From the Slone Epidemiology Unit, Brookline, MA.
Address correspondence to: Martha Werler, Slone Epidemiology Unit, 1371 Beacon Street, Brookline, MA 02446. Fogarty Senior International Fellow, Institute for Child Health Research, Perbin, Australia.
In this issue, two reports by Botto and colleagues shed more light on the question of what factors influence the development of congenital heart defects (CHDs). One study identified a 1.8-fold increased risk of CHDs for reported maternal fever, using data from the Atlanta Birth Defects Case Control Study (ABDCCS). 1 The other identified a greater than 9-fold increased risk of transposition of the great arteries for retinol intakes of 10,000 IU or more, using data from the Baltimore-Washington Infant Study (BWIS). 2 CHDs, as a group, are the most common structural malformation, occurring among an estimated 4–6 infants per 1,000 live births when prematurity-associated defects and spontaneously closing atrial septal defects are excluded. 3,4 A continuing challenge among birth defects epidemiologists is the classification of CHDs into etiologically meaningful groups. Over 30 specific anatomical defects are included under the label CHD, with varying epidemiologic features. 5 In consideration of this variability, Botto and colleagues appropriately focused their analyses on defects that were not associated with a known syndrome. The greater challenge is classifying cases with more than one cardiac defect and/or extra-cardiac defects. The approach used in both studies was to create a diagnostic hierarchy, giving priority to defects with the earliest embryologic disturbance, and then to categorize the case with the one diagnosis that is highest in the hierarchy. 5 This strategy represents a major advance in studying the epidemiology of CHDs, but there are limitations. First, the approach ignores the issue of complex cardiac defects by combining cases affected by multiple defects (particularly defects that are thought to arise from different pathogenetic mechanisms) with cases affected by only one defect. Second, the approach also ignores the fact that both single and multiple cardiac defects can occur along with defects of other organs and/or systems, possibly representing different etiologies. Although cases in the BWIS and the ABDCCS databases have previously been classified according to the presence of extra-cardiac defects, these subgroups were not presented in these reports, presumably owing to small numbers.
The list of known risk factors for specific CHDs is short, including only positive family history of CHD and maternal diabetes mellitus, epilepsy and/or anticonvulsant use, rubella, and isotretinoin use. 6,7 The paucity of known risk factors cannot be attributed to a lack of studies. Indeed, there are many “suspected” risk factors—those that have been implicated in some studies, but refuted in others. Both maternal fever and vitamin A intake fall into this category. 6,8–18 Many methodologic arguments (e.g., confounding, selection bias, definitions of outcomes, random misclassification, recall bias) can be invoked here to explain differences in study findings. Because specific CHDs are rare, the usual epidemiologic approach is retrospective, with concomitant challenges in accurate exposure measurement.
In an effort to assess validity of reported fever in the ABDCCS, Botto et al. compare its reported prevalence of fever with prevalences reported in other studies, 6,8,9 acknowledging differences in exposure definition. Such comparisons are further complicated by length of the exposure interval because the longer the interval the greater the opportunity for exposure to occur. While the ABDCCS examined a four-month interval, previous studies examined “the first trimester”, 6,8,9 which tends to be regarded by study subjects as an approximate 3 month interval from the time pregnancy is recognized to the third month. In addition, infectious diseases that cause fever will vary by geography and time, further obscuring comparisons across studies. Nevertheless, concerns about reporting accuracy are warranted. The ABDCCS collected exposure information by maternal interview two to fourteen years after the birth of the study child. Fever as an indicator of hyperthermia or infection is extremely difficult to measure accurately in retrospective studies, even when the interval between early gestation and data collection is comparatively short. I recently collected data that illustrate this point. Among control mothers who were interviewed within six months of birth, only 60% of those with a reported fever had taken their temperatures, and 80% of documented fevers were reported to be less than 102°F (38.9°C), the postulated threshold for teratogenesis in humans. 19 It is also likely that some women with infections do not report fever, when, in fact, temperatures were elevated. Thus, reported fever in the ABDCCS is likely an imperfect measure of hyperthermia or infection. Inaccurate measurement of similar magnitudes of maternal fever across both case and control groups (i.e., random misclassification) is not likely to explain the positive associations observed in the ABDCCS, but differential misclassification (i.e., recall bias) may play a role. Although there is little empirical evidence of the existence of recall bias in studies of birth defects, this may well be due to the difficulties in measuring its presence. Any influence it may have would most likely be exposure-specific. For example, if concerns are voiced about a particular exposure and its effects on pregnancy outcomes in the lay press, studies in which data collection closely follows such reports may be most vulnerable. Therefore, concerns of recall bias persist in studies of birth defects. Because nearly all categories of cardiac defects were found to be associated with fever in the Botto et al. study, 1 differential reporting of fever between cases and controls is suspect. In the previous retrospective studies, recall bias appears to be less likely because maternal fever was observed to increase risks of some, but not all, cardiac defects. 6,8,9
It is generally accepted that retinol affects fetal development at both extremely low and high doses. 20,21 The relation between risk and dose in humans is still being explored. Botto et al. present data suggesting that intakes of at least 10,000 IU of supplemental retinol affect the risk of one type of CHD, dextro-transposition of the great arteries. 2 This finding is contrary to three earlier studies that examined ≥10,000 IU of retinol, 13,14,18 but each was limited in size and precision, particularly for outflow tract defects. One previous report observed 2.4-fold increased risk of birth defects for the same level of supplemental retinol, 12 but it was not specific for outflow tract defects. The Botto et al. finding is supported by high quality diagnostic information and classification in the BWIS 5 and by experimental and clinical evidence that shows retinol, retinol esters, and isotretinoin affect cranial neural crest cell activity 7,21 and that shows the great arteries are at least partially derived from that cell population. 22 Nevertheless, the increased risk observed by Botto et al. is the first for this specific cardiac defect and this specific level of retinol, and, therefore, should be confirmed in other studies. It is worth noting that the BWIS data also suggest a slight increase in risk for low supplemental retinol intake, but the category of intake (<5,000 IU) is comprised mostly of non-supplementers with varying levels of dietary intakes. It would be interesting to examine risk for non-supplementing women with low dietary intakes. Botto et al. offer another important piece information: high intakes of beta-carotene did not affect outflow tract defect risks, which is consistent with the findings of experimental studies of beta-carotene and reproductive outcomes. 21 This lack of an association is worth highlighting because some of the vitamin A in multivitamins, particularly prenatal preparations, was changed from retinol to beta-carotene over the past decade because of concerns of retinol teratogenicity. 21 In fact, it is thought that intakes of ≥10,000 IU of retinol are rare due, in part, to this replacement, although it is difficult to evaluate because manufacturers tend to combine both beta-carotene and retinol doses on multivitamin labels.
Where do birth defect epidemiologists go from here? There are many avenues to pursue. As diagnostic techniques and our understanding of cardiovascular development improves, so to will classification of congenital cardiac defects. Also, there is always need for improvement on the measurement of exposures. In addition, the consideration of biologically sensible modifying factors (as did Botto et al. in their examination of fever risks within multivitamin supplementation subgroups 1) is also an important and relatively unexplored area in human studies of CHD. Indeed, one such possibility would be to combine the topics examined by Botto et al., 1,2 since vitamin A and hyperthermia have been shown to have a synergistic effect on birth defects in hamsters. 23 Use of antipyretic medication would be another potential modifying factor for fever and birth defect risks. Finally, because specific CHDs have recurrence risks in the range of 1% to 4%, 24 gene-environment interactions deserve much attention.
1. Botto LD, Lynberg MC, Erickson JD. Congenital heart defects, maternal febrile illness, and multivitamin use: a population-based study. Epidemiology 2001; 12: 485–490.
2. Botto LD, Loffredo C, Scanlon KS, Ferencz, Khoury MJ, Wilson PD, Correa A. Vitamin A and cardiac outflow tract defects. Epidemiology 2001; 12: 491–496.
3. Wren C, Richmond S, Donaldson L. Temporal variability in birth prevalences of cardiovascular malformations. Heart 2000; 83: 414–419.
4. Ferencz C, Rubin JD, McCarter RJ, Brenner JI, Neill CA, Perry LW, Hepner SI, Downing JW. Congenital heart disease: prevalence at birth. The Baltimore-Washington Study. Am J Epidemiol 1985; 121: 31–36.
5. Ferencz C, Rubin JD, Loffredo CA, Magee CA. Epidemiology of congenital heart disease: the Baltimore-Washington Study. 1981–1989. In: Perspectives in Pediatric Cardiology, vol. 4. Mt. Kisco, NY: Futura Publishing Co., Inc., 1993.
6. Ferencz C, Loffredo CA, Rubin JD, Magee CA. Genetic and environmental risk factors of major cardiovascular malformations. In: Perspectives in Pediatric Cardiology, vol. 5. Mt. Kisco, NY: Futura Publishing Co., Inc., 1997.
7. Lammer EJ, Chen DT, Hoar N, Agnish PJ, Benke JT, et al. Retinoic acid embryopathy. N Engl J Med 1985; 313: 837–841.
8. Tikkanen J, Heinonen OP. Maternal hyperthermia during pregnancy and cardiovascular malformations in the offspring. Eur J Epidemiol 1991; 7: 628–635.
9. Zhang J, Cai WW. Association of the common cold in the first trimester of pregnancy with birth defects. Pediatrics 1993; 92: 559–563.
10. Shaw GM, Malcoe LH, Swan SH, Cummins SK, and Schulman J. Congenital cardiac anomalies relative to selected maternal exposures and conditions during early pregnancy (Letter). Eur J Epidemiol 1992; 8: 757–760.
11. Chambers CD, Johnson KA, Dick LM, Felix RJ, Jones KL. Maternal fever and birth outcome: a prospective study. Teratology 1998; 58: 251–257.
12. Rothman KJ, Moore LL, Singer MR, Nguyen US, Mannino S, Milunsky A. Teratogenicity of high vitamin A intake. N Engl J Med 1995; 333: 1369–1373.
13. Martinas-Frias ML, Salvador J. Epidemiological aspects of prenatal exposure to high doses of vitamin A in Spain. Eur J Epidemiol 1990; 6: 118–123.
14. Mills JL, Simpson JL, Cunningham GC, Conley MR, Rhoads GG. Vitamin A and birth defects. Am J Obstet Gynecol 1997; 177: 31–36.
15. Shaw GM, Wasserman CR, Block G, Lammer EJ. High maternal vitamin A intake and risk of anomalies of structures with a cranial neural crest cell contribution (Letter). Lancet 1996; 347: 899–900.
16. Werler MM, Lammer EJ, Rosenberg L, Mitchell AA. Maternal vitamin A supplementation in relation to selected birth defects. Teratology 1990; 42: 497–503.
17. Khoury MJ, Moore CA, Mulinare J. Vitamin A and birth defects (Letter). Lancet 1996; 347: 322.
18. Mastroiacovo P, Mazzone T, Addis A, Elephant E, Carlier P, Vial T, Garbis H, Robert E, Bonati M, Ornoy A, Finardi A, Schaffer C, Caramelli L, Rodriguez-Pinilla E, Clementi M. High vitamin A intake in early pregnancy and major malformations: a multicenter prospective controlled study. Teratology 1999; 59: 7–11.
19. Graham JM, Edwards MJ, Edwards MJ. Teratogen Update: Gestational effects of maternal hyperthermia due to febrile illnesses and resultant patterns of defects in humans. Teratology 1998; 58: 209–221.
20. Bendich A, Langseth L. Safety of vitamin A. Am J Clin Nutr 1989: 358–371.
21. Public Affairs. Committee of the Teratology Society. Position paper by the Teratology Society: Vitamin A during pregnancy. Teratology 1987; 35: 267–268.
22. Kirby ML, Waldo KL. Neural crest and cardiovascular patterning. Circ Res 1995; 77: 211–215.
23. Ferm VH, Ferm RR. Teratogenic interaction of hyperthermia and vitamin A. Biol Neonate 1979; 36: 168–172.
24. Nora JJ, Nora AH. Update on counseling the family with a first-degree relative with a congenital heart defect. Am J Med Genet 1988; 29: 137–142.