Click on the links below to access all the ArticlePlus for this article.
Please note that ArticlePlus files may launch a viewer application outside of your web browser.
Nearly 100 years ago, scientists discovered that malformations could be induced in various animal species by exposing the mother to high core body temperatures during critical periods of gestation.1 Although a number of teratologic outcomes were produced, including fetal loss and various malformations, the brain was particularly sensitive.2,3 In humans, an elevated core body temperature can occur with fever caused by viral or bacterial illnesses. In addition, extreme exercise, saunas, hot tubs, heated beds, and electric blankets may all lead to increased body temperatures. Presently, there is no consensus as to the role of hyperthermia during pregnancy in causing malformations in humans. Many of the available studies are inconclusive due to insufficient sample size.
A fever can be of varying duration, as with exposure to other potential teratogens. A fever may be recognized and treated with antipyretics or it may remain unidentified or untreated. Serious illnesses accompanied by fevers (eg, malaria) may be associated with poor nutritional intake by the mother, which itself may be associated with poor pregnancy outcome. The infectious agent itself may be teratogenic (such as rubella, varicella or cytomegalovirus). The literature on the effects of maternal hyperthermia has investigated various outcomes, including pregnancy loss and specific malformations. Our objective was to investigate the association between maternal hyperthermia in early pregnancy and the risk for risk for neural tube defects (NTDs). We chose this outcome because there are a substantial number of published studies on this association, and because it is an outcome for which animal data lend biologic plausibility.
We searched MEDLINE and EMBASE to locate articles in any language in 1966 through August 2003. In addition, the reference sections from articles retrieved from the original database searches were inspected for additional publications. Published meeting abstracts and proceedings were also searched between 1980 and February 2003 for studies in progress or studies not published in the peer reviewed literature. The keywords “pregnancy,” “pregnancy outcome,” “birth defect,” “teratogen,” and “congenital abnormalities” were searched. The intersection of this search with articles retrieved by using the keyword “hyperthermia” generated the final set of articles retrieved for evaluation.
Two investigators familiar with clinical study design acted as reviewers, selecting studies for inclusion in the meta-analysis based on preset inclusion criteria and following a standardized checklist (Appendix 1, available with the electronic version of this article). Reviewers received only the Methods sections; they were unaware of the authors’ and journal names, and year of publication. Studies were excluded if they were not performed in humans, did not report pregnancy outcome, did not specifically extract information on neural tube defects, reported on fewer than 5 subjects, did not have a control group, were not cohort or case–control studies, or did not report hyperthermia exposure occurring in the first trimester. Hyperthermia or fever was included if it was either internal (such as maternal fever caused by illness) or external (exposures to hot tubs, heated beds, saunas or electric blankets). Each reviewer listed the reason for exclusion.
The reviewers were then given the Results section of each of the included papers, with identifying information removed. Reviewers extracted the relevant data from the text and tables, and entered these numbers into 2 × 2 tables. The following neural tube defects were included: anencephaly, spina bifida, encephalocele, myelomeningocele, exencephaly, myeloschisis, iniencephaly, and craniorachischisis. Studies were included if they reported the total number of NTDs, even if the numbers of specific NTDs were not specified. Internal hyperthermia was defined as documented maternal fever (either receiving treatment or not) or documentation of maternal illnesses know to be associated with fever (“febrile illnesses”) in the first trimester of pregnancy. External hyperthermia was defined as exposure to at least 15 minutes of high temperature exposure via hot tub, sauna, electric blanket or heated waterbed in the first trimester of pregnancy.
Each reviewer extracted the data independently. For studies about which there was disagreement on inclusion or data extraction, the reviewers discussed the rationale for their choices and came to consensus. The extracted data were entered into Cochrane's Review Manager (version 4.1, Oxford: Cochrane Collaboration). For summaries that included case-control studies, we calculated individual and summary odds ratios (ORs) and 95% confidence intervals (CIs) by the Mantel-Haenszel method. For cohort studies we used the relative risk (RR) and 95% CI. Included studies were tested for heterogeneity using the χ2 test. Publication bias was assessed by visual examination of the funnel plot (effect size versus log study size) and by the trim-and-fill method of Duval and Tweedie.4,5
Initially, 42 studies were identified and presented to the reviewers for possible inclusion. Of these, we subsequently excluded 24 articles because they did not meet the inclusion criteria or because the Methods and Results sections did not provide sufficient information. For example, a study from a Japanese group6 was excluded because it described a retrospective cohort with no control group. Another study7 reported sauna use in all cases and all controls and focused on comparing habits of sauna use in the 2 groups. One study8 was excluded because the report did not give the total number of prospectively collected subjects but rather described only the affected cases in detail. An additional 3 studies were excluded because they were publication of duplicate data.9–11 A full list of excluded articles is available from the authors (Appendix 2, available with the electronic version of this article).
Details regarding the 15 included studies are shown in Table 1. 12–26 There were 9 case–control studies, with a total of 1601 NTD cases and 5149 controls. The overall OR was 1.93 (95% CI = 1.53–2.42) for NTDs associated with maternal hyperthermia exposure (Fig. 1). There were 6 prospective cohort studies reporting on 8,798 exposed infants and 24,069 unexposed infants, yielding an overall RR of 1.95 (CI = 1.30–2.92) (Fig. 2). When combined, the overall OR of all 15 studies was 1.92 (CI = 1.61-2.29). The χ2 test for heterogeneity was 13.14 (df = 13; P = 0.44), with no statistical evidence for heterogeneity. One cohort study24 did not contribute to the summary rates because there were no NTDs detected, and thus the individual RR/OR could not be estimated. Findings did not vary by year of publication. Some publications reported on other potential sources for variation in findings (eg, maternal characteristics, degree of fever and folate intake). However, there were not sufficient data to conduct secondary analysis on these parameters. A funnel plot arrangement of the results was asymmetric, suggesting the presence of publication bias. The adjusted odds ratio (using the trim-and-fill method) was 1.86 (95% CI = 1.54–2.24).
Hyperthermia was the first environmental teratogen identified by scientists in animal models. Even so, there has not been a consensus about its effect in humans. In animals, one can deliver precise amount of hyperthermia in terms of timing, length, and degree. Evaluation of exposure in humans has been substantially more complicated. In humans, hyperthermia is often associated with an infectious disease that may pose a risk to the fetus both in terms of the pathogen itself and through other changes in maternal well being and nutritional status.
We focused on NTDs because this was the endpoint of most human studies on pregnancy outcomes after exposure to hyperthermia. Most studies were conducted before the association between risk for NTD and low folate intake was established and, therefore, data on daily folate or blood levels of this vitamin were not included. Four studies made note of maternal use of multivitamins or folic acid specifically.18–20,25 In 2 of these publications,18,19 elevated risks for NTD persisted even after adjusting for maternal multivitamin use; in the more recent article by Shaw et al,20 the ORs for the association of hyperthermia with NTDs were somewhat lower among vitamin users but still elevated. In contrast, Milunsky et al25 showed that only hot tub use affected NTD risk after controlling for folic acid, while the risks for NTD following maternal fever, sauna or electric blanket use in pregnancy were minimal.
In the analysis reported here, the heat source for most studies was influenza or other febrile illnesses. However, a similar increased risk was found in studies where hyperthermia was due to exposure to external heat sources (eg, Milunsky et al25). These studies suggest that, as in the animal data, it is core body temperature and not confounding effects of viral infections that cause NTDs. Of note, Lynberg et al18 reported that influenza was associated with an increased risk for NTDs, whereas other febrile illnesses did not appear to increase the NTD risk after adjusting for confounders. Only 3 studies reported on outcome after both internal fever and external hyperthermia13,14,25; however, the sample size was very small in 2 of these studies,13,14 and in the third,25 internal and external heat sources were not separated.
There was some indication of a dose response. Milunsky et al25 reported that risks were higher among women who had exposure to more than one heat source, and Chambers et al26 found all cases of NTDs were in the high-fever group. Even so, given the limited data addressing specific details of magnitude and duration of the exposures, it remains unclear whether the hyperthermia itself or the maternal illness was the cause of the teratogenic outcomes.
Several studies explicitly reported the pharmacologic treatment of the maternal fevers,14,16,18,19,26 although data were rarely presented in a way we could include in our analysis. Lynberg et al18 found that ORs for NTDs were increased if medications were taken, whereas Shaw et al19 reported that use of medications decreased risks for NTD. We found very similar effects in case–control and cohort studies, lending further credibility to the observed association. The studies were statistically homogeneous, suggesting that pooling the data is appropriate.
One of the concerns regarding meta-analysis is the combination of studies with a range of quality. As suggested originally by Glass27,28 and confirmed by us,29 the quality of the study does not appear to affect the directional of the results. However, there are too few empiric observations to support such generalizations.
Publication bias also was a concern in meta-analysis. Arranging the results in a funnel plot should produce an inverted funnel if publication bias is not present. In this case the funnel plot was not symmetrical, with the lower left quadrant absent, suggesting a deficit of smaller negative studies. Among the methods for quantifying the degree of asymmetry (and thus publication bias) the most widely used is the trim–and-fill method proposed by Duval and Tweedie.4,5 In this analysis the assessment did not change the direction or significance of the results, suggesting that publication bias is not an important factor in these results.
In the studies reviewed here, hyperthermia met many of the criteria for a human teratogen as established by Shepard.30 The exposures occurred at a critical time in prenatal development; only studies that defined exposures in the first trimester were included in our meta-analysis. The findings were consistent across epidemiologic studies of high quality. Despite the fact that some of the studies did not control for confounders or biases, all studies showed the same directional of risk. Finally, as described earlier, similar consistent findings have been shown in animals, which lends biologic plausibility to teratogenic causality. To strengthen the evidence for causality, future studies should attempt to define exposures with more detail and precision, while incorporating measures of folate intake to correct for this potential effect modifier.
1. Warkany J. Teratogen update: hyperthermia. Teratology
2. Edwards MJ. Congenital defects in guinea pigs: prenatal retardation of brain growth of guinea pigs following hyperthermia during gestation. Teratology
3. Edwards MJ. Congenital defects in guinea pigs: fetal resorptions, abortions, and malformations following induced hyperthermia during early gestation. Teratology
4. Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics
5. Duval S, Tweedie R. A nonparametric “trim and fill” method of accounting for publication bias in meta-analysis. J Am Stat Assoc
6. Hirayama Y, Suzuki H, Arima M. Maternal hyperthermia during pregnancy in severely multi-disabled children [Japanese]. No To Hattatsu
7. Saxen L, Holmberg PC, Nurminen M, Kuosma E. Sauna and congenital defects. Teratology
8. Smith DW, Clarren SK, Harvey MA. Hyperthermia as a possible teratogenic agent. J Pediatr
9. Layde PM, Edmonds LD, Erickson JD. Maternal fever and neural tube defects. Teratology
10. Erickson JD. Risk factors for birth defects: data from the Atlanta Birth Defects Case-Control Study. Teratology
11. Botto LD, Erickson JD, Mulinare J, Lynberg MC, Liu Y. Maternal fever, multivitamin use, and selected birth defects: evidence of interaction? Epidemiology
12. Chance PF, Smith DW. Hyperthermia and meningomyelocele and anencephaly (letter). Lancet
13. Halperin LR, Wilroy RS Jr. Maternal hyperthermia and neural-tube defects (letter). Lancet
14. Miller P, Smith DW, Shepard TH. Maternal hyperthermia as a possible cause of anencephaly. Lancet
15. Fisher NL, Smith DW. Occipital encephalocele and early gestational hyperthermia. Pediatrics
16. Shiota K. Neural tube defects and maternal hyperthermia in early pregnancy: epidemiology in a human embryo population. Am J Med Genet
17. Dlugosz L, Vena J, Byers T, Sever L, Bracken M, Marshall E. Congenital defects and electric bed heating in New York State: a register-based case-control study. Am J Epidemiol
18. Lynberg MC, Khoury MJ, Lu X, Cocian T. Maternal flu, fever, and the risk of neural tube defects: a population- based case-control study. Am J Epidemiol
19. Shaw GM, Todoroff K, Velie EM, Lammer EJ. Maternal illness, including fever and medication use as risk factors for neural tube defects. Teratology
20. Shaw GM, Nelson V, Carmichael SL, Lammer EJ, Finnell RH, Rosenquist TH. Maternal periconceptional vitamins: interactions with selected factors and congenital anomalies? Epidemiology
21. Coffey VP, Jessop WJE. Maternal influenza and congenital deformities: A prospective study. Lancet
22. McDonald AD. Maternal health in early pregnancy and congenital defect. Br J Prev Soc Med
23. Kleinebrecht J, Michaelis H, Michaelis J, Koller S. Fever in pregnancy and congenital anomalies (letter). Lancet
24. Little BB, Ghali FE, Snell LM, Knoll KA, Johnston W, Gilstrap LC III. Is hyperthermia teratogenic in the human? Am J Perinatol
25. Milunsky A, Ulcickas M, Rothman KJ, Willett W, Jick SS, Jick H. Maternal heat exposure and neural tube defects. JAMA
26. Chambers CD, Johnson KA, Dick LM, Felix RJ, Jones KL. Maternal fever and birth outcome: a prospective study. Teratology
27. Glass GV. Primary, secondary, and meta-analysis of research. Educ Res
28. Glass GV. Integrating findings: the meta-analysis of research. Rev Res Educ
29. Seto A, Einarson T, Koren G. Pregnancy outcome following first trimester exposure to antihistamines: meta-analysis. Am J Perinatol
30. Shepard. TH. Catalog of Teratogenic Agents
. 10th ed. Baltimore: Johns Hopkins University Press; 2001.