Lie RT, Wilcox AJ, Skjærven R. Survival and Reproduction Among Males With Birth Defects and Risk of Recurrence in Their Children. JAMA. 2001;285(6):755–760. doi:10.1001/jama.285.6.755
Author Affiliations: Section for Medical Statistics and Medical Birth Registry of Norway, University of Bergen, Norway (Drs Lie and Skjærven); and the National Institute of Environmental Health Sciences, Research Triangle Park, NC (Dr Wilcox).
Context Few systematic data exist on survival and reproduction among males with
birth defects and their contribution to occurrence of birth defects in the
Objective To estimate survival of males with registered birth defects, their subsequent
reproduction rates, and their risk of transmitting birth defects to their
Design and Setting Population-based cohort study of data from the Medical Birth Registry
Subjects A total of 486 207 males born in Norway between 1967 and 1982,
12 292 of whom had a recorded birth defect.
Main Outcome Measures Survival rates through 1992, reproduction rates through 1998, and risk
of birth defects among offspring of males with vs without birth defects.
Results Survival through 1992 was lower among males with birth defects (84%
vs 97%). Compared with males without birth defects, affected males were 28%
less likely to have had a child. Among offspring of affected males, 5.1% had
a registered birth defect compared with 2.1% of offspring of males without
birth defects (relative risk [RR], 2.4; 95% confidence interval [CI], 1.9-3.0).
Offspring of affected fathers had an increased risk of the same defect as
their fathers (RR, 6.5; 95% CI, 4.0-10.4) and an increased risk of dissimilar
defects (RR, 1.8; 95% CI, 1.3-2.5).
Conclusions Compared with unaffected males, males with birth defects have higher
mortality and survivors are less likely to have a child. Fathers with birth
defects are significantly more likely than unaffected fathers to have an affected
A father's contribution to his children's risk of birth defects is not
well established. If the father has a birth defect, he may pass on to his
children genes that increase their risk of a birth defect. However, the extent
to which children of affected males are at higher risk is unknown except in
a few specific types of birth defects. Furthermore, if a father has a particular
birth defect, it is unknown whether his offspring are at increased risk of
having any other kinds of birth defects.
We observed a cohort of nearly a half million males from birth to adulthood,
using a population-based registry in Norway. Males with registered birth defects
were compared with other males regarding survival, their probability of having
offspring, and risk of birth defects in their offspring. We have previously
reported similar data for a cohort of females with birth defects.1
The population-based Medical Birth Registry of Norway records all births
in Norway since 1967 (about 1.8 million births). All live births and stillbirths
at a gestational age of at least 16 weeks are included in the registry. There
were 486 207 live and stillborn male infants delivered in Norway between
1967 and 1982, which we defined as our study cohort. In our analyses, the
cohort was divided into those with a registered birth defect (n = 12 292
males [2.5%]) and the remaining males without birth defects (the reference
group). The cohort was followed up for survival through 1992, when the most
recent linkage with mortality records was carried out, and for reproduction
through September 1998.
The Medical Birth Registry of Norway includes unique personal identification
numbers for all births in Norway as well as the identification number of the
father and mother. These identifiers permit linkage of males born early in
the cohort with their offspring born later (1983-1998). This linkage does
not ensure 100% detection of the cohort's offspring. During the later period,
5.6% of the births lacked a father's identity and, therefore, were excluded
from this analysis. The lack of identified fathers was associated with fetal
death; the percentage of stillbirths that did not have an identified father
was 51%. Also, younger mothers were more likely to have children with an unidentified
father, which presumably led to selective loss of recorded births for younger
fathers. Finally, children born outside of Norway to Norwegian-born parents
are not captured by the registry (although outmigration during this period
was extremely low2). By 1998, the registry
had accumulated 110 327 births that could be linked to fathers born in
the study cohort.
The Medical Birth Registry of Norway records birth defects that are
diagnosed at delivery or by pediatric examination during the initial hospitalization
(≥5 days). In this study, 24 categories of birth defects were used, the
same as in 2 previous studies.1,3
Categories of birth defects were based on the 3-digit codes of the International Classification of Diseases, Eighth Revision (ICD-8), with minor modifications. Most of the birth defects included
are major; however, minor defects may be included in some categories and cannot
be distinguished by ICD-8 codes. For example, the
heterogeneous category of limb defects includes serious reduction deformities
and, possibly, minor soft-tissue syndactylies.
Most affected children had only 1 specific birth defect diagnosis. Cases
with multiple defects were pooled in a separate category, except that when
spina bifida was present with anencephaly, only anencephaly was counted, and
when hydrocephalus was present with spina bifida, only spina bifida was counted.
Isolated cleft palate was separated from the 3-digit ICD-8 code for cleft lip and palate as a distinct category. Similarly, Down
syndrome was separated as a category distinct from other syndromes.
The Medical Birth Registry of Norway, like other registries based on
routine medical birth records, does not capture all birth defects. Estimates
of ascertainment vary by defect category. For example, the Medical Birth Registry
of Norway captures an estimated 80% of cleft lip and an estimated 60% of Down
syndrome cases.4 The Norwegian Health Inspectorate
has instructed that therapeutic abortions should be reported to the registry
as stillbirths, including information on any diagnoses of birth defects. The
completeness of such reporting is not known, however.
In analyzing the risk of birth defects in children whose fathers also
had birth defects, we defined a similar defect as
one in the same ICD-8 category as the father's and other defects as all others.3
Males with birth defects were compared with males without birth defects.
Survival and reproduction were based on standard actuarial life-table methods
(6-month intervals). Reproduction among males with specific birth defects
was calculated as a ratio relative to males who had no reported birth defects,
with the inherent limitation of a slight underascertainment of offspring.
Estimates of reproduction are based on all males surviving to age 15 years.
The 95% confidence interval (CI) for relative reproduction between groups
was calculated using a Cox proportional hazards model.5
The expected number of birth defects in offspring of fathers with a
specific defect was based on risk of birth defects among offspring of males
without recorded birth defects. Relative risks (RRs) of recurrence of similar
or dissimilar birth defects (the observed-expected ratio [O/E]) were summarized
across all paternal defect categories using a stratified approach. Exact P values and 95% CIs were calculated using StatXact.6
The prevalence of registered defects at birth (including stillbirths)
was 2.5% in this cohort of 486 207 males. Eighteen percent (n = 85 574)
of survivors had at least 1 recorded offspring by September 1998 (Table 1). The percentage who had become
fathers increased with age, reaching 46% by age 28 to 31 years.
Figure 1A shows survival for males with and without birth defects. Males with a birth
defect had lower survival to age 20 years (84% compared with 97%, including
all male fetuses aged ≥16 completed weeks' gestation). Affected males had
higher mortality at all ages up to 14 years, with the highest RR in the first
year of life (Table 2).
If they survived, males with birth defects were less likely than other
males to have a child at any age (Figure 1B). Using a Cox regression model, the rate of reproduction among
surviving males with birth defects was estimated to be 72% of the reproduction
rate among unaffected males (95% CI, 68%-77%). Taking into account both the
higher mortality rates and the lower reproduction rates among survivors, a
male with birth defects was, on average, only 63% as likely to reproduce by
age 30 years as an unaffected male.
Figure 2 shows total survival
for each of the 24 birth defect categories and reproduction among males in
each category relative to unaffected males. The defects that were least likely
to cause death were also least likely to reduce a man's probability of having
a child if he survived. However, some birth defect categories with high survival
rates had substantially reduced reproduction rates and vice versa.
Of the 12 292 males with birth defects, 850 had a total of 1265
children. Sixty-four (5.1%) of these children had birth defects. Fathers without
birth defects had a total of 109 162 children, of whom 2326 (2.1%) had
birth defects. Thus, the overall risk of a birth defect was 2.4 times higher
among children of affected fathers (95% CI, 1.9-3.0). This higher risk of
birth defects did not differ by sex of the offspring (P = .79).
The excess contribution made by affected fathers to occurrence of birth
defects in the next generation was a combination of 2 factors: the proportion
of fathers who had birth defects and the increased risk of defects in their
children. Measured in attributable risk,7 affected
fathers contributed 16 of 1000 registered birth defects in the next generation.
Table 3 shows the risk of
similar and dissimilar birth defects among offspring of affected fathers.
Twenty-one children had the same defect as their fathers compared with an
expected number of 3.2. The most common recurring defects were cleft lip,
genitalia defects, limb defects, and clubfoot. The recurrence risk (O/E) was
38-fold for cleft lip, 3.8-fold for genitalia defects, and 12-fold for limb
defects. The recurrence risk for clubfoot was 3.4-fold but was not significantly
different from 1. The pooled RR for similar defects in children was 6.5 (95%
Forty-three children had defects different from the father's compared
with an expected number of 23.7. The O/Es for specific defects were all greater
than 1 and, for 2 categories (cleft lip and abdominal wall), this increase
was statistically significant. The pooled O/E for a dissimilar defect was
1.8 (95% CI, 1.3-2.5).
The Medical Birth Registry of Norway is a population-based registry
that permits follow-up of persons born since 1967. In a previous article,
we identified a cohort of females born between 1967 and 1982 and compared
the survival and reproduction of females with and without birth defects.1 In the present article, we provide similar data for
males with and without birth defects.
Males with birth defects had higher mortality than unaffected males
through infancy and childhood. This surprising persistence of mortality risk
presumably reflects the ongoing complications related to their defects.
Males with birth defects were 28% less likely than unaffected males
to have a child. This presumably reflects social as well as biological consequences
of their defects. The reduced probability of fathering a child varied substantially
across defect categories, with all but the small category of other central
nervous system defects having a reduced tendency to reproduce.
The total risk of birth defects was 5.1% among offspring of fathers
with defects and 2.1% among offspring of fathers without defects, for an RR
of 2.4. No single category of defect explained the higher risk in affected
As expected, the risk of birth defects in offspring was increased mainly
for the same type of defect as the father's. Fairly precise estimation of
the recurrence risk was possible for the most common types of defects. A recurrence
risk of the same defect is consistent with a shared genetic etiology, although
shared environmental causes cannot be ruled out. Offspring of affected fathers
also had an increased risk of dissimilar defects of all types. The explanation
for this excess is not as clear, and the possibility of bias must be considered.
The Medical Birth Registry of Norway, like other registries based on
routine medical birth records, does not capture all birth defects. Some are
simply underregistered at birth.4 Others, such
as defects of the heart or kidney, are often ascertained too late to be captured
by the registry. This is reflected in the relatively low number of such defects
in our data.
Underregistration creates an opportunity for bias. A father's birth
defect may make it more likely that a birth defect in his offspring would
be recorded in the medical record. If so, we would predict that, on average,
these recorded defects would tend to be milder. This was apparently not the
case. Among the 64 affected offspring of affected fathers, 9.4% died within
the first year of life, while among the 2326 affected offspring with unaffected
fathers, only 5.9% died in the first year. It appears that affected offspring
of affected fathers had more serious defects compared with affected offspring
of unaffected fathers. A tendency for more minor defects to be ascertained
in offspring of affected fathers is possible but could not be demonstrated.
Our previous analysis of affected mothers provides some further information
on the issue of selective diagnosis. Previously, we found no increase among
the offspring of affected mothers in the RR for dissimilar defects.1 An increased ascertainment of defects would presumably
occur among offspring regardless of which parent was affected. Thus, we observed
no evidence to support ascertainment bias as the reason for an excess of dissimilar
defects among offspring of affected fathers.
If male neonates are routinely inspected more closely for defects than
female neonates, this could lead to ascertainment of more minor defects for
males within each defect category. If such minor defects had a stronger tendency
to result in a dissimilar defect in offspring, this could contribute to a
higher risk in offspring of affected fathers. Some children with 1 registered
birth defect probably also have other unascertained birth defects. If such
underascertainment were more frequent for males, this also could lead to an
increased risk of apparently dissimilar defects.
Although the recurrence risk from father to offspring was substantial,
the number of affected fathers was too low to contribute many birth defects
to the next generation. Only 1.6% of birth defects in the second generation
could be attributed to a defect in fathers. Males with birth defects were
slightly more likely than other males to find a partner with a birth defect
(1.6% vs 0.9%). However, this difference is too small to explain the excess
risk of dissimilar defects. Of the 15 pairs of parents in which both had birth
defects, only 1 had an affected child (father, mother, and child all had cleft
Given the structure of the cohort (comprising all male births in 1967-1982),
only a small proportion of the members of the cohort have reached age 31 years.
Caution must be used in making longitudinal interpretations of these cross-sectional
data. A complete reproductive history will be required to make a more definitive
inference about male reproduction and risk of birth defects in offspring.
This study had incomplete ascertainment of offspring among males in
the cohort. During the period when these births occurred, nearly 6% of birth
certificates had no recorded father and, as a consequence, some males in the
cohort became fathers without evidence in the registry. Since the father's
information was lacking for half of stillbirths, and children with birth defects
are more likely than others to be stillborn, the total number of children
with birth defects born to the study cohort of males is probably underascertained.
Furthermore, offspring of affected fathers may be selectively underascertained
to the extent that these offspring more often have birth defects and, therefore,
more often are stillborn. The effect could be that we slightly underestimate
the fertility of affected fathers and, more importantly, that we may underestimate
the percentage of birth defects among their offspring.
For an unknown percentage of birth certificates, the recorded father
is not the biological father. Such errors would reduce the contrast in rates
of recurrence between affected and unaffected males.
Our estimates of recurrence risk do not take into account the slight
statistical dependence of outcomes within the family structure, ie, that some
fathers are represented with more than 1 child.8
While this may widen some CIs, the effects are minimal and should have little
or no impact on the risk estimates themselves. The results were similar when
analyses were restricted to only 1 (the first) child per father.
There are similarities but also apparent dissimilarities between these
data for males and the data for females reported previously.1
Birth defects generally were more prevalent among males than females. This
male excess of defects also has been observed in other cohorts.9
Infant and childhood mortality rates were lower for affected males than
for affected females (16% vs 20%).1 This counters
the usual survival advantage of females in infancy and childhood. The higher
mortality for affected females was not explained by any specific category
of defects. Given the higher rate of recorded defects among males, it is possible
that less severe manifestations of a given defect are more likely to be registered
for males than for females.
The 28% reduced reproduction tendency among affected males is very similar
to the reduction previously seen among affected females (27%).1
Given that reproduction is a more complex function for mothers than for fathers,
we might have expected a smaller reduction among affected males. Similarly,
if the excess birth defects among males reflect the selective recording of
a greater number of milder defects, we might have expected less reproductive
impairment among males than females. Neither was the case.
Affected males were slightly more likely than affected females to contribute
offspring to the next generation (63% vs 60%). Given that more males than
females had registered birth defects, there were substantially more affected
fathers than affected mothers contributing to the next generation. This result
was not fully apparent at the follow-up in 1998 because males tend to become
parents later in life than females, and the present data are truncated at
a relatively young age.
The RR for birth defects among offspring was significantly higher for
affected males (2.4) than for affected females (1.6) (test of homogeneity
of odds ratios, P = .03). This difference hypothetically
may be explained by a greater contribution of paternal genes to risk of birth
defects (genomic imprinting).
The pooled RR of a woman having a child with the same defect was 6.8,1 similar to the father's risk of 6.5. However, affected
mothers had no apparent increase in risk of offspring with other defects (RR,
1.0),1 in contrast with the substantial increase
in the father's risk of other defects (RR, 1.8). The higher risk of dissimilar
defects with affected fathers apparently accounts for the entire excess of
birth defects among offspring of affected fathers compared with offspring
of affected mothers. A syndrome could hypothetically be expressed as different
defects in different affected family members. It is unclear whether this could
explain a higher tendency of males to pass on a different defect to the next
There are few data in the literature that compare recurrence risk for
fathers and mothers, and these data are restricted to recurrence of the same
defect carried by the parent. Mothers have been reported to be more likely
than fathers to pass on a heart defect to a child.10,11
In a study of spina bifida, mothers of affected children were more likely
than fathers to report a family history of spina bifida.12
The authors interpret this as evidence of preferential transmission of these
defects through females, although this could also be because of more complete
reporting by mothers.
Affected males contributed more affected offspring than affected females,
not merely because of their increased risk of having affected offspring but
because there were more affected fathers than mothers. The attributable risk
of birth defects in the next generation from affected fathers was 3 times
that from affected mothers (1.6% vs 0.5%).1
The sum of 2 attributable risks is usually an overestimate of their total,
so the combined contribution from affected parents is presumably no more than
In this report, we have shown that males with birth defects have increased
mortality throughout childhood compared with unaffected males and that males
with birth defects are less likely to have a child. Furthermore, our data
suggest that fathers with birth defects are significantly more likely than
unaffected fathers to have children with birth defects. In many respects,
these findings are similar to previously published data for females. However,
affected fathers appear to contribute more birth defects than affected mothers
to the next generation.