bothmaternal illnesses and the medications taken to treatthem.

We examined data from a large population-basedcase-control study to investigate if selected maternalillnesses, or medications used, during the periconcep-tional period increased the risk for NTD-affected preg-nancies. This study contributes both to the existingevidence concerning previously studied maternal ill-nesses and medications used, as well as investigatesmaternal illnesses and medications used that have notbeen previously studied as potential risk factors forNTDs.

Contract grant sponsor: California Department of Health Services.

*Correspondence to: Dr. Gary M. Shaw, California Birth DefectsMonitoring Program, 1900 Powell St., Suite 1050, Emeryville, CA94608.

Received 8 August 1997; Accepted 11 November 1997

TERATOLOGY 57:1–7 (1998)

r 1998 WILEY-LISS, INC.

SUBJECTS AND METHODS

For this case-control study, described elsewhere (Shawet al., ’95), infants or fetuses with a NTD (anencephaly,spina bifida cystica, craniorachischisis, or iniencephaly)were actively ascertained by the California Birth De-fects Monitoring Program (Croen et al., ’91). Medicalrecords including prenatal ultrasonography, were re-viewed at all hospitals and genetic clinics for infants orfetuses delivered in selected California counties, andwhose mother gave her residence as California. Eligiblewere singleton fetuses, including those electively termi-nated, and liveborn infants diagnosed with an NTDamong the cohort of 708,129 births (includes fetaldeaths) between June 1989 and May 1991. Ascertainedwere 653 singleton infants or fetuses with an NTD.Controls were randomly selected from each area hospi-tal in proportion to the hospital’s estimated contribu-tion to the total population of infants born alive in agiven month from June 1989 to May 1991. Eligible were644 singleton infants who were born without a report-able congenital anomaly (Croen et al., ’91) and whosemother was a California resident.

The study protocol was approved by the CaliforniaDepartment of Health Services institutional reviewboard. Interviews were conducted with mothers inEnglish (74%) or Spanish (26%), primarily face-to-face(95%). We excluded women who only spoke languagesother than English or Spanish (29 cases and 32 con-trols), as well as 12 women who had a previous NTD-affected pregnancy. Interviews were completed withmothers of 538 (87.8%) cases and of 539 (88.2%) con-trols, an average of 4.9 months for cases and 4.6 monthsfor controls after the actual or estimated date of termdelivery. Information was unavailable for about 6% ofcase and control mothers who refused to be inter-viewed, and from approximately 5% who could not belocated.

An interviewer presented each woman with an indi-vidualized calendar marked by 4 time periods to assisther in remembering exposures/events: 3 months beforeconception, 3 months after conception, and the subse-quent 2 trimesters of pregnancy. Date of conception wasestimated primarily from date of last menstrual period,and in some circumstances from prenatal ultrasoundresults or newborn physical examination. The average2-hour interview elicited information on medical, repro-ductive, and family history, as well as aspects aboutmaternal employment, hobbies, recreational drug use,and nutritional intake. Regarding maternal illness, foreach of the following: diabetes, asthma, blood clots,cancer or tumors, epilepsy or seizures, high bloodpressure, and thyroid condition, a woman was asked,‘‘Have you ever been told by a doctor that you had[ILLNESS]?’’ If she responded affirmatively, she wasfurther asked the date of diagnosis, whether she took amedication for the illness in the period 3 months beforeconception until the end of pregnancy, specifically whenshe took the medication, and for the name of themedication(s) used. For each of the following illnesses:

heatstroke or heat exhaustion, bladder infection, fever(degree of temperature increase not measured or asked),vaginal or genital infection, a woman was specificallyasked, ‘‘During the time periods marked on the calen-dar did you have [ILLNESS]?’’ If she responded affirma-tively, she was further asked for the specific calendarperiod(s) she had the illness, whether she took medica-tion for the illness, the calendar period(s) she took themedication, and the name of the medication used.Following these inquiries, a woman was asked a gen-eral question about whether she contacted a healthpractitioner about any other injury or medical problem,before conception or during pregnancy, and was probedsimilarly to the above inquiries regarding timing ofillness and medication used. Using the collected illnessinformation, the following were considered ‘‘febrile’’illnesses: bronchopneumonia, pneumonia, influenza,common cold, bronchitis, otitis media, tonsillitis, strep-tococcal pharyngitis, abscessed tooth, rubella vaginitis,pelvic inflammatory disease, hyperthyroidism, heatexhaustion, herpes zoster, amebiasis, hepatitis, skininfection, giardiasis, and third-degree burns. This defi-nition is similar to that others have used to definefebrile illness (Layde et al., ’80; Shiota ’82). Women whoonly had bladder infections were considered unknownfor this febrile illness category. Unlike for fever andother illnesses studied, a woman was not specificallyasked whether she had a ‘‘febrile’’ illness. The assign-ment to a febrile illness category was made by one of us(EJL) blinded to case or control status.

In addition to the medication information elicitedfrom women who reported specific illnesses, womenwere also asked more generally, ‘‘Did you take anymedicines which you haven’t yet told me about? Thisincludes over-the-counter medicines as well as thosepurchased with a prescription or borrowed from friendsor family members?’’ If she responded affirmatively, shewas further asked details about the name of the medi-cine and the specific calendar period of use.

All prescription and over-the-counter medicines usedby study women were coded and matched to a medicinelinkage system that includes individual and multicom-ponent medicinal agents. This system was created bythe Slone Epidemiology Unit, Boston University Schoolof Medicine (personal communication, Tom Kelley ’96).This matching permitted the grouping of medicationsinto coalitions that reflected similar pharmacologicaction, class, chemical structure, or indication.

Risks were estimated by the odds ratio (OR), and theprecision of the OR assessed by its 95% confidenceinterval. The risk of an NTD-affected pregnancy wasestimated for each maternal illness relative to thosepersons without that particular illness. Risk estima-tions involving specific medication use for a maternalillness were made relative to those persons withoutthat particular illness and who also did not use thatparticular medication. Risk estimations involving spe-cific medication coalitions were made relative to those

2 G.M. SHAW ET AL.

persons who did not use a medication included in thatparticular coalition. ORs were computed for compari-sons that involved 15 or more cases and controlscombined. The time period considered relevant for mostof the studied illnesses and medications used was thefirst trimester. Considered as potential covariates insome analyses were maternal race/ethnicity (non-Hispanic, White; foreign-born Latina; US-born Latina;other), age (,20, 20–24, 25–29, 30–34, and .34 years),education (,high school graduate, high school gradu-ate, 2–4-year college graduate), body mass index (#29kg/m2 or .29 kg/m2), and use of vitamins containingfolic acid (for the period 3 months before and afterconception).

RESULTS

Compared to control mothers, case mothers weremore likely to be foreign-born Latina, to be less thanage 25 years, to have a household income under $10,000,and were less likely to have graduated from college,been employed, or to have used vitamins, cigarettes, oralcohol in the 6-month periconceptional period 3 monthsbefore and after conception (Table 1).

Risks of having an NTD-affected pregnancy fromselected maternal illnesses, and for some, from theirassociated medical treatments, are shown in Table 2.Many of the reported illnesses were relatively rare. Forcomparisons involving at least 15 cases and controls(combined), all ORs, except for bladder infection in thefirst trimester, were less than 1.5. Use of therapeuticmedications associated with the studied illnesses didnot reveal substantially different risk estimates fromtheir associated conditions. Single variable adjustmentfor maternal race/ethnicity, age, education, body massindex, or vitamin use did not yield substantially differ-ent risk estimates for any illness, or its treatment witha medication.

A potential association with maternal fever or febrileillness in the first trimester was explored. ElevatedNTD risks were observed for both fever and febrileillness experienced by the mother in the first trimester(Table 3). NTD risks associated with maternal feverwere lower among women who used a fever-reducingmedication (approximately 95% of women who usedsuch a medication used acetaminophen). Analyses forthe 2 common NTD phenotypes (anencephaly and spinabifida) revealed risks associated for fever and febrileillness in the first trimester to be somewhat higher foranencephaly, OR 5 2.64 (1.57–3.72) for fever, and OR 52.48 (1.31–4.69) for febrile illness, than for spina bifida,OR 5 1.69 (1.12–2.55) for fever, and OR 5 1.74 (0.94–3.20) for febrile illness. NTD risks for fever or febrileillness were not greatly influenced by simultaneousadjustment for maternal race/ethnicity, age, education,body mass index, and vitamin use, adjusted OR 5 1.99(1.37–2.90) for fever and adjusted OR 5 1.99 (1.12–3.46) for febrile illness. NTD risks were also not substan-tially changed if women who had bladder infectionswere included in the febrile illness category.

We also explored the potential risks associated withmaternal use of medications irrespective of their indica-tion. As noted above (see Methods), this was accom-plished by examining selected medication coalitions asthe analytic ‘‘exposure’’ unit. The same woman could berepresented in more than one coalition for a givenmedication used. Risks of having an NTD-affectedpregnancy associated with first trimester use of se-lected medication coalitions are shown in Table 4. Forpresentation purposes, only medication coalitions thatincluded 15 or more women (cases and controls com-bined) are shown. ORs exceeding 1.5 were observedonly for the guaifenesin coalition. The 95% confidenceintervals for both estimates, however, included 1.0,suggesting limited precision. ORs of 0.5 or less, suggest-ing a reduced NTD risk, were observed for use ofcalcium-containing products. Risk estimates for these

TABLE 1. Maternal characteristics according to casecontrol study group (%)*

Characteristic

Casemothers(n 5 538)

Controlmothers(n 5 539)

Race/ethnicityNon-Hispanic, White 40.0 50.3Foreign-born Latina 35.7 21.7U.S.-born Latina 13.8 14.8Other 10.4 13.0

Age (years),20 13.0 8.720–24 29.6 24.925–29 28.8 30.830–34 20.3 24.7$35 8.4 10.9

Education,High school 35.3 26.0High school graduate 47.2 45.6College graduate (includes

junior college) 17.1 28.2Household income (U.S. $)

,9,999 25.1 15.010,000–29,999 27.5 30.230,000–49,999 18.2 22.8$50,000 17.1 24.5

Gravidity1 24.0 19.72 26.8 29.73 22.1 21.241 27.1 29.5

Employed 3 months before orafter conception 65.6 73.4

Use of vitamins containingfolic acid

None 38.4 27.63 months before conception 15.4 17.4Started use in first tri-

mester 43.9 51.9Cigarette use in first trimester

None 82.7 78.5Any 17.1 21.5

Alcohol use in first trimesterNone 71.9 65.7Any 27.9 34.3

*Percentages may not add to 100 because of rounding ormissing information for some subjects.

MATERNAL ILLNESS AND NTD RISK 3

medicine coalitions were fairly precise. Simultaneousadjustment for maternal race/ethnicity, age, education,body mass index, and vitamin use did not revealsubstantially different results for most coalitions, al-though adjustment did result in ORs exceeding 1.5 forpseudophedrine use, OR 5 1.59 (0.86–2.95).

In addition to their illnesses and medication usenoted above, women were queried about whether they

used oral contraceptives, received general anesthesia,or were exposed to diagnostic x-rays in the periconcep-tional period (3 months before and after conception).Compared to women who did not use oral contracep-tives, those who did (69 case and 65 control mothers) inthe 3 months before conception revealed an OR of 1.10(0.76–1.57). Those women who reported use within 2months after conception (26 case and 20 control moth-

TABLE 2. Relative risk (odds ratios) for an NTD-affected pregnancy from selectedmaternal illnesses and from some associated treatments

IllnessesNo.

casesNo.

controlsOddsratio1

95%confidence

interval

Diabetes2 7 4 1.76 0.52–5.44Juvenile 1 1 1.00 0.10–9.66Adult, insulin dependent 3 1 3.02 0.35–15.95Adult, noninsulin dependent 3 2 1.51 0.28–7.15

Epilepsy/Seizure2 9 7 1.29 0.49–3.36Nonepileptic 2 5 0.40 0.10–2.04Epileptic 7 2 3.51 0.71–12.64

Medication used3 2 1 2.00 0.22–12.69No medication used 5 1 5.01 0.60–22.35

Hypothyroidism 10 12 0.88 0.37–1.94Medication used3 4 4 1.02 0.28–3.74No medication use 6 8 0.76 0.28–2.18

Thyrotoxicosis 6 7 0.87 0.31–2.50Medication used3 2 2 1.02 0.18–5.80No medication used 4 5 0.82 0.24–2.92

Goiter 5 2 2.55 0.50–10.05Medication used3 1 0 — —No medication used 4 2 2.04 0.39–8.66

Asthma2 39 52 0.73 0.47–1.12Medication used3 14 15 0.91 0.44–1.88No medication used 25 37 0.66 0.39–1.10

High blood pressure2 84 84 1.00 0.72–1.39Medication used3 4 7 0.57 0.18–1.85No medication used 80 77 1.05 0.74–1.46

Blood clots2 15 13 1.16 0.56–2.42Medication used3 2 0 — —No medication used 13 13 1.01 0.47–2.15

Tumor2 46 38 1.24 0.79–1.93Ever treated by radiation or

chemotherapy 10 10 1.02 0.43–2.41

Bladder infection, first trimester 46 32 1.50 0.94–2.38Medication used3 32 24 1.39 0.81–2.38No medication used 14 8 1.82 0.78–4.28

Vaginal infection, first trimester 42 46 0.90 0.58–1.39Medication used3 28 33 0.84 0.50–1.40No medication used 14 13 1.06 0.50–2.24

Yeast infection, first trimester 15 25 0.61 0.32–1.17Medication used3 13 20 0.61 0.33–1.34

Genital herpes, first trimester 2 3Genital warts, first trimester 6 2Chlamydia, first trimester 3 1Influenza, first trimester 4 1Heat exhaustion, first trimester 6 11Reference group for calculation of the odds ratio is persons without a particular condition.For an odds ratio associated with medication use, the reference group is persons who did notuse that particular medication and who did not have that particular condition.2Refers to ever having a history of that condition not limited to the first trimester.3Medication used in the first trimester.

4 G.M. SHAW ET AL.

ers) revealed an OR of 1.34 (0.74–2.42). With respect togeneral anesthesia, 7 case and 7 control mothers re-ported receiving anesthesia (3 for dental related proce-dures and 11 for other surgical procedures) indicatingno increased NTD risk, OR 5 1.00 (0.36–2.76). Someevidence for an increased NTD risk was observed amongwomen (39 case and 24 control mothers) who had beenexposed to dental x-rays, OR 5 1.68 (1.00–2.83). However,the risk estimate was not as elevated among women whowere exposed to diagnostic x-rays for other (nondental)indications, OR 5 1.26 (0.72–2.20).

DISCUSSION

Overall, these results suggest that most maternalillnesses (acute or chronic) and the medications used totreat them in the periconceptional period did not contrib-ute substantially to the occurrence of NTDs in thisCalifornia population. In fact, maternal fever and fe-brile illness were the only acute conditions that affectedmore than 2% of case mothers, and had an associated

risk of 1.5 or more. Although this study was advantagedby its size, its population-based ascertainment of casesand controls, its high participation, and its control ofrelevant confounders, it was still hindered by theinfrequency of any specific illness or use of any particu-lar medication among participants.

Epidemiologic studies specifically investigating NTDrisks associated with maternal illness or medicationuse are relatively few. Increased NTD risks have beenassociated with maternal conditions and treatment,such as insulin-dependent diabetes (Myrianthopoulosand Melnick, ’87; Becerra et al., ’90); epilepsy treatedwith valproic acid (Lindhout and Schmidt, ’86); pericon-ceptional use of oral contraceptives (Kasan and An-drews, ’87; Bracken et al., ’78); ovulation-inducingmedications (Greenland and Ackerman, ’95); and fever,febrile illness, or other sources of heat exposure such assauna use (Layde et al., ’80; Shiota, ’82; Coffey andJessop, ’59; Hakosalo and Saxen, ’71; Halperin andWilroy, ’78; Chance and Smith, ’78; Miller et al., ’78;

TABLE 3. Relative risk (odds ratios) for an NTD-affected pregnancyfrom maternal fever or maternal febrile illness

Maternal fever/febrile illnessNo.

casesNo.

controlsOddsratio2

95%confidence

interval

Neither, in first trimester 367 427 Reference —First trimester

Fever 97 59 1.91 1.35–2.72No medication used1 55 22 2.91 1.75–4.84Fever-reducing medication used 38 33 1.34 0.83–2.17

Febrile Illness 40 23 2.02 1.20–3.43No medication used1 23 14 1.91 0.98–3.73Fever-reducing medication used 4 1Other medication used 16 9 2.07 0.92–4.64

1No medication used means woman did not use a fever reducing medication or anothermedication to treat cause of fever.2Odds ratio estimates only shown for comparisons that included 15 or more case and controlsubjects.

TABLE 4. Relative risk (odds ratios) for an NTD-affected pregnancy associatedwith use of selected maternal medication coalitions*

Medication coalitionNo.

casesNo.

controlsOddsratio

95%confidence

interval

Ampicillin/amoxicillin 13 16 0.81 0.39–1.68Penicillin (natural penicillin) 9 10 0.90 0.37–2.17Antifungals 10 15 0.66 0.30–1.46Antihistamines 21 31 0.67 0.38–1.17Decongestants

Phenylpropanolamine 10 11 0.91 0.39–2.11Pseudophenedrine 22 26 1.20 0.67–2.13

Aspirin 22 19 1.17 0.63–2.18Ibuprofen 20 22 0.91 0.50–1.68Acetaminophen 144 141 1.05 0.80–1.38Codeine 8 9 0.89 0.35–2.24Antacids 16 27 0.58 0.31–1.09Calcium-containing products 5 13 0.38 0.14–1.03Guaifenesin 12 6 2.04 0.79–5.28

*Medication coalitions, developed by Slone Epidemiology Unit, Boston University, reflectsimilar pharmacologic action, class, or chemical structure.

MATERNAL ILLNESS AND NTD RISK 5

Hunter, ’84; Sandford et al., ’92; Milunsky et al., ’92;Zhang and Cai, ’93; Lynberg et al., ’94; Werler et al.,’95). Among these risk factors, the conditions in thecurrent study that had sufficient sample size and wereassociated with elevated risks were limited to fever andfebrile illness.

Our finding an increased risk for fever and febrileillness is supported by animal studies (Edwards et al.,’95) and is consistent with numerous human studies(Layde et al., ’80; Shiota, ’82; Coffey and Jessup, ’59;Hakosalo and Saxen, ’71; Halperin and Wilroy, ’78;Chance and Smith, ’78; Miller et al., ’78; Hunter, ’84;Sandford et al., ’92; Milunsky et al., ’92; Zhang and Cai,’93; Lynberg et al., ’94; Werler et al., ’95), that haveimplicated maternal hyperthermia or fever-producingillness as a risk factor for NTDs. The data from animalstudies suggest that hyperthermia itself is the terato-gen, whereas the data from epidemiologic studies can-not fully disentangle whether elevated temperature orthe illness responsible for the elevated temperature isthe potential teratogenic mechanism. However, Milun-sky et al. (’92) have recently noted that women exposedto heat solely through hot tub use were 3 times morelikely to have an NTD-affected pregnancy. We foundthat a maternal fever, or a febrile illness, episode in thefirst trimester was associated with approximately a2-fold increased NTD risk. These increased risks per-sisted even after controlling for potentially relevantcovariates. From the current data, we could not deter-mine whether the observed increased risk was attribut-able to the elevated temperature or the illness underly-ing the association. The magnitude of the increasedrisks for fever, however, was somewhat reduced amongthose who took a fever reducing medication. The lowerrisk among those women who used a medication may beindicative of a reduction in core temperature followinguse of the medication. Alternatively, the more elevatedrisk among those who reported fever, but did not use afever reducing medication, may be due to recall bias;i.e., it may be indicative of case mothers being morelikely than control mothers to recall less severe feverepisodes that did not warrant medication use. Thus, ininterpreting our findings for fever, and perhaps forfebrile illness, we cannot exclude the possibility thatthe increased NTD risks observed were attributable tobiased reporting; i.e., case mothers misreporting thatthey had a fever (or febrile illness) or control mothersmisreporting that they did not have a fever episode.This is a particular concern given that the degree ofelevated temperature denoting a fever was not mea-sured or asked about. Some indication that reportingbias may explain some of the increased risk associatedwith fever was evidenced by the fact that we observedan OR of 1.32 (0.79–2.21) for NTDs among women whoreported a fever episode in the second trimester. Thesecond trimester is believed to be an irrelevant embry-onic period for neural tube development because theneural tube closes approximately 28 days after concep-tion. This risk estimate could be indicative of reporting

errors in general, could be a chance finding, or could beindirect evidence for the less accepted theory thatNTDs arise as a result of a reopening of a previouslyclosed neural tube (Hook, ’92). The risk associated withmaternal febrile illness in the second trimester, how-ever, was not as elevated OR 5 1.10 (0.64–1.92), whichargues against reporting bias as the alternative expla-nation of our findings. Further, the differences in risksobserved by NTD phenotype may be indirect evidencein favor of our results not being solely attributable toreporting bias. There is little reason to assume thatmothers of infants/fetuses with anencephaly wouldreport fever or febrile illness less accurately thanmothers of infants/fetuses with spina bifida.

Overall, our results suggest that the attributable riskfor NTDs associated with many of the illnesses andmedications used among reproductive aged women isrelatively low. However, because of the infrequency ofwomen reporting any specific illness or medicationused, it will be difficult to adequately estimate therelative risk for NTDs associated with any maternalillness or medication used, even in large case-controlstudies such as the current one. Estimation of relativerisks, therefore, will likely be possible only in largemulticentered studies.

ACKNOWLEDGMENTS

The authors thank Dr. Allen Mitchell for his insight-ful comments on an earlier version of this paper. Thiswork was supported by funds provided to the March ofDimes and the California Birth Defects MonitoringProgram from the California Department of HealthServices.

LITERATURE CITEDBecerra, J.E., M.J. Khoury, J.F. Cordero, and J.D. Erickson (1990)

Diabetes mellitus during pregnancy and risks for specific birthdefects: A population based case-control study. Pediatrics, 85:1–9.

Bracken, M.B., T.R. Holford, C. White, and J.L. Kelsey (1978) Role oforal contraception in congenital malformation of offspring. Int. J.Epidemiol., 7:309–317.

Chance, P.F., and D.W. Smith (1978) Hyperthemia and meningomyelo-cele and anencephaly. Lancet, 1:769–770.

Coffey, V.P., and W.J.E. Jessop (1959) Maternal influenza and congeni-tal deformities: A prospective study. Lancet, 2:935–938.

Croen, L.A., G.M. Shaw, N.J. Jensvold, and J.A. Harris (1991) Birthdefects monitoring in California: A resource for epidemiologicalresearch. Ped. Perinatol. Epidemiol., 5:423–427.

Czeizel, A.E., and I. Dudas (1992) Prevention of the first occurrence ofneural-tube defects by periconceptional vitamin supplementation.N. Engl. J. Med. 327:1832–1835.

Edwards, M.J., K. Shiota, M.S.R. Smith, and D.A. Walsh (1995)Hyperthemia and birth defects. Repro. Toxicol., 9:411–425.

Elwood, J.M., J. Little, and J.H. Elwood (1992) Epidemiology andControl of Neural Tube Defects. Oxford, England: Oxford UniversityPress.

Greenland, S., and D.L. Ackerman (1995) Clomiphene and neural-tubedefects: a pooled analysis of controlled epidemiologic studies andrecommendations for future studies. Fertility Sterility, 64:936–941.

Hakosalo, J., and L. Saxen (1971) Influenza epidemic and congenitaldefects. Lancet, 2:1346–1347.

Halperin, L.R., and R.S. Wilroy, Jr. (1978) Maternal hyperthermia andneural-tube defects. Lancet 2:212–213.

6 G.M. SHAW ET AL.

Hook, E.B. (1992) Neural tube rupture as a cause of neural tubedefects. Lancet, 339:1000.

Hunter, A.G.W. (1984) Neural tube defects in eastern Ontario andwestern Quebec: Demography and family data. Am. J. Med. Genet.,19:45–63.

Kasan, P.N. and J. Andrews (1980) Oral contraception and congenitalabnormalities. Br. J. Obstet. Gynecol. 87:545–551.

Layde, P.M., L.D. Edmonds, and J.D. Erickson (1980) Maternal feverand neural tube defects. Teratology, 21:105–108.

Lindhout, D., and D. Schmidt (1986) In-utero exposure to valproateand neural tube defects. Lancet, 1:1392–1393.

Lynberg, M.C., M.J. Khoury, X. Lu, and T. Cocian (1994) Maternal flu,fever, and the risk of neural tube defects: A population-basedcase-control study. Am. J. Epidemiol., 140:244–255.

Medical Research Council (Vitamin Study Research Group) (1991)Prevention of neural tube defects: Results of the Medical ResearchCouncil vitamin study. Lancet, 338:131–137.

Miller, P., D.W. Smith, and T.H. Shepard (1978) Maternal hyperther-mia as a possible cause of anencephaly. Lancet, 1:519–521.

Milunsky, A., M. Ulcickas, K.J. Rothman, W. Willett, S.S. Jick, and H.Jick (1992) Maternal heat exposure and neural tube defects. JAMA,268:882–885.

Myrianthopoulos, N.C., and M. Melnick (1987) Studies in neural tubedefects. I. Epidemiologic and etiologic aspects. Am. J. Med. Genet.,26:783–796.

Sandford, M.K., G.E. Kissling, and P.E. Joubert (1992) Neural tubedefect etiology: New evidence concerning maternal hyperthermia,health and diet. Dev. Med. Child. Neurol., 34:661–675.

Shaw, G.M., D. Schaffer, E.M. Velie, K. Morland, and J.A. Harris(1995) Periconceptional vitamin use, dietary folate, and the occur-rence of neural tube defects. Epidemiol., 6:219–226.

Shiota, K. (1982) Neural tube defects and maternal hyperthermia inearly pregnancy: Epidemiology in a human embryo population. Am.J. Med. Genet., 12:281–288.

Werler, M.M., C. Louik, and A.A. Mitchell (1995) Heat source exposureand neural tube defect risk. Am. J. Epidemiol., 142:369.

Zhang, J., and W-W. Cai (1993) Association of the common cold in thefirst trimester of pregnancy with birth defects. Pediatrics, 92:559–563.

MATERNAL ILLNESS AND NTD RISK 7