Human Reproduction Vol.17, No.10 pp.

27622768, 2002

Fetal sex and preterm birth: are males at greater risk?

Jennifer Zeitlin1,4, Marie-Josephe Saurel-Cubizolles1, Jaques de Mouzon2, Lucile Rivera3,

Pierre-Yves Ancel1, Beatrice Blondel1 and Monique Kaminski1
1Epidemiological Research Unit on Perinatal and Womens Health, U149 INSERM, Paris, 2FIVNAT Registry and Epidemiological
Unit for Human Reproduction, U292 INSERM, Kremlin Bicetre and 3Service de Protection Maternelle et Infantile, Conseil General
de Seine-Saint-Denis, France
4To whom correspondence should be addressed. E-mail: zeitlin@cochin.inserm.fr

BACKGROUND: The existence of a male excess among preterm births is interesting because it could shed light
on the aetiology of preterm birth. Possible mechanisms are greater body weight, increased susceptibility to
complications of pregnancy, sex-linked biochemical processes and earlier conception in the fertile cycle. METHODS:
We measured the association between fetal sex and preterm birth in four original datasets, including a cohort of
births after IVF, and 20 populations extracted from published birthweight references. The original samples were
also analysed by mode of onset. RESULTS: There were more males among preterm and early preterm births than
among term births in most populations, including IVF births (odds ratio: 1.091.24). No male excess was observed
for two cohorts of black births, induced preterm births in the general population, and spontaneous onset births
after IVF. CONCLUSIONS: The proportion of male births declines with increasing gestation, even when time of
conception is known. This male excess appears to be strongest for spontaneous preterm births. Studying the sex
ratio of preterm births by medical risk factors may clarify why the male excess is absent in some populations. The
possibility that obstetric decision-making affects the sex ratio of indicated births must be considered.

Key words: IVF births/preterm birth/risk factors/sex ratio

Introduction production from androgen precursors or by interleukin-1 that

The greater mortality risks for males during pregnancy and
infancy are well known. Boys have higher rates of fetal
and neonatal mortality and are more vulnerable to long-
term neurological and motor impairments after preterm birth
(Khoury et al., 1985; Verloove-Vanhorick et al., 1994; Smith,
2000; Stevenson et al., 2000). In the past decade, several
studies using large datasets have found an excess in the
proportion of boys among preterm versus term births
(McGregor et al., 1992; Cooperstock and Campbell, 1996;
Astolfi and Zonta, 1999), building on earlier, less conclusive,
research on this subject. The higher proportion of preterm
births among males could contribute to an explanation for
their higher mortality in infancy. However, the question of the
male excess is interesting primarily because of the light that
it could shed on the aetiology of preterm birth.
Several mechanisms have been proposed to explain why
pregnancies carrying male fetuses could have a higher risk of
preterm birth: (i) heavier average body weight which increases
the probability of preterm labour (Hall and Carr-Hill, 1982;
McGregor et al., 1992); (ii) a greater susceptibility to certain
medical complications associated with preterm birth, such as
pregnancy-induced hypertension or infection (Campbell et al.,
1983; MacGillivray and Davey, 1985); and (iii) sex-linked
biochemical processes, including labour promoted by estrogen
may differ between male and female fetuses (Cooperstock and
Campbell, 1996). Another explanation, which is not based on
a greater male susceptibility to preterm birth, is that boys are
more likely to be conceived in the beginning of the fertile period
in contrast with girls who are associated with inseminations in
the middle of the cycle (James, 1994, 2000).
This analysis explores the association between fetal sex and
the risk of preterm birth in 24 populations of singleton births.
We test for the presence of a male excess among all preterm
births and early preterm births. The inclusion of births from
the French National IVF Registry makes it possible to test for
a male excess when time of conception is known with certainty.
We undertook further analyses within four samples by mode
of onset of the delivery. Explanations based on labour-
inducing processes predict a greater male excess for spon-
taneous births while those positing a greater susceptibility to
medical complications during pregnancy could lead us to
expect a greater excess among induced preterm births.
Materials and methods
This analysis used data from four original data sources, one European
and three French, as well as data extracted from published articles to
explore the sex distribution of preterm versus term births. All analyses
were undertaken using singleton pregnancies only. When possible,

2762 European Society of Human Reproduction and Embryology


stillbirths were excluded since the predominance of male fetuses
among stillbirths was not the subject of this analysis.
The EUROPOP study
The European Program of Occupational Risks and Pregnancy Outcome
(EUROPOP) was a European casecontrol study to explore the
relationship between occupational risk factors and preterm birth
(Saurel-Cubizolles et al., 1997). Data were from the Czech Republic,
Finland, France, Germany, Greece, Hungary, Ireland, Italy, The
Netherlands, Poland, Romania, Russia, Scotland (UK), Slovenia,
Spain and Sweden. Cases were all preterm singleton live and stillbirths
between 22 and 36 completed weeks of gestation born in participating
maternity hospitals. The control group included every 10th singleton
live birth or stillbirth at 37 completed weeks of gestation over the
same time period. Gestational age was based on obstetric estimates,
derived from last menstrual period and ultrasound measures. These
analyses used data on 5024 preterm and 8197 full-term singleton
livebirths.
French National Perinatal Surveys (FNPS)
The French National Perinatal Surveys collect data on all births with
a gestational age of 22 weeks or a birthweight of 500 g in France
during a 1 week period with the aim of monitoring the evolution of
indicators of perinatal health and medical practice (Foix-LHelias and
Blondel, 2000; Blondel et al., 2001). Data are collected through
interviews with the mother and from medical records. Gestational
age is based on the best estimate available in the medical records,
including the results of ultrasound scans. The total sample of 26 011
singleton live births combined data from the national survey in 1995
(a week in February, n 12 885) and the national survey in 1998 (a
week in November, n 13 126). Less than 1% of births had missing
data on gestational age or sex (n 246). The percentage preterm
was 4.6%.
Seine-Saint-Denis Experimental Health Certificates (SSD Registry)
This database included all births that occur in the district (departement)
of Seine-Saint-Denis in France. It was constituted from 8th day health
certificates (Certificats de Sante du 8eme jour), a certificate that must
be filled out for each infant before the 8th day of life. The data used
for this analysis included all live births that occurred between October
1998 and the end of December 1999. Sixty-three cases had missing
gestational age and 29 data on fetal sex. The total sample numbers
were 24 554 term and 1382 preterm singleton live births. The
percentage preterm was 5.3%.
National IVF Registry: Fecondation In Vitro National (FIVNAT)
FIVNAT was set up in January of 1986 and includes most French
IVF centres (de Mouzon et al., 1993; Fecondation In Vitro National,
1995). These centres participate in the registry on a voluntary basis.
Data are collected using three forms: one for infertility diagnosis and
the IVF cycle, one for thawed embryo transfers and one for obstetric
and paediatric data. These forms are completed by each centre and
centralized in the registry. Gestational age at birth was computed in
theoretical weeks of amenorrhoea by adding 14 days to the difference
between the date of oocyte retrieval and the date of delivery
(Fecondation In Vitro National, 1995). The data used for these
analyses were from 13 819 singleton live births resulting from IVF
from the registrys inception in 1986 until 1997. The preterm birth
rate for this cohort was 8.4%.
Birthweight reference populations
Datasets were constructed from birthweight reference populations
abstracted from published articles. A Medline search on combinations
Fetal sex and preterm birth: are males at greater risk?
of the following key words fetal/neonatal/birthweight and refer-
ences/norms/curves identified 54 articles with published birthweight
reference standards. Data were abstracted from articles where the
numbers of male and female births at each gestational age were
given. For these samples, it was possible to calculate the ratio of
males to females at each gestational age. All population-based samples
were selected for the analysis and hospital-based samples were also
included, with the exception of two samples that had 250 preterm
births. Three articles contained data on several reference populations:
in Canada (Arbuckle and Sherman, 1989), data were given for two
distinct periods and in two US samples data were presented separately
for white and black births (Amini et al., 1994; Roberts et al., 1996).
A published article on birthweight and the risk of stillbirth in Scotland
provided the number of live male and female births at each gestational
age and was also included (Smith, 2000).
The 20 birthweight reference populations, abstracted from 18
published articles, are listed and referenced in Table I, which shows
whether the sample is hospital- (H) or population- (P) based, the
gestational ages included in the studies, the sample sizes for preterm
and term births, and the percentage preterm. In general, these
birth cohorts have large sample sizes since this is necessary for
establishing birthweight percentiles (ranges: 65091 000 preterm
births and 41501 612 382 term births). With the exception of a
population from Denver and Norway, all births were live births.
Exclusion criteria for the studies are specified in Table I. All studies
eliminated infants with missing gestational age and birthweight and
used various exclusion criteria for birthweight outliers. In most
populations, only cases with missing data on birthweight or gestational
age or those with extreme values for these variables were excluded.
Many articles did not provide information on the percentage of cases
excluded and none of the articles provided information on the sex
ratio of excluded cases. Stricter exclusion criteria were used for the
Swedish analysis where 20% of births were excluded to obtain a
healthy population. In the earlier Scottish data sets, only legitimate
births were included. Hospital samples are more likely to include
higher risk populations. In the East Midland sample the authors
excluded all in-utero transfers to minimize this bias. They also
excluded pregnancies without a dating scan before 24 weeks as well
as congenital anomalies. In the sample from the hospital in Cleveland,
Ohio, however, both white and black births were high risk (15%
preterm birth rate). These exclusion criteria and the gestational age
limits affect the preterm birth rates, which ranged from 3 to 15%.
For the analysis, a subset of studies was defined that were population
based with minimal exclusion criteria. This subset included 10
birthweight reference populations plus the three original population-
based datasets.
Analysis
The analysis contrasted the sex distribution by preterm and term
status. The association between male sex and the risk of preterm
birth was expressed as an odds ratio (OR) with female sex as the
reference category. OR were also calculated for early preterm births,
defined as births before 33 weeks of gestation. Data on early preterm
births were available for all but four of the birth cohorts. 2-tests for
trend were done for the proportion male with rising gestational age
up until term. We used random effects meta-analysis to estimate a
combined measure of the association between fetal sex and preterm
birth in the 24 populations as well as the extent to which subgroups
explained heterogeneity in the pooled estimate. The latter analysis
relates study co-variates to the outcome, assuming a normal
distribution for the residual errors with both a within-study and an
additive between-studies component of variance. The within-study
variance (SE) was derived from each study and the overall variance
2763
2764

J.Zeitlin et al.
Table I. Populations of singleton live births from published articlesa

Location, year Reference Hospital (H) or GA Preterm Term % preterm Exclusion or inclusion criteria
population (P) n n (% excluded, if available)
sample

Denver, Colorado, Lubchenco et al. (1966) H (one centre) 2442 1485 4150 NAb Hospitalized preterm and all live term
USA, 19481961 white births. Gross pathological
conditions and missing GA excluded
(7%)
Aberdeen, Scotland, Thomson et al. (1968) P 3242 2575 43 786 5.6 Legitimate births only. Missing GA
19481964 (10%). Macerated stillbirths and
congenital anomalies excluded
Scotland, 19731979 Forbes and Smalls (1983) P 3242 7175 61 434 10.5 Legitimate births only
Glasgow, Scotland, Forbes and Smalls (1983) P 3242 2484 52 903 4.5 Legitimate births only
19751979
Norway, 19671998 Skjaerven et al. (2000) P 2445 90 825 1 612 382 5.3 Includes stillbirths
Canada, 1972 Arbuckle and Sherman (1989) P 2542 26 589 297 428 8.2 None mentioned
Cleveland, Ohio, USA Amini et al. (1994) H (one centre) 2444 White 3941 White 22 212 White 15.1 Excluded if discordance between
19751992 Black 3911 Black 20 983 Black 15.7 estimated GA and Dubowitz score
2 weeks (7.4%)
Germany, 19761985 Roemer et al. (1990) H (one centre) 2742 1288 16 224 7.4 Congenital anomalies excluded
Sweden, 19771981 Lawrence et al. (1989) P 2744 12 224 358 249 3.3 Excluded congenital malformations,
maternal infections, diabetes, other
disorders (20%)
Aberdeen, Scotland Campbell et al. (1993) P 3242 653 13 566 4.6 None mentioned
19791983
Scotland, 19801996 Smith (2000) P 2843 27 573 438 948 5.9 Includes first births only
Italy, 19841985 Parazzini et al. (1991) P 2842 64 842 1 101 591 5.6 None mentioned
France, 19841988 Mamelle et al. (1996) H 2844 6834 90 388 7.0 Elimination of extremes of weight
distribution
Canada, 1986 Arbuckle and Sherman (1989) P 2542 18 333 335 170 5.2 Excluded birthweights 500 g or
5000 g
East Midlands, UK, Wilcox et al. (1993) H (3 centres) 2442 2983 40 624 7.2 Excluded if no scan 24 wks (7.5%),
19861991 congenital anomalies and in-utero
transfers
Connecticut, USA, Roberts et al. (1996) P 2244 White 10 946 White 190 895 White 5.4 Missing GA (11%) and extreme
19881993 Black 3423 Black 26 009 Black 11.6 birthweights (0.5%)
Australia, 19911994 Roberts and Lancaster (1999) P 2244 38 513 695 517 5.2 Non-indigenous, Australian-born
women only
Canada, 19941996 Kramer et al. (2001) P 2243 39 491 636 418 5.8 Cases with discordant GA and birthweight
excluded based on statistical model

aTwo populations included stillbirths, as noted in inclusion criteria.

bTerm births were only included up to 1955.
GA gestational age; NA not available.
 Fetal sex and preterm birth: are males at greater risk?

Table II. Percentage male by preterm status and odds ratios (OR) for preterm birth associated with male sex

Data source Term All preterm births (37 weeks) Early preterm births (33 weeks)
% male
(n) % male OR 95% CI % male OR 95% CI
(n) (n)

EUROPOP, 19941997 52.1 55.4 1.14 1.061.22 55.8 1.15 1.041.30

(8197) (5024) (1375)
FNPS, 19951998 51.0 54.2 1.13 1.011.27 53.2 1.09 0.821.14
(24 583) (1182) (186)
SSD Registry, 19981999 51.1 54.5 1.14 1.031.27 52.1 1.04 0.831.31
(24 552) (1370) (292)
FIVNAT Registry, 19861997 51.3 53.4 1.09 0.961.23 53.3 1.08 0.831.14
(12 652) (1167) (229)
Birthweight reference populations
Denver, 19481961 50.6 55.4 1.22 1.081.37 55.9 1.24 1.071.44
Aberdeen, 19481964 51.4 55.3 1.17 1.081.27 NA
Scotland, 19731979 51.7 54.2 1.10 1.051.16 NA
Glasgow, 19751979 51.2 53.6 1.10 1.151.19 NA
Norway, 19671998 51.2 55.8 1.21 1.191.22 55.9 1.21 1.171.24
Canada, 1972 51.4 52.6 1.05 1.021.08 53.7 1.10 1.031.17
Cleveland, 19751992 (white) 51.4 53.7 1.10 1.031.18 54.7 1.14 1.031.27
Cleveland, 19751992 (black) 50.4 48.8 0.94 0.871.00 51.7 1.05 0.941.18
Germany, 19761985 50.9 53.7 1.12 1.001.26 53.3 1.10 0.871.39
Sweden, 19771981 50.8 55.4 1.20 1.161.25 56.3 1.25 1.131.35
Aberdeen, 19791983 51.5 56.8 1.24 1.061.45 NA
Scotland, 19801996 51.2 54.9 1.16 1.131.19 55.3 1.18 1.111.25
Italy, 19841985 51.3 53.6 1.09 1.081.11 53.5 1.09 1.051.13
France, 19841988 50.9 51.8 1.04 0.991.09 52.3 1.05 0.951.17
Canada, 1986 51.0 55.6 1.21 1.171.24 56.6 1.25 1.171.34
East Midlands, UK 19861991 51.1 54.0 1.12 1.041.21 56.5 1.24 1.041.48
Connecticut, 19881993 (white) 51.0 54.4 1.14 1.101.19 53.8 1.12 1.021.22
Connecticut, 19881993 (black) 51.0 49.9 0.97 0.901.04 52.5 1.07 0.941.22
Australia, 19911994 51.3 55.0 1.16 1.141.19 55.9 1.20 1.151.26
Canada, 19941996 51.1 54.9 1.16 1.141.19 54.9 1.16 1.131.18

CI confidence interval.

was estimated by an iterative procedure, using an estimate which was
based on restricted maximum likelihood. Analyses were done using
the meta and metareg commands in STATA v6.0.
Further analyses were undertaken according to whether the birth
was spontaneous or induced in the EUROPOP sample and the three
French datasets. The category of induced births includes induced
deliveries and Caesarean sections before the onset of labour. The
comparison group was all term births. In the EUROPOP study and
the French national perinatal surveys, 99% of all births had
information on mode of onset. In the Seine-Saint-Denis registry,
however, 8.5% of the births had missing information on mode of
onset. Missing data were more frequent in preterm deliveries (11.4
versus 8.4%), but did not differ by sex (8.5% for pregnancies with
boys versus 8.7% for pregnancies with girls). For the FIVNAT
registry, 7% of the births have missing information on the onset of
labour. Missing data are not more frequent among preterm versus
term births (6.9 and 7.0% respectively), nor among pregnancies with
male or female fetuses (7.2 and 6.8% respectively).
Results
Table II displays male births as a percentage of total births
for term, preterm and very preterm births as well as correspond-
ing OR for male sex in 24 populations of singleton births.
Sample sizes for the birthweight reference populations are
presented in Table I. Almost everywhere, there was a
higher percentage of males among preterm compared with
term births. OR in 20 out of the 24 populations ranged from
1.09 to 1.24. This excess risk translated into preterm birth
rates 0.51% higher for boys versus girls as, for example, in
Norway in 19671998 (4.8% for girls and 5.8% for boys),
Canada in 1986 (4.7 and 5.6%), Italy in 19841985 (5.3 and
5.8%), France in 19951998 (4.3 and 4.9%), Scotland in 1980
1996 (5.5 and 6.3%), Australia in 19911994 (4.9 and 5.6%),
and in Canada in 19941996 (5.4 and 6.2%).
In four samples, the confidence interval for the OR included
one. For the first of these, the FIVNAT registry, the OR was
within the lower limit of effects observed in other populations:
1.09. For the other samples, however, a hospital-based sample
in France in 19841988, and samples of black births (one
population-based and one hospital-based) from Cleveland and
Connecticut, OR were notably lower: 1.04 0.94, 0.97. For a
Canadian population in 1972 the OR is 1.05 although the
confidence interval does not include one.
The male excess existed for early preterm birth in most
samples; OR associated with early preterm birth were almost
identical to those within the whole sample. Two exceptions
are the French National Perinatal Surveys and the Seine-Saint-
Denis Registry where the OR for early preterm birth were
lower than the OR for all preterm births and were not
significantly different from 1.
Figure 1 plots the percentage of male births by gestational
age group starting with the group 3336 weeks gestation. We
omitted early preterm births since this information is not
2765
J.Zeitlin et al.
Figure 1. Percentage of male births by gestational group.
available for all samples, and is presented in Table II. Figure
1 illustrates a common pattern of decreasing proportions of
male births with increasing gestational age. The trend was
significant in all samples with OR for male sex significantly
greater than 1. The data from the sample of IVF births also
adhere to this pattern (large diamond-shaped plots) and the
declining percentage of boys with gestational age was signi-
ficant (2-test for trend, P 0.001). This trend was also
significant for the Canadian 1972 sample (P 0.001) and the
French 1988 sample (P 0.001), the two samples with lower
OR for the association between fetal sex and preterm birth. In
contrast, the trend was not significant for the samples of black
births (presented as dashed lines)
A combined estimate of the association between male sex
and preterm birth from the 24 studies was calculated using
random effects meta-analysis. The combined measure was
1.12 (1.091.15) and was highly significant. The test for
heterogeneity was significant (P 0.0001). Two populations
of black births contributed significantly to the observed hetero-
geneity (Z 4.542, P 0.0001); if these two populations
were excluded, the combined measure was 1.14 (1.111.17).
The test for heterogeneity remained significant (P 0.0001).
The combined estimate for the subgroup of population-based
studies with minimal exclusions was similar to the overall
estimate: 1.13 (1.101.17). Whether the studies had significant
versus minimal exclusions did not contribute to an explanation
of the heterogeneity. For early preterm births, the combined
estimate was 1.14 (1.101.18).
Table III presents results by mode of onset using the four
datasets where this information was available. The risk of
preterm birth associated with carrying a male fetus was higher
for spontaneous preterm births in the three samples of naturally
conceived births (OR: 1.17 for EUROPOP; OR: 1.29 for the
French National Perinatal Surveys; OR: 1.29 for the Seine
Saint-Denis registry). In fact, male sex was not significantly
related to the risk of induced preterm birth in any of these
samples. In contrast, the opposite pattern emerges from the
sample of IVF births: the excess risk associated with male sex
was apparent only in induced preterm deliveries and there was
no excess of boys among spontaneous preterm births.
The analysis of early preterm births in Table III showed
that for the two French samples (National Perinatal Surveys
and the SSD Registry), the overall OR associated with male
sex (shown in Table II) represented an average of a strong
2766
association with spontaneous preterm births and none for
induced preterm births. In these samples, about one-half of
early preterm births were induced. In the FIVNAT sample, a
male excess was evidentalthough not statistically signi-
ficantfor spontaneous early preterm births.
Discussion
Males have a greater risk of being born before term in a wide
range of populations across time and space. OR associated
with male sex were between 1.09 and 1.24 in 21 out of 24
populations. Results of other studies on this topic from New
England, Aberdeen and Italy are within this range [calculated
crude OR: 1.16, 1.20 and 1.11 respectively (Hall and Carr-
Hill, 1982; McGregor et al., 1992; Astolfi and Zonta, 1999)]
as are those from studies including fetal sex as a co-variate in
risk factor analyses [adjusted OR of 1.10 and 1.28 (van den
Berg and Oechsli, 1984; Shiono and Klebanoff, 1986)]. The
combined estimate of the OR for preterm birth from the 24
birth cohorts is 1.12 (1.091.14). Large sample sizes provide
high precision for individual estimates and there is significant
heterogeneity in this association over a relatively narrow range
of values.
The risk associated with fetal sex for early preterm births
is of a similar magnitude. Two exceptions are the French
samples from the late 1990s, where the OR for early preterm
births are lower and not significantly different from 1. In all
but two populations, the proportion of male births declines
with gestational age up until ~40 weeks gestation.
These data sources come from Europe and North America
over the last 50 years. The 20 birthweight reference populations
were selected from all referenced articles that provided data
to calculate the percentage of male births at each gestational
age. These studies used varying exclusion criteria. Overall,
exclusions on medical and social criteria were likely to result
in a healthier population of births. However, limiting the
analysis to birth samples with minimal exclusions did not
change the overall association between fetal sex and preterm
birth.
The risk of preterm birth was not affected by fetal sex for
the two populations of black births included in this analysis:
neither OR was significantly different from 1 and the overall
trend with gestational age was not significant. This corroborates
findings from a previous study that found a 7.2% excess of
males among white singleton preterm births versus only a
2.8% excess among black singleton births in a New England
sample of births (Cooperstock and Campbell, 1996). To explain
their finding, the authors hypothesize that the causes of preterm
birth in high-risk black populations, such as infection or pre-
eclampsia, could be sex independent. They conclude that a
labour-inducing mechanism influenced by fetal sex hormones
is most likely responsible for the male excess among preterm
births and that this excess would be most epidemiologically
evident in women at low risk of preterm birth (Cooperstock
and Campbell, 1996). This interpretation is consistent with
rates of preterm births observed in the sample from Connecticut,
where the pretem birth rate is much lower for whites than
blacks (5.4 versus 11.6%), but not with the sample from the
 Fetal sex and preterm birth: are males at greater risk?

Table III. Percentage male by preterm status and odds ratios (OR) for preterm birth associated with male sex, by mode of onset

Term Preterm Preterm Early preterm Early preterm

births
% male Spontaneous Induced Spontaneous Induced OR Spontaneous Induced Spontaneous Induced
(n) % male % male OR (95% CI) (95% CI) % male % male OR (95% CI) OR (95% CI)
(n) (n) (n) (n)

EUROPOP Study 52.1 56.0 54.0 1.17 (1.081.27) 1.08 (0.961.20) 56.0 55.8 1.17 (1.031.33) 1.16 (0.981.36)
(8197) (3384) (1627) (1144) (658)
French National 51.0 57.3 48.4 1.29 (1.111.49) 0.90 (0.741.10) 60.7 46.9 1.48 (0.961.80) 0.85 (0.5711.27)
Perinatal Surveys (FNPS) (24 583) (759) (417) (89) (96)
SeineSaint-Denis 51.1 57.3 51.1 1.28 (1.111.48) 1.00 (0.831.21) 58.6 47.9 1.35 (0.981.87) 0.88 (0.621.26)
Registry (SSD) (24 552) (784) (444) (152) (121)
FIVNAT Registry 51.3 51.6 55.7 1.01 (0.861.20) 1.19 (1.001.43) 54.1 52.6 1.12 (0.731.72) 1.06 (0.731.53)
(12 652) (579) (508) (85) (114)

CI confidence interval.

Cleveland, Ohio hospital where preterm birth rates are identical
for white and black births (15.1 and 15.7% respectively). No
further information is available on the medical risk factors
associated with preterm birth in this hospital; analyses to see
if these differ by race would be needed in order to confirm or
reject the hypothesis put forth by Cooperstock and Campbell.
Births from the French registry of IVF births were included
in this analysis to test the theory that boys have a greater
probability of being conceived earlier in the fertile cycle.
Support for this theory comes from studies of other mammalian
species and small studies on the sex of human offspring and
the cycle day of natural insemination (James, 1994, 2000). We
posited that if the excess of males was due solely to a difference
in the timing of conception, it should disappear completely in
IVF births for which the moment of conception is known with
certainty for both sexes. This does not appear to be the case
in the overall population of IVF births. Although the association
between fetal sex and preterm birth is of lesser magnitude, the
same population trends between the proportion of male births
and gestational age are observed.
Our analyses of spontaneous versus induced onset of delivery
in the four original data sets were intended to shed further
light on underlying mechanisms. A greater male excess for
spontaneous onset preterm births would add support to an
explanation based on labour-inducing processes associated
with fetal sex, whereas a greater male excess among medically
indicated preterm births would favour an explanation based
on a greater male susceptibility to certain complications of
pregnancy, such as hypertension or growth restriction. A
Scottish study from the early 1980s previously documented a
higher proportion of males in spontaneous versus induced
preterm births, leading the authors to posit that the labour-
inducing effects of fetal sex hormones or the effects of fetal
weight were most likely causal mechanisms (Hall and Carr-
Hill, 1982). One limit of analyses based on mode of onset is
that they do not separate out preterm births associated with
preterm premature rupture of membranes (PROM). These
can have either a spontaneous or medically decided onset,
depending on how the PROM was managed.
An excess of males in preterm births following spontaneous
onset of labour was clearly observed in the EUROPOP sample
and the two French population-based samples. No significant
excess was present for induced births. These results favour an
explanation based on a labour-inducing mechanism. Our results
for the IVF sample are not in line with the three other
samples, however. An excess of males is not apparent among
spontaneous IVF preterm birthsthe excess of males is
observed only for induced births. While it is possible that
there may be different medical risk factors associated with
spontaneous and induced preterm birth in IVF and naturally
conceived births, these conflicting results raise the question of
how to interpret the sex ratio of deliveries with a medically
decided onset.
To draw conclusions about the natural sex ratio from these
results, we must assume that medical decisions do not affect
the sex ratio. While fetal sex is unlikely to determine the
decision to terminate a pregnancy before term, it is related to
birthweight, which is a key parameter in these decisions. For
instance, obstetric decision-making could have the effect of
attenuating the natural sex ratio of indicated preterm births if
single sex reference curves were used for the diagnosis of
intrauterine growth restriction. Results from the two French
population-based samples from the late 1990s provide some
support for this possibility. No male excess is observed for
early preterm birthsin contrast to almost all other birth
samples. Almost half of all early preterm births result from a
medical decision to induce delivery and the low OR for male
sex reflects an average of a strong male excess for spontaneous
preterm births and no male excess for induced preterm births.
On the other hand, an excess of males among indicated
preterm births could occur if heavier babies were more likely
to be considered above the threshold where the adverse
consequences of continued pregnancy outweigh those associ-
ated with preterm birth. This would be more likely to occur
for closely monitored pregnancies and for moderate preterm
births. This is the case for the IVF births: about half of preterm
IVF births result from a medical decision, in contrast with a
third for naturally conceived births, and the excess of males
2767
J.Zeitlin et al.
for induced IVF births is evident only among preterm births
after 33 weeks of gestation.
Although the results of analyses by mode of onset provide
most support for a labour-inducing mechanism, the possibility
that medical decisions could affect the sex ratio of preterm
births prevents us from completely rejecting the other hypo-
theses. For instance, in the absence of medical decisions to
induce delivery, would we find no excess of males among IVF
births, as posited by the early conception hypothesis? Similarly,
could induction practices in naturally conceived populations
make it impossible to observe a male excess among those
infants whose delivery is medically decided?
In conclusion, these results provide strong evidence that
boys are more likely to be born before term. This effect is
observed in a wide range of populations, is evident among
early preterm births, and appears to be strongest for
spontaneous preterm births. To shed light on the mechanisms
underlying this effect, and in particular to clarify the reasons
that no male excess is evident in certain groups of preterm
births, future analyses should explore the sex ratio of
preterm births by aetiology, for both spontaneous and
induced deliveries. The possibility that obstetric decision-
making affects the sex ratios of induced births needs to be
taken into consideration when analysing the sex ratios of
births by mode of onset.
Acknowledgements
The EUROPOP study was financed by the European Union BIOMED
project BMH1-CT94-1041. The authors wish to thank the members
of the EUROPOP group. The French National Perinatal Surveys were
partly funded by the Ministry of Health (Direction generale de la
sante). The Seine-Saint-Denis experimental birth certificates are
financed by the Conseil General de la Seine-Saint-Denis. The authors
wish to thank the maternal and child protection services for their
collaboration. French National Registry on IVF is organized as a
collaboration between the FIVNAT association and U292, INSERM
and is partly funded by the pharmaceutical industry (Laboratoire
Organon) and the French Ministry of Health.
References
Amini, S.B., Catalano, P.M., Hirsch, V. and Mann, L.I. (1994) An analysis of
birth weight by gestational age using a computerized perinatal data base,
19751992. Obstet. Gynecol., 83, 34252.
Arbuckle, T.E. and Sherman, G.J. (1989) An analysis of birth weight by
gestational age in Canada. Can. Med. Assoc. J., 140, 157160, 165.
Astolfi, P. and Zonta, L.A. (1999) Risks of preterm delivery and association
with maternal age, birth order, and fetal gender. Hum. Reprod., 14,
28912894.
Blondel, B., Norton, J., du Mazaubrun, C., Breart, G. pour la coordination
national des Enquetes Nationales perinatales (2001) Evolution des principaux
indicateurs de sante perinatale en France metropolitaine entre 1995 et 1998.
J. Gynecol. Obstet. Biol. Reprod., 30, 552564.
Campbell, D., Hall, M., Lemon, J., Carr-Hill, R., Pritchard, C. and
Samphier, M. (1993) Clinical birthweight standards for a total population
in the 1980s. Br. J. Obstet. Gynaecol, 100, 43645.
Campbell, D.M., MacGillivray, I., Carr-Hill, R. and Samphier, M. (1983)
Fetal sex and pre-eclampsia in primigravidae. Br. J. Obstet. Gynaecol., 90,
2627.
Cooperstock, M. and Campbell, J. (1996) Excess males in preterm birth:
interactions with gestational age, race, and multiple birth. Obstet. Gynecol.,
88, 189193.
2768
de Mouzon, J., Bachelot, A. and Spira, A. (1993) Establishing a national
in vitro fertilization registry: methodological problems and analysis of
success rates. Statist. Med., 12, 3950.
Fecondation In Vitro National (1995) Pregnancies and births resulting from
in vitro fertilization: French national registry, analysis of data 1986 to 1990.
FIVNAT (French In Vitro National). Fertil. Steril., 64, 746756.
Foix-LHelias, L. and Blondel, B. (2000) Changes in risk factors of preterm
delivery in France between 1981 and 1995. Paediat. Perinat. Epidemiol.,
14, 314323.
Forbes, J.F. and Smalls, M.J. (1983) A comparative analysis of birthweight
for gestational age standards. Br. J. Obstet. Gynaecol., 90, 297303.
Hall, M.H. and Carr-Hill, R. (1982) Impact of sex ratio on onset and
management of labour. Br. Med. J. (Clin. Res.), 285, 401403.
James, W.H. (1994) Cycle day of insemination, sex ratio of offspring and
duration of gestation. Ann. Hum. Biol., 21, 263266.
James, W.H. (2000) Why are boys more likely to be preterm than girls? Plus
other related conundrums in human reproduction. Hum. Reprod., 15,
21082111.
Khoury, M.J., Marks, J.S., McCarthy, B.J. and Zaro, S.M. (1985) Factors
affecting the sex differential in neonatal mortality: the role of respiratory
distress syndrome. Am. J. Obstet. Gynecol., 151, 777782.
Kramer, M., Platt, R., Wen, S., Joseph, K., Allen, A., Abrahamowicz, M.,
Blondel, B. and Breart, G. (2001) A new and improved population-based
Canadian reference for birth weight for gestational age. Pediatrics, 108, 17.
Lawrence, C., Fryer, J.G., Karlberg, P., Niklasson, A. and Ericson, A. (1989)
Modelling of reference values for size at birth. Acta Paediat. Scand. Suppl.,
350, 5569.
Lubchenco, L.O., Hansman, C. and Boyd, E. (1966) Intrauterine growth in
length and head circumference as estimated from live births at gestational
ages from 26 to 42 weeks. Pediatrics, 37, 403408.
MacGillivray, I. and Davey, D.A. (1985) The influence of fetal sex on rupture
of the membranes and preterm labor [letter]. Am. J. Obstet. Gynecol., 153,
814815.
Mamelle, N., Munoz, F. and Grandjean, H. (1996) Croissance fetale a partir
de letude AUDIPOG. I. Etablissement des courbes de reference. J. Gynecol.
Obstet. Biol. Reprod., 25, 6170.
McGregor, J.A., Leff, M., Orleans, M. and Baron, A. (1992) Fetal gender
differences in preterm birth: findings in a North American cohort. Am. J.
Perinatol., 9, 4348.
Parazzini, F., Cortinovis, I., Bortolus, R. and Fedele, L. (1991) [Standards of
birth weight in Italy]. Ann. Ostet. Ginecol. Med. Perinat., 112, 203246.
Roberts, C., Mueller, L. and Hadler, J. (1996) Birth-weight percentiles by
gestational age, Connecticut 19881993. Conn. Med., 60, 131140.
Roberts, C.L. and Lancaster, P.A. (1999) Australian national birthweight
percentiles by gestational age. Med. J. Aust., 170, 114118.
Roemer, V.M., Buhler, K. and Kieback, D.G. (1990) Gestationszeit und
Geburtsgewicht. I. Mitteilung:Intrauterine Wachstumskurven. Z.
Geburtshilfe Perinatol., 194, 241253.
Saurel-Cubizolles, M.J., Renzo, G.C.D., Breart, G. and the EUROPOP Group
(1997) Womens work and preterm birth: epidemiological knowledge and
description of a European project. Prenat. Neonat. Med., 2, 161180.
Shiono, P.H. and Klebanoff, M.A. (1986) Ethnic differences in preterm and
very preterm delivery. Am. J. Publ. Health, 76, 13171321.
Skjaerven, R., Gjessing, H. and Bakketeig, L. (2000) Birthweight by gestational
age in Norway. Acta Obstet. Gynecol. Scand., 79, 440449.
Smith, G.C. (2000) Sex, birth weight, and the risk of stillbirth in Scotland,
19801996. Am. J. Epidemiol., 151, 614619.
Stevenson, D.K., Verter, J., Fanaroff, A.A., Oh, W., Ehrenkranz, R.A.,
Shankaran, S., Donovan, E.F., Wright, L.L., Lemons, J.A., Tyson, J.E. et al.
(2000) Sex differences in outcomes of very low birthweight infants: the
newborn male disadvantage. Arch. Dis. Child., 83, F182185.
Thomson, A.M., Billewicz, W.Z. and Hytten, F.E. (1968) The assessment of
fetal growth. J. Obstet. Gynaecol. Br. Commonw., 75, 903916.
van den Berg, B. and Oechsli, F. (1984) Prematurity. In Bracken, M. (ed.),
Perinatal Epidemiology. Oxford University Press, New York, pp. 6985.
Verloove-Vanhorick, S.P., Veen, S., Ens-Dokkum, M.H., Schreuder, A.M.,
Brand, R. and Ruys, J.H., (1994) Sex difference in disability and handicap
at five years of age in children born at very short gestation. Pediatrics, 93,
576579.
Wilcox, M., J.Gardosi, Mongelli, M., Ray, C. and Johnsom, I. (1993) Birth
weight from pregnancies dated by ultrasonography in a multicultural British
population. Br. Med. J., 307, 588591.
Submitted on April 2, 2002; accepted on May 31, 2002