Human Reproduction vol.15 no.1 pp.

210217, 2000

Birth weight in pre-eclamptic and normotensive twin

pregnancies: an analysis of discordance and growth
restriction

Sergio Ferrazzani1, Annamaria Merola,

Sara De Carolis, Brigida Carducci,
Giancarlo Paradisi and Alessandro Caruso
Department of Obstetrics and Gynaecology, Catholic University,
School of Medicine, 00168 Rome, Italy
1To

whom correspondence should be addressed

The aim of this study was to verify whether twin pregnancies complicated by pre-eclampsia were associated with a
higher rate of inter-twin weight discordance or an increased
prevalence of small for gestational age (SGA) neonates than
in normotensive twin pregnancies. A 17 year retrospective
study was undertaken by examining 76 twin pregnancies
complicated by pre-eclampsia and comparing them with
400 normotensive twin pregnancies. The case notes were
reviewed in reference to birth weight differences, birth
order, pregnancy outcome and inter-twin birth weight
discordance. Statistical analyses were performed with t-test,
contingency tables, regression curves, rank sum test and
non-parametric survival plots. Power analysis was also
carried out. Pre-eclamptic twin pregnancies were delivered
at similar weeks of gestation to normotensive. They resulted
in a smaller size for the second twin the earlier the delivery
week, while in normotensive twin pregnancies no significant
difference occurred at any week. Twin pregnancies complicated by pre-eclampsia showed higher rates of SGA neonates
among second twins than those with normal pressure. The
>25% discordance was associated with lower gestational
age at delivery in each group [mean (range) 33 weeks (27
38) versus 37 (2941), P < 0.005 pre-eclampsia and 35
weeks (2541) versus 38 (2542), P < 0.001 normotensive].
In pre-eclampsia the concomitant occurrence of SGA
second twin and the discordance >25% was associated with
shorter gestation while the presence of SGA second twin
alone was not.
Key words: birth weight discordance/growth restriction/preeclampsia/twin pregnancy

Introduction
It is well known that singleton pregnancies complicated by
pre-eclampsia are characterized more frequently by low birth
weight and small for gestational age (SGA) neonates than
normotensive pregnancies (Ananth et al., 1995). This increased
prevalence of low birth weight has also been reported in twin
pregnancies complicated by pre-eclampsia (McMullan et al.,
1984; Long and Oats, 1987), although some authors did not
find any difference in the rate of low birth weight compared
210

with normotensive twin pregnancies (Campbell, 1995). Recent

studies have shown that birth weight patterns in twin pregnancies should be analysed as inter-twin weight discordance
(Blickstein and Lancet, 1988), rather than as absolute weight.
In fact, in a twin pregnancy, the increase of weight discordance
between the two twins is associated with a worsening of
perinatal outcome (Blickstein and Lancet, 1988). This is
probably a result of the higher rate of SGA neonates (OBrien
et al., 1986). Therefore, it could be argued that the co-existence
of two conditions, such as pre-eclampsia and twin pregnancy,
characterized by high rate of low birth weight, could determine
an increased perinatal risk.
The present study was carried out to verify whether twin
pregnancies complicated by pre-eclampsia are associated with:
(i) negative fetal outcome in terms of low birth weight
and preterm delivery; (ii) higher rate of inter-twin weight
discordance; and (iii) higher prevalence of SGA compared
with normotensive twin pregnancies.

Materials and methods

A 17 year retrospective study of twin pregnancies complicated by
pre-eclampsia in comparison with normotensive twin pregnancies
was carried out at the Department of Obstetrics and Gynaecology at
the Catholic University of Rome, covering the period January 1, 1980
to December 31, 1996. A total of 661 consecutive twin births were
taken into consideration in a total of 47 664 deliveries. Among the
twin pregnancies, 76 complicated by pre-eclampsia were studied and
400 normotensive pregnancies were considered as controls. The
remaining 185 twin pregnancies were excluded due to the following:
gestational hypertension (n 135), major fetal malformations or
aneuploidies (n 16), delivery before 25 weeks gestation (n 4),
uncomplicated chronic hypertension (n 6), renal disease (n 2),
proteinuria without hypertension (n 7), incomplete information or
uncertain gestational age (n 4), HELLP (haemolytic anaemia,
elevated liver enzymes, low platelet counts) syndrome without hypertension (n 1) and twintwin transfusion syndrome (n 10).
Gestational hypertension was defined as a diastolic blood pressure
90 mmHg on two or more consecutive occasions, at 6 h apart,
developing after 20 weeks of gestation in a previously normotensive
patient. Pre-eclampsia was considered when gestational hypertension
was associated with proteinuria. Chronic hypertension was hypertension documented before pregnancy or before 20 weeks of gestation.
Proteinuria was defined as one urine collection with a total protein
excretion 300 mg/24 h or 1 g/l (or with dipstick) in a random
sample, without urinary infection. Blood pressure was taken with a
standard mercury sphygmomanometer, using phases one and four of
the Korotkoff sounds before delivery for systolic and diastolic blood
pressure, respectively, and recording with the patient in a semirecumbent position.
The study population was of ethnic uniformity due to the large
European Society of Human Reproduction and Embryology

Birth weight discordance in twins

predominance of white Italians (97%). During the time period covered

by this study, no policy of special antenatal clinical care was enforced
for twin gestations other than encouraging bed rest. Across all
subjects, the indications for planning delivery were: severe preeclampsia according to ACOG (American College of Obstetricians
and Gynecologists, 1986), presence of fetal distress as indicated by
fetal heart rate recording, oligo-anhydramnios, premature rupture of
membranes, cholestasis in presence of a mature lecithin/sphingomyeline ratio, chorionamnionitis or any other general obstetric indication.
The finding of intrauterine fetal discordance was not an indication
for planned delivery if not associated with other signs of fetal distress.
In case of intrauterine demise of one fetus, delivery was immediately
planned after 27 weeks gestation, when placentation was monochorial,
or as soon as lecithin/sphingomyeline ratio was more than 2 when
the placentation was dichorial. Corticosteroids were used to accelerate
fetal lung maturation in both pre-eclamptic and normotensive twin
pregnancies.
For the purpose of data analysis, information was extracted from
the records and entered into a personal computer database. Information
included aspects of prenatal care, previous medical history, complications, intrapartum care, delivery, fetal outcome and post-partum
course. Data entries were then accumulated and organized in tabular
form on a spreadsheet database. Case notes on the twins were
reviewed with reference to birth weight differences, birth order and
sex combination. In each case, the membrane separating the amniotic
sacs had been examined to determine whether it was monochorionic
or dichorionic. In the case of same sex twins and fused placentae,
the membranes separating the amniotic sacs were microscopically
examined by our pathologists to exclude the presence of chorionic
tissue. Although examination of the placentae and membranes is not
the most accurate way to diagnose specific twin type, the margin of
error was considered negligible and other methods were not used.
Where clinical and ultrasound evidence of twintwin transfusion
syndrome was available, a difference in haemoglobin concentration
of 5 g/dl at birth was considered as corroborative when the placental
type was monochorionic.
Eighty-nine per cent of all pregnancies during the study period
under consideration were evaluated with routine ultrasound in the
early second trimester. No correction was made when significant
differences (more than 7 days) were not evident in comparing
gestational age between ultrasound and last menstrual period. A case
was excluded from analysis in the absence of ultrasound or when
this was only performed near delivery with uncertain information on
gestational age.
SGA was diagnosed as a birth weight lower than the 10th percentile
according to either a national standard curve for singleton births
(Gagliardi et al., 1975) or a recently elaborated standard curve for
twin births from white population (Cohen et al., 1997).
Inter-twin birth weight difference was reported as weight discordance. Twin A was defined as first born and twin B as second born.
The absolute weight discordance (AWD) between two twins was
defined as heavier twin minus lighter twin and expressed in grams.
The percentage weight discordance (PWD) was defined as (heavier
twin lighter twin)100/(the heavier of the two) and expressed as
a percentage. Signed percentage weight discordance (SPWD) was
defined as (twin A twin B)100/(the heavier of the two twins) and
expressed as signed percentage. The latter, unlike other measures of
weight discordance, indicates which twin is heavier according to
delivery order. The twin B/twin A SGA rate ratio was calculated to
allow a direct comparison of frequencies between the two groups of
twin pregnancies.
Statistical analysis included the Students t-test for comparison of
averages, the Wilcoxon rank sum test for comparison of medians

when the distribution of the data was not normal and the 2 or
Fishers exact test when appropriate for comparison of frequencies.
P values 0.05 (two-tailed) were considered significant. A power
analysis was performed for statistical inference. The value reported
in the tables represents the power of the statistical test used to
detect the observed difference between two sample mean values or
frequencies. This value is derived from assessing 0.05 and a
sample size of 76 when the t-test is used (Glanz, 1987) or a 5.3 ratio
(400/76) when the 2 is applied (Marascuilo and McSweeney, 1977).
Power determination in regression analysis was obtained according
to Cohen (Cohen, 1977) by using the Fisher z transformation for the
critical value of r and for sample r. Survival statistics and plots were
obtained by using the KaplanMeier non-parametric method and the
P value was obtained through the log rank (MantelCox). Multiple
comparisons with Bonferroni correction were used to isolate differences between groups. All computations were performed on an Apple
Macintosh Performa 6400/200 with appropriate software (Microsoft
Excel, Abacus StatView).

Results
Table I shows some main clinical characteristics and pregnancy
outcomes in the two groups. No significant differences regarding height, weight before pregnancy or maternal age were
observed. The rate of nulliparas was 19% higher in preeclamptic twin pregnancies as compared to normotensive.
Systolic and diastolic blood pressures were higher in preeclamptic twin pregnancies, as expected. Delivery week rate
of premature delivery (under 37 weeks) and rate of low
birth weight neonates were similar in both groups. The
Caesarean section rate was higher in pre-eclamptic pregnancies, as expected. Twin pre-eclamptic women had a higher
prevalence of SGA neonates. Stillbirth prevalence showed no
significant difference between pre-eclamptic and normotensive
twin pregnancies. In pre-eclamptic twin pregnancies, stillbirth
involved both twins in two cases, twin A in one case and twin
B in two cases. In normotensive twin pregnancies stillbirth
occurred in both twins in two cases, twin A in five cases and
twin B in 11 cases. The involvement rate of twin B was double
that of twin A in each group.
Table II shows some detailed birth weight features in the
two groups. No significant differences were observed between
the two groups of pregnancies in regard to birth weight
according to delivery order. However, a higher proportion of
SGA neonates in pre-eclamptic pregnancies involved the
second twin. This pattern remained significant even after
exclusion of stillbirths (P 0.03). There was also a significantly higher inter-twin absolute (1.3 times), percentage (1.4
times) and signed percentage weight discordance (3.7 times)
in the pre-eclamptic group than in the normotensive. These
findings were observed despite similarities in total birth weight
and in the monochorionic placental type distribution between
the two groups.
Figure 1 shows that in pre-eclamptic twin pregnancies there
was a significant negative correlation between inter-twin signed
percentage weight discordance and the week of delivery. This
correlation was even more significant after exclusion of cases
of intrauterine death (r 0.33, P 0.005, power
0.80). On the other hand, the same regression line drawn in
normotensive twin pregnancies was not significantly different
211

S.Ferrazzani et al.

Table I. Main maternal clinical features and pregnancy outcome in twin pregnancies

Height (cm)
Weight before pregnancy (kg)
Age (years)
Nulliparas n (%)
Highest SBP (mmHg)b
Highest DBP (mmHg)b
Week of delivery
median (range)
Delivery 37 weeks n (%)
Caesarean section n (%)
Birth weight 2500 g n of
neonates (%)
Birth weight 10th percentile n
of neonates (%)c
Birth weight 10th percentile n
of neonates (%)d
Stillbirth n of neonates (%)

Pre-eclamptic
(n 76)

Normotensive
(n 400)

Powera

161.3 6.9
59.6 9.1
30.0 5.1
40 (66.7)
148.4 15.9
94.4 12.4
37 (2741)

NS
NS
NS
0.001
0.001
0.001
NS

162.1 6.5
58.9 9.0
29.8 4.4
167 (47.6)
118.1 11.7
72.6 9.5
37 (2542)

0.12
0.08
0.06
0.85
1.00
1.00

47 (61.8)
57 (75.0)
94 (61.8)

NS
0.001
NS

201 (50.3)
202 (50.5)
433 (54.2)

0.40
0.98
0.37

61 (40.2)

0.05

251 (31.4)

0.52

25 (16.7)

0.05

78 (10.6)

0.52

7 (4.6)

NS

20 (2.5)

0.26

Where not otherwise specified, values are expressed as mean SD; NS not significant; SBP systolic
blood pressure; DBP diasystolic blood pressure.
aSee text for details.
bValues without anti-hypertensive therapy and before delivery.
cAccording to singleton growth curve.
dAccording to twin growth curve (one pre-eclamptic and 19 normotensive pregnancies were excluded
because out of the limits of the reference curve, 2841 weeks).

Table II. Some birth weight features of twin pregnancies

1st born twin (g)

1st twin 10th percentile n (%)b
1st twin 10th percentile n (%)c
2nd born twin (g)
2nd twin 10th percentile n (%)b
2nd twin 10th percentile n (%)c
Total birth weight (g)
Inter-twin absolute weight
difference (g)
Inter-twin % weight difference
Inter-twin signed % weight
difference
Monochorionic placentae n (%)

Pre-eclamptic
(n 76)

Normotensive
(n 400)

Powera

2322 645
24 (31.6)
9 (12.0)
2133 739
37 (48.7)
16 (21.3)
4454 1303
396 323

NS
NS
NS
NS
0.03
0.05
NS
0.01

2348 636
117 (29.3)
31 (8.1)
2290 662
134 (33.5)
47 (12.2)
4638 1231
306 278

0.06
0.03
0.17
0.30
0.40
0.49
0.15
0.48

17.9 16.3
8.9 22.6

0.01
0.01

13.1 12.6
2.4 18.0

0.63
0.55

21 (27.6)

NS

83 (21.0)

0.22

Where not otherwise specified, values are expressed as mean SD; NS not significant.
aSee text for details.
bAccording to singleton growth curve.
cAccording to twin growth curve (one pre-eclamptic and 19 normotensive pregnancies were excluded
because they were out of the limits of the reference curve, 2841 weeks).

from the equality line between twin pairs, even after excluding
intrauterine deaths (r 0.0004, P 0.99, power 0.97).
Figure 2 reports the cumulative survival plots for the
gestational duration related to four degrees of percentage
weight discordance (25%, 2515%, 1510% and 10%).
As shown, in both groups of twin pregnancies, the discordance
25% was associated with shorter gestational duration. This
association was particularly evident in pre-eclamptic twin
pregnancies where the survival function of the discordance
25% was significantly different from all the other functions.
Conversely, in normotensive twin pregnancies the survival
function of the discordance 25% was only different from
discordance 10%. The median (range) weeks of gestation at
212

delivery in women with inter-twin weight discordance 25%

when compared to those with discordance 25% was 33 (27
38) versus 37 (2941), P 0.005 and 35 (2541) versus
38 (2542), P 0.001, in pre-eclamptic and normotensive
pregnancies respectively.
The distribution of the four degrees of discordance was not
significantly different between the two groups (Table III).
Figure 3 shows percentage distribution of pregnancies with
various combinations of SGA neonates according to the degree
of inter-twin percentage weight discordance in both groups.
Using both singleton and twin growth curves, the rate of
pregnancies with SGA twin B rose as the degree of discordance
increased, while the rate of pregnancies with both twins

Birth weight discordance in twins

Figure 1. Regression lines of signed percent birthweight discordance (SPWD) drawn against duration of gestation.

Figure 2. Survival analysis. Cumulative plot for duration of gestation in relation to four arbitrary degrees of percent weight discordance
(PWD). Only significant results are reported.

Table III. Distribution of the four degrees of discordance between preeclamptic and normotensive women
Degrees of
discordance

Pre-eclamptic

Normotensive

25%
2515%
1510%
10%
Total

15
17
14
30
76

47
79
81
193
400

(19.7)
(22.4)
(18.4)
(39.5)
(100.0)

(11.8)
(19.7)
(20.3)
(48.2)
(100.0)

Frequencies (%) are reported. 2 4.5; P NS.

adequate for gestational age (AGA) declined as the degree of

discordance increased (according to singleton growth curve:
pre-eclamptic, 2 20.1, P 0.02; normotensive, 2 98.9, P
0.001; according to twin growth curve: pre-eclamptic, 2 22.5,
P 0.01; normotensive, 2 96.9, P 0.001). On the other
hand, in the two graphs plotted using singleton growth curve

the degree of discordance appeared to have no effect on the

rate of pregnancies with SGA twin A or with both SGA twins.
In contrast, in the two graphs plotted using twin growth curve,
the greater the degree of discordance, the higher was the rate
of pregnancies with all the combinations of SGA.
Figure 4 describes the twin B/twin A SGA rate ratio as a
relative measure of the involvement of twin B over twin A.
According to the singleton growth curve, in normotensive twin
pregnancy, a clear involvement of twin B over twin A was
reached with a degree of weight discordance of 2515%. In
pre-eclamptic twin pregnancies, this involvement was already
evident with a relatively low degree of weight discordance of
1510%. According to the twin growth curve, in either
pre-eclamptic and normotensive twin pregnancies, a clear
involvement of twin B over twin A was reached with a degree
of discordance of 2515%. In both graphs, twin B was
constantly more frequently involved than twin A in preeclamptic than in normotensive twin pregnancy.
213

S.Ferrazzani et al.

Figure 3. Percent distribution of various combinations of small for gestational age (SGA) neonates according to the degree of inter-twin
percentage weight discordance in the two groups, according to singleton and twin growth curves. Twin A is defined as first born and twin B
as second born.

Figure 4. The histograms describe twin B/twin A small for gestational age (SGA) rate ratio as a relative measure of the involvement of
twin B over twin A. Only significant results are reported.

Figure 5 gives cumulative survival plots for gestational

duration in pregnancies with various combinations of SGA
neonates among twin pairs. With pre-eclampsia no significant
differences were observed among the survival functions. In
normotensive pregnancies with SGA in both twins delivery
was significantly closer to term, with respect to that of
214

pregnancies with AGA in both twins. This finding was present

only when SGA was defined according to the singleton
growth curve.
Figure 6 shows survival plots for the gestational duration
in relation to pregnancies with concomitance of both SGA
twin B and discordance (according to the various degrees).

Birth weight discordance in twins

Figure 5. Survival analysis. Cumulative plots for duration of gestation in presence of various combinations of small for gestational age
(SGA) twin pairs according to singleton and twin growth curves. Only significant results are reported.

Pre-eclamptic pregnancies with SGA twin B in the context

of discordance 25% were significantly related to shorter
gestational duration than pregnancies with AGA twin B. This
association was more evident in pre-eclampsia when the
definition of SGA was made according to twin growth curve.
In normotensive twin pregnancies there was no significant
difference among the survival functions.
Discussion
In the present study, twin pregnancies complicated by preeclampsia gave birth at similar weeks, having similar rates of
low birth weight neonates and stillbirths as compared to the
normotensive twin pregnancies. However, pre-eclamptic twin
pregnancies were subject to a higher rate of SGA neonates.
The use of individualized growth curves has shown that twins
are capable of achieving a normal growth potential (Xu et al.,
1985; Simon et al., 1989). Consistently, accelerated post-natal
growth of low birth weight twins suggests that in-utero
malnutrition is a result of limited maternal potential to nurture
multiple fetuses (McKeown and Record, 1952; Naeye, 1964;
Leveno et al., 1980). Given the equivocal consensus on whether
to evaluate growth restriction in twins on the basis of singleton
or twin standards, it was decided to use both singleton
(Bronsteen et al., 1989; Nellson, 1989) and twin growth curves
for the comparison of twin birth weight.
The higher rate of SGA neonates in pre-eclamptic twin

pregnancies suggested more detailed analysis of the birth

weight characteristics in each twin pregnancy. Despite a total
birth weight difference of about 200 g between the two study
groups (not significant), some significant differences were
found in inter-twin weight discordance however expressed
(whether as absolute or percent or signed percent). Like other
studies (Taylor et al., 1998), in normotensive twin pregnancies
no significant differences were observed in the birth weight
within pairs at any week. While lack of significance could be
ascribed to a type II error, it still indicates that if a mean
size difference should exist, it would be negligible. In twin
pregnancies with pre-eclampsia this difference became clear.
It has previously been reported that early onset pre-eclampsia
in singleton pregnancies was associated with higher frequency
of SGA neonates (Ferrazzani et al., 1990). Similarly, in twin
pregnancies it was observed that early onset pre-eclampsia
was associated with a higher degree of percentage discordance
as a consequence of lower size in second twins. It is important
to note that in our sample the median time interval between
the first observation of proteinuria and delivery was 3 days
(range 025). So early onset pre-eclampsia was approximately
coincident with early birth.
Observed from a different point of view, the findings argue
that growth discordance higher than 25% was associated with
lower gestational age at delivery in both pre-eclampsia and
normotensive twin pregnancies. No other authors have
observed this relationship between growth discordance and
215

S.Ferrazzani et al.

Figure 6. Survival analysis. Cumulative plot for duration of gestation related to pregnancies with the concomitance of both small for
gestational age (SGA) twin B and discordance (according to the various degrees) according to singleton and twin growth curves. Only
significant results are reported.

gestational age (Blickstein et al., 1987; Taylor et al., 1998),

probably because they studied only normotensive twin pregnancies where this association is much less evident. Since this
was a retrospective study, it was not possible to control for
timing of delivery and factors that may have contributed to
the decision to deliver preterm twins. All twin pregnancies
during the study period, regardless of indication for delivery
were considered. To a certain extent we were able to minimize
some of these concerns by documenting a comparable gestational age distribution in both groups.
Twin pregnancies with more than 25% growth discordance
were delivered (or had to be delivered) earlier. The low
gestational age at delivery in the presence of severe percent
weight discordance could be related to the high rate of SGA
among second twins. This association was much more evident
in pre-eclampsia than in normotensive twin pregnancies, probably because in women with pre-eclampsia there is an increased
likelihood of the second twin being SGA. It is well known
that in twin pregnancies, second twin outcome is more jeopardized than the first (Friedman et al., 1977; Essel and Opai-Tetteh,
1994). The findings of this study confirm this assertion by
indicating that the rate of SGA and stillbirth increased in
second twins in both groups. However, even if not in a
significant manner, the stillbirth rate in pre-eclampsia was
almost double that in normotensive twin pregnancies. It seems
216

reasonable to consider this as relevant to the higher rate of

SGA among second twins in pre-eclampsia.
The findings of this study highlight that when both twins
are SGA without discordance, delivery is closer to term in
normotensive pregnancy. On the other hand, when both twins
are AGA without discordance, delivery is more distant from
term. The paradox is only apparent and only evident when
SGA is defined according to the singleton growth curve. In
fact, it is well known that the birth weight of an uneventful
twin pregnancy progressively diverges from singleton median
birth weight as it approaches term. In other words, neonates
born from an uncomplicated twin pregnancy at term are SGA
when compared to singleton standards. When the week of
delivery is earlier, birth weight may still be within normal
limits for singleton standards. The reason for a premature
delivery when both twins are AGA could simply be ascribed
to premature labour, the most frequent complication in twin
pregnancies. In pre-eclampsia there was no significant relationship between the occurrence of SGA neonates and duration of
gestation, probably because the maternal indication for timing
of delivery could have interfered with the natural evolution of
pregnancy.
In conclusion, the rate of inter-twin birth weight discordance
is similar between pre-eclamptic and normotensive pregnancies. In the presence of similar gestational weeks at delivery,

Birth weight discordance in twins

pre-eclampsia shows a higher rate of SGA among second

twins than in normotensive pregnancies. In both pre-eclamptic
and normotensive twin pregnancies, earlier delivery weeks
were related to birth weight discordance 25% or to the
concomitant occurrence of SGA second twin and discordance
25%, but not to the presence of SGA second twin alone. These
associations were more strongly evident in pre-eclampsia.
The term SGA, deriving from a statistical concept, does not
distinguish normal and healthy fetuses that have a weight
below the 10th percentile from those who are small because
of intrauterine malnutrition. The healthy SGA could be
defined as the neonate being constitutionally small with no
deprivation of in-utero nutrients, and the at-risk SGA could
be identified as the malnourished fetus that is small because
of in-utero deprivation of nutrients and oxygen, as a consequence of placental insufficiency. In pre-eclampsia placental
insufficiency is frequently observed and it can be associated
with endothelial activation and damage (Myatt et al., 1997;
Daniel et al., 1998). In the light of the findings by other
authors (Blickstein and Lancet, 1988), indicating that the
smallest discordant twin delivered at term has normal outcome,
the data presented here suggest that also in twin pregnancies
there are two types of SGA neonates: healthy (probably
constitutional) and at risk (malnourished). In general, the
healthy SGA neonate can be identified in twin pregnancies
where one (generally the first presenting) or both twins are
under the 10th percentile and pregnancy can end close to term.
The at-risk SGA neonate is to be found prevalently among
second twins when growth discordance is evident.
Prospective studies are needed to confirm the findings of
this study.

Gagliardi, L., Preve, C.U., Cordero di Montezemolo, C. et al. (1975)

Accrescimento intrauterino ed eta` gestazionale in un campione di 9774
casi. Ann. Ost. Gin. Med. Perin. Anno XCVI N, 3, 147157.
Glanz, S.A. (1987) Primer of Bio-statistics, 2nd edn. McGraw-Hill, New
York, pp. 144162.
Leveno, K.J., Santos-Ramos, R., Duenhelter, J.H. et al. (1980) Sonar
cephalometry in twin pregnancy: discordancy of the biparietal diameter
after 28 weeks gestation. Am. J. Obstet. Gynecol., 138, 615619.
Long, P.A. and Oats, J.N. (1987) Pre-eclampsia in twin pregnancy severity
and pathogenesis. Aust. N.Z. J. Obstet. Gynaecol., 27, 15.
Marascuilo, L.A. and McSweeney, M. (1977) Nonparametric and DistributionFree Methods for the Social Sciences. Brooks/Cole, Monterey, p. 556.
McKeown, T. and Record, R.G. (1952) Observations of fetal growth in
multiple pregnancy in man. J. Endocrinol., 8, 386401.
McMullan, P.F., Norman, R.J. and Marivate, M. (1984) Pregnancy-induced
hypertension in twin pregnancy. Br. J. Obstet. Gynecol., 91, 240243.
Myatt, L., Eis, A.L.W., Brockman, D.E. et al. (1997) Endothelial nitric
oxide synthase in placental villous tissue from normal, pre-eclamptic and
intrauterine growth restricted pregnancies. Hum. Reprod., 12, 167172.
Naeye, R.L. (1964) The fetal and neonatal development of twins. Pediatrics,
33, 546551.
Nellson, J.P. (1989) Fetal growth in twin pregnancies. Acta Genet. Med.
Gemellol., 37, 409416.
OBrien, W.F., Knuppel, R.A., Scerbo, J.C. and Rattan, P.K. (1986) Birth
weight in twins: an analysis of discordancy and growth retardation. Obstet.
Gynecol., 67, 483486.
Simon, N.V., Deter, R., Hassinger, K. et al. (1989) Evaluation of fetal growth
by ultrasonography in twin pregnancy. J. Clin. Ultrasound, 17, 633640.
Taylor, G.M., Owen, P. and Mires, G.J. (1998) Foetal growth velocities in
twin pregnancies. Twin Research, 1, 914.
Xu, B., Deter, R., Milner, L. and Hill, R. (1985) Evaluation of twin growth
status at birth using individualised growth assessment. J. Clin. Ultrasound,
23, 277286.
Received on May 17, 1999; accepted on September 27, 1999

References
ACOG (1986) American College of Obstetricians and Gynecologists Technical
Bulletin, Washington, DC, No. 91.
Ananth, C.V., Peedicayil, A. and Savitz, D.A. (1995) Effect of hypertensive
diseases in pregnancy on birthweight, gestational duration, and small-forgestational-age births. Epidemiology, 6, 391395.
Blickstein, I. and Lancet, M. (1988) The growth discordant twin. Obstet.
Gynecol. Surv., 43, 509515.
Blickstein, I., Shoham-Schwartz, Z., Lancet, M. et al. (1987) Characterization
of the growth-discordant twin. Obstet. Gynecol., 70, 1115.
Blickstein, I., Shoham-Schwartz, Z. and Lancet, M. (1988) Growth discordancy
in appropriate for gestational age, term twins. Obstet. Gynecol., 72, 582584.
Bronsteen, R., Govert, G. and Bottoms, S. (1989) Classification of twins and
neonatal morbidity. Obstet. Gynecol., 74, 98101.
Campbell, D.M. (1995) Pre-eclampsia in twin pregnancy (abst). Hypertens.
Pregn., 14, 94.
Cohen, S.B., Dulitzky, M., Lipitz, S. et al. (1997) New birth weight nomograms
for twin gestation on the basis of accurate gestational age. Am. J. Obstet.
Gynecol., 177, 11011104.
Cohen, J. (1977) Statistical Power Analysis for the Behavioural Sciences.
Academic Press, New York.
Daniel, Y., Kupferminc, M.J., Baram, A. et al. (1998) Plasma soluble
endothelial selectin is elevated in women with pre-eclampsia. Hum. Reprod.,
13, 35373541.
Essel, J.K. and Opai-Tetteh, E.T. (1994) Twin birth-weight discordancy in
Transkei. S. Afr. Med. J., 84, 6971.
Ferrazzani, S., Caruso, A., De Carolis, S. et al. (1990) Proteinuria and outcome
of 444 pregnancies complicated by hypertension. Am. J. Obstet. Gynecol.,
162, 366371.
Friedman, E.A., Sachtleben, M.R. and Friedman, L.M. (1977) Relative birth
weights of twins. Obstet. Gynecol., 49, 717720.

217