|
|
| |||||||||||||||||||||||||||||||||||||||
Background Certain birth defects are known to recur in families, but most estimates of the risk of recurrence have come from clinic-based studies. In this study we estimated the risk of recurrent birth defects using a population-based registry.
Methods The study was based on the records of the first and second infants delivered by 371,933 women from 1967 through 1989. The records are maintained by the Medical Birth Registry of Norway. For the 9192 women whose first infant had a birth defect, we determined the relative risk of similar and dissimilar defects in the second infant. The reference population was women whose first infant had no defect.
Results Among first infants, 2.5 percent had a birth defect. The mothers of affected first infants were 2.4 times as likely as other women to have second infants with any registered defect. This increased risk was due primarily to an increased (7.6 times higher) risk of the same defect in the second infant as in the first (95 percent confidence interval, 6.5 to 8.8) and secondarily to a slightly increased (1.5 times higher) risk of a different defect in the second infant (95 percent confidence interval, 1.3 to 1.7). Among the women who lived in the same municipality during both pregnancies, the relative risk of having a second infant with the same defect was 11.6, as compared with 5.1 among the women who moved to another municipality after the birth of their first infant (P<0.001).
Conclusions Among women whose first infant has a birth defect, the risk of the same defect in the second infant is substantially increased and the risk of a different defect in the second infant is slightly increased. Environment plays a strong part in repeated defects.
A high risk of having infants with birth defects can result from maternal or paternal genes, dietary patterns, or long-term exposure to environmental teratogens. Women may also vary in their likelihood of carrying a malformed fetus to birth. Although nongenetic factors are plausible, risks that are specific to women are usually considered to be due to genetic factors3,4.
In this study, we used a population-based registry to estimate a mother's risk of having a second infant with a defect after having given birth to a first infant with a defect. We also examined how these risks may be affected when the mother changes her partner or her city of residence between her first and second completed pregnancies.
Methods
The Medical Birth Registry of Norway includes all live births and stillbirths with a gestational age of at least 16 weeks. Registration of births is compulsory and is performed by midwives. About 1.5 million births have been registered since the registry was established in 1967. A number is assigned to each record of a birth, allowing that record to be linked with the record of the infant's mother and (for most babies) father. This study is based on all 371,933 mothers whose first and second births were recorded in the registry from 1967 through 1989, excluding mothers with twins or other multiple births. Restricting the study to the first two births minimizes interpretational problems related to selective fertility, which increase with successive births5.
Classification of Birth Defects
The Medical Birth Registry of Norway records birth defects identified during the initial hospitalization, which lasts a minimum of five days. Birth defects were grouped according to the three-digit codes of the International Classification of Diseases, 8th Revision (ICD-8), with minor modifications. All registered birth defects except congenital dislocation of the hips were considered in the analysis. Hip dislocation was excluded because of the vague diagnostic criteria for this condition6 and its high prevalence. For most infants, only a single defect was reported, and it was assumed to be an isolated defect. When infants had multiple reported defects, we treated them as single defects. When spina bifida was present with anencephaly, only anencephaly was counted; similarly, when hydrocephalus was present with spina bifida, only spina bifida was counted. All other multiple defects (in 5.6 percent of affected first infants) were combined as a separate category (see below). There was a separate category for isolated cleft palate. Down's syndrome was also categorized separately, whereas other syndromes (chromosomal or mendelian) were pooled in another category. This system of classification resulted in a total of 24 mutually exclusive categories, 23 containing isolated defects and 1 containing multiple defects.
Number of Recurrences
There were 13,607 women in the registry whose first infant had a defect at birth; 9192 of these women had a second infant. These 9192 women formed the index-case group, for which we estimated the risk of a birth defect in the second infant. A similar defect in a second infant was defined as a defect with the same ICD code as the defect in the first infant. A dissimilar defect in the second infant was defined as a defect with an ICD code that did not overlap with the ICD code of the defect in the first infant. Of the second infants with defects, 7.2 percent had more than one recorded defect. If one of the multiple defects in the second infant had the same ICD code as the defect in the first infant, we counted it as a recurrence of a similar defect.
Multiple Defects in the First Infant
The 538 mothers whose first infant had more than one defect (but no known syndrome) were treated as a separate group and included in the calculation of combined relative risks of recurrence. In this group, a defect in the second infant was counted as a recurrence of a similar defect if it was the same as any one of the defects in the first infant. If the defect (or defects) in the second infant differed from all the defects in the first infant, it was included in the category of dissimilar defects.
Expected Number of Recurrences
Mothers whose first infants had no defects were the reference group. Thus, the expected risk of any given type of defect was assumed to be the risk of that defect in the second infant of a woman whose first infant was normal. The observed-to-expected ratio is an estimate of the relative risk of a woman having a second infant with a similar or dissimilar defect, given the occurrence of a defect in the first infant.
Combined Relative Risks
The total risk of recurrence of a defect in the second infant that is similar to the defect in the first infant is a weighted average of the risks of similar defects across all the categories of defects. The total expected risk is a similarly weighted average of the expected risks. Thus, the combined relative risk of a similar defect is the ratio of these observed and expected total risks. The combined relative risk of a dissimilar defect was calculated in the same way.
Paternity
Information on the identity of fathers was incomplete for infants under 28 weeks of gestational age. Accordingly, sibling pairs in which one or both infants were under 28 weeks of gestational age were excluded from the analyses that incorporated data on fathers. Among the remaining sibling pairs, 80 percent were registered as having the same father, and 4 percent were registered as having different fathers; for the other 16 percent, identification of one or both fathers was lacking. The relative risk of a recurrence of a defect in sibling pairs with unknown fathers was close to that in sibling pairs with different fathers. Therefore, siblings with unknown fathers were grouped with siblings who had different fathers. The extent of misclassification is unknown.
Statistical Methods
Confidence intervals for the relative risk of recurrent birth defects in second infants were calculated for each defect category with the use of the Fortran implementation of the algorithm of Vollset et al.,7 with the assumption that odds ratios approximate relative risks. There is no simple method to calculate a confidence interval for the total relative risk of a similar or dissimilar defect. We developed an approximate method, a description of which is available from the National Auxiliary Publications Service (*). Statistical comparisons of relative risks between groups were based on chi-square tests. All tests were two-tailed. Analyses were performed with the LE program in the BMDP statistical software package8.
Results
Birth defects were recorded in 2.5 percent of first infants and 2.1 percent of second infants. The mothers of affected first infants were 2.4 times as likely as other women to have second infants with defects. The risk of a similar or dissimilar defect in the second infants is shown in Table 1 for all defects that recurred in at least three pairs of siblings. There were six such categories, accounting for more than two thirds of all the defects. (A table showing the risk of recurrence for all 23 categories of isolated defects is available from *.) The estimated relative risks for the most commonly recurring defects in the second infants were 31.4 for cleft lip, 11.3 for limb defects, 7.3 for clubfoot, and 4.9 for defects of the genitalia. The risks of dissimilar defects in second infants after each of these defects in first infants were 1.2, 2.4, 1.4, and 1.5, respectively.
|
We examined genetic and environmental influences on the risk of recurrence in the following ways. First, we looked at paternal inheritance. On average, full siblings share half their genes, whereas half-siblings share only one quarter of their genes. There were 1734 mothers who were classified as having changed their partners between their first and second pregnancies. Among the women who had the same partner for both pregnancies, the relative risk of a similar defect in the second infant was 8.8, whereas among the women who changed partners, the relative risk was 6.2 (P = 0.10).
There may also be nongenetic factors at work in this comparison. Women who changed partners may also have made changes in their diet, residence, or other environmental factors that contributed to their risk of having an affected first infant. Therefore, we studied the effect of a change in the city (or municipality) of residence, which is recorded at each birth. To control for paternal influences, we first included only those women whose partners were the same for both infants. Among the women who remained in the same municipality for both pregnancies, the relative risk of a similar defect in the second infant was 11.6, whereas among women who changed municipalities between pregnancies, the relative risk was 5.1 (P<0.001) (Table 2). Among the women who changed partners, a change in municipality also affected the relative risk of a similar defect in the second infant, but to a smaller degree (7.3 among the women who did not move vs. 4.9 among those who did, P = 0.37). The overall reduction in risk after a change in residence was significant (P<0.001). After adjustment for a change of residence, the reduction in the risk of recurrence among women who changed partners was not significant. To determine how subtle problems of ascertainment or selection may have affected these trends, we calculated the risk of dissimilar defects among the same groups. No trends were apparent for either a change of residence or a change of partner.
|
Most birth defects have no known cause4. Possible explanations include polygenic inheritance,2 an interaction between genetic and environmental factors,9 and purely random mechanisms10. Given the etiologic uncertainty, it is striking to find a persistent risk of specific defects in the siblings of affected infants.
A combination of genetic and environmental factors may cause a persistent risk of similar defects in siblings. The risk of a similar defect was reduced somewhat if the two infants had different fathers. Although such a difference would be consistent with a genetic cause of the defects, it may reflect changes in environmental factors as well. However, among the women who did not change partners, the reduction in risk associated with a change in municipality must have been purely an environmental effect, if not one due to chance or bias.
The reduced risk of a similar defect after a change of residence may be explained in part by a lower rate of case ascertainment for the same type of defect in the new residential area. It seems unlikely, however, that a birth defect would be overlooked in the case of a mother who already had an infant with a similar defect. Clubfoot, which accounted for nearly a third of all defects, may be misdiagnosed more often than more serious defects. The exclusion of clubfoot from the analysis, however, did not weaken the effect of a change in residence.
There was also a slightly increased risk of a dissimilar defect in the second infant after the occurrence of any defect in the first. There are at least three possible explanations for this increased risk. First, there is a general underascertainment of defects11. Ascertainment of any defect in a second infant may be more likely if the mother has already had an infant with a birth defect. Second, particular clusters of seemingly dissimilar defects have been observed in individual infants12,13. If apparently different defects have some causative agent in common, they may likewise occur in sibships. A third explanation for the increase in the risk of a dissimilar defect in the second infant concerns the ability of women to abort abnormal fetuses spontaneously14,15. A woman with one affected infant may be less able than other women to abort defective fetuses and may therefore be at increased risk of carrying to term another fetus with any type of defect. We cannot distinguish among these hypotheses with our data.
The risk of recurrence may be affected by self-selection among women. We estimated the rate of second births among all the women who had had their first infant before 1985, providing at least five years of follow-up. Among the women with a normal first infant, 82 percent had a second infant. Among the women with a malformed first infant, 73 to 96 percent (mean, 79 percent) had a second infant, with the highest rates occurring among women whose first infant had one of the defects associated with the highest mortality rates. This reflects the strong tendency of women who lose their first child to become pregnant again5. The mean maternal age at the time of each birth did not differ between mothers of infants with birth defects and other mothers, making maternal age an unlikely confounder.
Most previous estimates of the risk of recurrence were based on surveys of families in treatment centers. Clinic-based estimates can easily be distorted by recall bias16 and selection bias. Population-based estimates are preferable in principle, because they are not subject to these problems. However, population-based registries depend on routinely collected birth records, a requirement that limits the specificity, consistency, and completeness of diagnoses. With a double-sampling approach, ascertainment of Down's syndrome, central nervous system defects, hypospadias, and facial clefts in the Medical Birth Registry of Norway was estimated to be 60, 90, 65, and 80 percent, respectively, for the period from 1985 to 198811. In addition, multiple defects are likely to be underestimated. Even if one accepts these limitations, there are still few population-based record systems suitable for estimating the risk of recurrent birth defects, because few of these systems are able to link infants to sibships. The Medical Birth Registry of Norway provides the largest data set assembled for the estimation of the relative risk of recurrence. In the Norwegian sample, mothers with affected first infants had a 2.2 percent chance, on average, of having second infants with the same defects. The numbers reported in the literature for various categories of defects typically range from 2 to 8 percent2.
One other population-based analysis has been published, using a slightly different categorization of defects and a much smaller number of births17. The relative risk of recurrence was 17.8 for a similar defect and 1.8 for a dissimilar defect. Confidence intervals were not provided, but randomness may account in part for the difference between these estimates and ours. Differences in reporting and categorization of defects, as well as in genetic background or duration of exposure to environmental factors, may also contribute to the differences between the two sets of estimates.
In summary, there is a strong tendency for specific birth defects to recur in families, indicating that specific, persistent causal factors are at work. Although genetic explanations have usually been preferred,18,19 we find strong, if indirect, evidence that environmental factors contribute to the familial risk of birth defects, suggesting that important environmental teratogens have yet to be discovered.
Supported by a grant from the Norwegian Research Council.
We are indebted to Dr. Donna Baird, Professor Helge Boman, Dr. Beth Gladen, Dr. John Harris, Professor Ivar Heuch, Professor Lorentz M. Irgens, Professor Bengt Kallen, Dr. Muin J. Khoury, Dr. Ruth Little, Dr. Andrew Rowland, Dr. Dale Sandler, Dr. Jack Taylor, Dr. Kristi Tolo, Dr. David Umbach, Dr. Stein Emil Vollset, and Dr. Clarice Weinberg for constructive comments.
Source Information
From the Medical Birth Registry of Norway and the Section for Medical Informatics and Statistics, University of Bergen, Bergen, Norway (R.T.L., R.S.); the National Institute of Public Health, Oslo, Norway (R.T.L.); and the Epidemiology Branch, National Institute of Environmental Health Sciences, Research Triangle Park, N.C. (A.J.W.).
Address reprint requests to Dr. Lie at the Medical Birth Registry of Norway, M.F.H. Bldg., N-5021 Haukeland Sykehus, Bergen, Norway, or to Dr. Wilcox at the Epidemiology Branch, National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Risk of Recurrence of Birth Defects
Hook E. B., Hughes-Davies T.H., Lie R. T., Wilcox A. J., Skjaerven R.
Extract |
Full Text
N Engl J Med 1994;
331:1309-1310, Nov 10, 1994.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
