Human Reproduction Vol.16, No.10 pp.

21242129, 2001
Risk factors for high-order multiple implantation after
ovarian stimulation with gonadotrophins: evidence from a
large series of 1878 consecutive pregnancies in a single
centre
Rosa Tur
1,3
, Pedro N.Barri
1
, Buenaventura Coroleu
1
, Rosario Buxaderas
1
,
Francisca Mart nez
1
and Juan Balasch
2
1
Service of Reproductive Medicine, Department of Obstetrics and Gynaecology, Institut Universitari Dexeus, and
2
Institut Clinic of
Obstetrics and Gynaecology, Faculty of Medicine-University of Barcelona, Hospital Cl nic-Institut dInvestigacions Biome`diques
August Pi i Sunyer (IDIBAPS), Barcelona, Spain
3
To whom correspondence should be addressed at: Service of Reproductive Medicine, Institut Universitari Dexeus. Pso Bonanova 67,
08017-Barcelona, Spain. E-mail: [email protected]
BACKGROUND: High-order multiple pregnancies (triplets or more) have a large adverse impact on perinatal
morbidity and mortality as well as important economic consequences. Most triplets and higher births are due to
ovulation induction alone or in combination with intrauterine insemination (IUI) rather than to in-vitro fertilization
(IVF). The present investigation was undertaken to determine whether there were specic variables that related to
patient clinical characteristics (age of the woman, duration of infertility, type of infertility, body mass index, basal
FSH and LH concentrations), treatment characteristics (initial dose of gonadotrophins, total dose of gonadotrophins
administered, number of days of ovarian stimulation, insemination procedure, number of spermatozoa inseminated
in patients undergoing IUI, type of luteal support), and ovarian response (oestradiol serum concentrations, number
and size of follicles) that might be associated with the occurrence of high-order multiple implantation in order to
develop a prediction model. METHODS: This study employed univariate, multivariate and receiver-operating
characteristic (ROC) analysis of a large series of 1878 consecutive pregnancies obtained in cycles stimulated with
gonadotrophins. Of them, 1771 (94.3%) were low-order pregnancies (1477 singletons and 294 pairs of twins) and
107 (5.7%) were high-order pregnancies. RESULTS: Predictive variables in the multivariate analysis were age of
the woman, serum oestradiol concentrations and number of follicles >10 mm on the day of HCG injection.
Stratication of the number of follicles into three categories (1 to 3, 4 to 5, and >5 follicles respectively), peak
serum oestradiol and womans age according to the ROC curves, showed that the risk of high-order multiple
implantation correlated signicantly with increasing total number of follicles and was signicantly increased in
women with a serum oestradiol >862 pg/ml and aged 32 years. CONCLUSIONS: This three-variable model can
help to identify patients at high-risk for high-order multiple pregnancy in ovulation induction cycles.
Key words: gonadotrophins/high-order gestation/multiple pregnancy/ovulation induction
Introduction
There is general agreement that the incidence of multiple
pregnancies has increased dramatically during the past 20
years in association with the modern management of infertility,
mainly the widespread use of ovarian stimulants and assisted
reproduction techniques (CDC, 2000; The ESHRE Capri
Workshop Group, 2000; Jones and Schnorr, 2001). Interest-
ingly, according to population-based studies, as much as two
thirds of iatrogenic multiple pregnancies, mainly high-order
gestations (triplets), may be attributable to ovulation-indu-
cing drugs without IVF or related techniques (Levene et al.,
1992; Derom et al., 1993; Evans et al., 1995; Corchia et al.,
2124 European Society of Human Reproduction and Embryology
1996; Wilcox et al., 1996), a gure that could be even higher
considering that compared with IVF programmes, no reporting
system on the use of ovulation-inducing drugs not associated
with IVF is available (CDC, 2000; Jones and Schnorr, 2001).
In fact, there has been greater control of IVF than ovulation
stimulation (Evans et al., 1995; Jones and Schnorr, 2001).
Therefore, identication of reliable predictors of multiple
pregnancy during ovulation induction cycles is clearly
necessary.
A number of previous studies have analysed risk factors for
multiple gestation in gonadotrophin induced cycles focusing
mainly on the major methods of monitoring follicular develop-
Multiple implantation after ovulation induction
ment, i.e. ovarian ultrasonography and determination of serum
oestradiol concentration. However, data regarding the predict-
ive value for multiple pregnancy of both methods of monitoring
ovarian response to gonadotrophin treatment have been clearly
contradictory. Thus, some reports favour the use of oestradiol
determination (Farhi et al., 1996; Valbuena et al., 1996;
Pasqualotto et al., 1999; Gleicher et al., 2000; Dickey et al.,
2001) but others disagree (Kurachi et al., 1985; Shelden et al.,
1988; Dodson et al., 1988; Navot et al., 1991b; Yu et al.,
1991; Ben-Nun et al., 1993; Goldenberg et al., 1994; Goldfarb
et al., 1997). Similarly, the usefulness of ovarian ultrasono-
graphy in identifying cycles with high-risk for multiple preg-
nancy is stressed in some reports (Dickey et al., 1991, 2001;
Navot et al., 1991b; Farhi et al., 1996; Valbuena et al., 1996;
Gleicher et al., 2000) but not all reports (Kurachi et al., 1985;
Dodson et al., 1988; Shelden et al., 1988; Yu et al., 1991;
Ben-Nun et al., 1993; Godenberg et al., 1994; Goldfarb et al.,
1997; Pasqualotto et al., 1999).
Furthermore, those previous studies have usually included
a limited number of multiple pregnancies which, in addition,
were twins in the great majority of cases. Thus, there is little
available data in the literature regarding risk factors for high-
order multiple pregnancy. Risks associated with multiple
pregnancy depend, obviously, on the size of multiple gestation,
with twin pregnancies carrying only a slightly increased risk
over singletons, while multiple gestations of triplets or more
represent the majority of the risk usually associated with
multiple birth (Sassoon et al., 1990; Seoud et al., 1992;
Gleicher et al., 1995; Berkowitz, 1996; Evans et al., 1998).
However, women with infertility rst and foremost demand
high rates of pregnancy, which may entail a relatively high
risk of low-order multiple pregnancy (Gleicher et al., 1995).
Finally, it has been stressed that an attempt to eliminate most
twin pregnancies after ovulation induction would imply a
signicant decline in the overall pregnancy rate (Goldenberg
et al., 1994).
Therefore, on the above evidence, the present investigation
was undertaken to determine whether there were specic
characteristics of the cycle in which conception occurred, the
treatment modality, or the patient, that might be associated
with the occurrence of high-order multiple implantation in
order to develop a prediction model. To this end, this study
employed retrospective, univariate, multivariate and receiver-
operating characteristic (ROC) analysis of a large series of
1878 consecutive pregnancies obtained in cycles stimulated
with gonadotrophins in a single centre.
Materials and methods
Patients
Between January 1988 and December 1998, 1878 intrauterine pregnan-
cies in 1781 women were obtained after gonadotrophin ovarian
stimulation or induction of ovulation without IVF at the Reproductive
Medicine Service, Department of Obstetrics and Gynaecology, Institut
Universitari Dexeus. In patients achieving more than one pregnancy
in different treatment cycles, each cycle was considered as an
independent event for the analysis of results. Treatment cycles leading
to biochemical pregnancies or ectopic gestations were excluded.
2125
Women were either anovulatory or were undergoing ovarian stimula-
tion on an empirical basis, usually in conjunction with intrauterine
insemination (IUI). Thus, of the 1878 conceptional cycles, 1493
resulted from IUI with husbands (n 1041) or donor frozenthawed
(n 452) spermatozoa and the remaining 385 followed ovulation
induction and timed intercourse.
Treatment protocols
All pregnancy cycles were treated with gonadotrophins. The prepara-
tion used was human menopausal gonadotrophin (HMG) (Pergonal;
Serono S.A., Madrid, Spain) in 1201 (64%) cycles and highly puried
follicle-stimulating hormone (FSH-HP) (Neo-Fertinorm; Serono S.A.)
in the remaining 677 (36%) cycles. The regimen of gonadotrophin
administration used was either the conventional step-up dose approach
characterized by initial daily doses of two ampoules of gonadotrophin
(containing 75 IU FSH each in the form of HMG or FSH-HP) which
is increased every 35 days until an ovarian response occurs (Wang
and Gemzell, 1980), or the chronic low-dose step-up regimen where
the starting dose of FSH is lower and followed by small weekly
increments (Balasch et al., 1996). The ovarian response was monitored
by serial vaginal ultrasonographic follicular measurements and serum
oestradiol determinations. Ovulation was triggered with the injection
of human chorionic gonadotrophin (HCG) (5000 IU i.m. Profasi;
Serono S.A.) when at least one leading follicle measuring 17 mm
in diameter was detected in association with a consistent rise in serum
oestradiol concentration. Current institutional guidelines recommend
the cancellation of cycles of gonadotrophin stimulation if 4 follicles
measuring 14 mm in diameter are present in association with
oestradiol serum concentrations 1000 pg/ml. However, there was
no strict adherence to these guidelines throughout the study period
because of a number of changes in the clinic policy over time.
Ovulatory HCG injection was followed by timed intercourse or
IUI as appropriate. For IUI only one insemination was performed
3640 h after HCG administration using husband or donor sperm
samples processed with Percoll gradients as reported previously (Pardo
et al., 1988).
The luteal phase was supported with two additional doses of
2500 IU HCG (administered 4 and 7 days after the HCG ovulatory
injection respectively) in 1088 (58%) cycles. 598 (32%) cycles were
given intravaginal micronized progesterone (300 mg/day until menses
occurred or until pregnancy was diagnosed), and no luteal support
was provided in the remaining 192 (10%) cycles. Pregnancy was
diagnosed by positive urine and/or blood tests and the subsequent
demonstration of at least one intrauterine gestational sac by trans-
vaginal ultrasonography at 6 weeks gestation. The order of the
multiple pregnancy was classied according to the highest number
of gestational sacs observed by ultrasound imaging, including
pregnancy sacs which did not contain an embryonic pole, and is
referred to in the text as a multiple conception. The subsequent
outcome of pregnancy was not considered for the specic purpose of
this study.
Hormone analysis and ultrasonography
Serum FSH, LH and oestradiol were determined by radioimmunoassay
(Nichols Institute, S.Juan Capistrano, CA, USA for FSH and LH;
Orion Diagnostica, Espoo, Finland for oestradiol). The inter- and
intra-assay coefcients of variation were 5.4 and 2.9% for FSH, 5.4
and 2.6% for LH and 10.2 and 9.7% for oestradiol respectively.
Pelvic ultrasound was performed using a Combison 320 (Kretztechnik,
Zips, Austria) with a 5 mHz vaginal transducer in patients treated
between 1988 and 1992, and with a 6 mHz vaginal transducer
attached to a Sonolayer SSA-270A (Toshiba Co., Tokyo, Japan) in
those treated between 1993 and 1999. Follicular size, as measured
R.Tur et al.
by ultrasound, was divided into three categories: leading follicles
(17 mm), intermediate follicles (14 to 17 mm), and small follicles
(10 to 14 mm). Follicle sizes are the average of two dimensions
measured from the outer wall of one side of the follicle to the inner
wall of the other and corresponding to the maximum diameters of
the follicle measured in both longitudinal and transverse scan planes.
Measurements and statistical analysis
Descriptive data are presented as means SEM. Variables related
to patients clinical characteristics, treatment characteristics and
ovarian response that have been proposed as potential predictive
factors of multiple pregnancy and that are readily available to the
clinician were used to develop the prediction model. Clinical variables
included age of the woman, duration of infertility, type of infertility
(primary or secondary), body mass index and basal (days 35 of a
spontaneous or induced menses in the three months preceding
treatment) FSH and LH concentrations. Treatment characteristics
included the following: initial dose of gonadotrophins; total dose of
gonadotrophins administered (in terms of ampoules of 75 IU FSH);
number of days of ovarian stimulation; insemination procedure (IUI
versus timed intercourse); number of spermatozoa inseminated in
patients undergoing IUI; and type of luteal support. Variables related
to ovarian response included serum oestradiol concentration and the
number and size of follicles detected by ultrasonography on the day
of the HCG injection.
We used a univariate logistic regression model to test the above
variables for their association with high-order multiple gestation.
Variables that were statistically signicant associated with high-order
multiple pregnancy were retained for testing in a multivariate logistic
regression model (Hosmer and Lemeshow, 1989). Receiver operating
characteristic (ROC) curves (Hanley and McNeil, 1982; Zweig and
Campbell, 1993) were used to determine the cut-off point which best
discriminated between high-order multiple pregnancy (3 gestational
sacs) and low-order gestation (12 gestational sacs) for those variables
remaining statistically signicant after being entered into the logistic
regression model. Then, the multivariate ordinal logistic regression
with the proportional-odds model was performed to determine the
extent to which these latter variables were associated with the number
of gestational sacs. The area under the ROC curve (AUC
ROC
) was
used as a quantitative measure of accuracy.
Data were analysed by Statistics Package for Social Sciences
(SPSS, Chicago, IL, USA) and MedCalc Software (Mariakarke,
Belgium).
Results
Of the 1878 clinical intrauterine pregnancies, 1477 (78.6%)
resulted from the conception of singletons, 294 (15.6%) of
twins, 74 (3.9%) of triplets, 23 (1.2%) of quadruplets, 7 (0.4%)
of quintuplets, 2 (0.1%) of sextuplets, and 1 (0.05%) of
septuplets. Low-order pregnancies thus made up 1771 (94.3%)
of the pregnancies and high-order pregnancies 107 (5.7%)
of them.
Association of the different variables investigated with
high-order pregnancy as shown by the univariate analysis is
presented in Table I. Of them, nine were independently
associated with high-order pregnancy. When these nine vari-
ables were entered simultaneously into the logistic regression
model, only three of them remained statistically signicant:
the age of the woman; the peak serum oestradiol; and the total
number of growing follicles (10 mm) on the day of HCG
2126
administration. Their -coefcients were 0.92, 1.0, and 1.04
respectively. The diagnostic accuracy (predictive value of high-
order multiple implantation) of these variables was analysed
further by the AUC
ROC
curves determined with ROC analysis.
AUC
ROC
curves (95% CI) for age, number of growing follicles,
and peak serum oestradiol were 0.60 (0.570.63), 0.65
(0.630.67), and 0.70 (0.670.72) respectively. The three
variables had similar predictive properties and analysis of the
groups created by the simultaneous evaluation of age and
growing follicles and oestradiol did not delineate groups
beyond those provided by individual variables.
However, through use of the estimated variables from the
logistic regression model, point estimates can be generated for
the probability of high-order multiple implantation and to
stratify across the three prognostic variables in order to develop
a prediction model using three readily accessible clinical
parameters. Thus, the cut-off points for age, total number of
growing follicles, and oestradiol that discriminated the best
between low-order and high-order pregnancy cycles were 32
years, 3 follicles, and 862 pg/ml respectively according to
the ROC curves. These cut-off points were used to stratify
the three predictive variables (age of the woman, oestradiol
concentrations, and total number of growing follicles)
for the probability of high-order pregnancy as presented in
Table II. Thus, for example, when the total number of follicles
is 45 and the oestradiol concentration is 862 pg/ml in a
woman aged 32 years, the risk of a high-order multiple
pregnancy is 4.3%. In contrast, with the same number of
follicles in a woman 32 years old having oestradiol con-
centrations 862 pg/ml, the risk increases 3-fold to as much
as 13% (Table II). Overall, Table II clearly shows that the risk
increases with increasing serum oestradiol concentrations, with
younger age of the woman and increasing total number of
growing follicles.
Discussion
Stimulation of ovarian follicular development is, at present,
the most widely used therapeutic modality for treatment of the
infertile couple and ovulation induction has become one of
the most successful areas for the practising reproductive
endocrinologist. However, the dramatic increase in multiple
pregnancy during the past two decades has corresponded with,
and is largely the result of, both the tremendous increase in
the use of gonadotrophins as well as a major modication in
the original philosophy of their use. Initially, these agents were
used exclusively for the treatment of anovulation, oligo-
ovulation or ovulatory dysfunction in a relatively small patient
cohort (the refractory members of a group of all infertile
women numbering ~15%), with a relatively low (~20%)
incidence of multiple pregnancy and of triplet or larger multiple
gestation (5%) until ~20 years ago (Blazar and Seifer, 1998;
Hecht and Magoon, 1998). This trend has accelerated, with
approximate doubling of the incidence of multiple pregnancy,
beginning with the widening popularity of IVF and related
technologies in the 1980s and continuing with the expansion
of indications for the use of gonadotrophins, particularly for
unexplained infertility in the late 1980s. Thus, with increasing
Multiple implantation after ovulation induction
Table I. Results of univariate analysis of the association of the variables investigated and type of pregnancy
Variable Low-order pregnancy High-order pregnancy P
(n 1771) (n 107)
Patient characteristics
Age (years) 32.8 0.09 31.7 0.35 0.005
Duration of infertility (years) 3.8 0.06 4.1 0.2 NS
Primary infertility (n, %) 1176 (66%) 74 (69%) NS
Secondary infertility (n, %) 595 (34%) 33 (31%) NS
Body mass index (kg/m
2
) 22.1 0.08 22.4 0.3 NS
Basal FSH (IU/l) 6.2 0.1 5.7 0.3 NS
Basal LH (IU/l) 6.3 0.15 6.4 0.84 NS
Treatment characteristics
Initial dose of FSH (75 IU ampoules) 1.47 0.01 1.8 0.08 0.001
Total dose of FSH (75 IU ampoules) 13.2 0.1 16.6 0.99 0.005
Days of ovarian stimulation 8.5 0.07 8.6 0.3 NS
Insemination procedure NS
Timed intercourse (n, %) 354 (20%) 31 (28%)
IUI (n, %) 1416 (80%) 77 (72%)
Motile sperm inseminated 10
6
8.9 1.4 7.8 1.7 NS
Luteal phase support 0.005
HCG (n, %) 1044 (59%) 44 (41%)
Progesterone (n, %) 549 (31%) 49 (46%)
None (n, %) 178 (10%) 14 (13%)
Ovarian response (HCG day)
No. of follicles 10 to 14 mm 1.4 0.03 1.8 0.14 0.01
No. of follicles 14 to 17 mm 1.6 0.03 2.6 0.17 0.001
No. of follicles 17 mm 2.0 0.03 2.5 0.15 0.005
Total no. of follicles 3.8 0.03 5.6 0.3 0.001
Oestradiol (pg/ml) 636 11 951 50 0.001
Values are mean SEM.
NS not signicant.
Table II. Observed numbers of cycles with low-order and high-order pregnancy and predicted probability of
high-order pregnancy according to multivariate ordinal logistic regression analysis*
No. follicles 10 mm on HCG day Peak oestradiol 862 pg/ml Peak oestradiol 862 pg/ml
Age 32 Age 32 Age 32 Age 32
13 follicles
low-order pregnancy (n) 319 266 22 35
high-order pregnancy (n) 10 8 4 7
probability 0.033 0.054 0.082 0.117
45 follicles
low-order pregnancy (n) 87 85 29 35
high-order pregnancy (n) 4 9 4 3
probability 0.043 0.066 0.084 0.130
5 follicles
low-order pregnancy (n) 66 92 67 132
high-order pregnancy (n) 2 5 7 29
probability 0.052 0.087 0.126 0.189
*Cycles shown (n 1327) are those for which complete data sets were available for analysis.
frequency since the advent of IVF, ovarian stimulants have
been administered to ovulatory women with the purpose of
recruiting multiple follicles and to ensure that several oocytes
are available within a given treatment cycle, the so-called
superovulation therapy. With superovulation therapy, a
broader range of aggressiveness is observed, reected in the
quantity of gonadotrophin used, the oestradiol concentrations
achieved and the number of follicles produced. This practice
substantially increases the number of women exposed to the
possibility of iatrogenic multiple pregnancy. However, because
of this changing philosophy tolerance to one degree or another
2127
of hyperstimulation has become commonplace (Blazar and
Seifer, 1998; Hecht and Magoon, 1998).
As recently shown in long-term follow-up studies (Yudin
et al., 2001) high-order multiple pregnancies have a large
adverse impact on perinatal morbidity and mortality, as well
as important economic consequences (The ESHRE Capri
Workshop Group, 2000). Therefore, strategies to prevent such
pregnancies are eminently desirable. It has been pointed out
that policies that would cancel cycles of ovulation induction
when there are more than two mature follicles would be a
technically feasible and readily available means of prevention.
R.Tur et al.
Unfortunately, however, the indiscriminate application of such
policies would reduce the pregnancy rate per cycle and lead
to even higher costs per live birth (Collins, 1994; Gleicher
et al., 2000). Therefore, the need for appropriate stimulation
protocols, careful cycle monitoring and strict criteria for the
cancellation of treatment to avoid multiple implantation have
been advocated (te Velde and Cohlen, 1999). This notwithstand-
ing, no ofcial or unofcial body has been able to offer any
regulations or guidelines for avoiding high-order multiple
gestations due to ovulation induction. Data to accomplish this
simply does not exist and it has been stressed that until there
is a proven protocol or information similar to that used for
developing the guidelines on the number of embryos to transfer,
we will be unable to deal effectively with high-order multiples
in relation to superovulation (Soules et al., 2001). At the very
least, further studies into the source of multiple pregnancies
in ovulation induction cycles would identify the target for an
educational effort (Jones and Schnorr, 2001).
As discussed above (see Introduction section) a number of
previous reports have tried to answer the fundamental question
as to whether multiple pregnancy can be predicted or prevented
by proper patient selection, exible treatment regimens and
careful, up-to-date monitoring techniques. However, results
from those studies were clearly contradictory. Furthermore,
recommendations resulting from those reports arose in most
cases from the analysis of a low number of patients, grouping
twins and triplets or more as a sole group of multiple pregnancy
and usually applying statistical methods only for comparison
between groups (single versus multiple pregnancy), but not
statistical techniques aimed at establishing the predictive value
of variables potentially associated with multiple implantation.
In contrast, we present here the largest series of pregnancies
obtained following ovarian stimulation with gonadotrophins
which also includes the highest number of high-order multiple
pregnancies ever reported. We have, on the other hand, used
a statistical method using ROC curve analysis to test the
usefulness of a womans age and peak serum oestradiol
concentrations to discriminate between low-order and high-
order pregnancies. In ROC curve analysis many efciencies
of all decision levels can be calculated, resulting in an
overall quantication of accuracy which is not affected by the
prevalence of a condition (Hanley and McNeil, 1982; Zweig
and Campbell, 1993). Thus, our analysis shows clearly that the
possibility of high-order multiple gestation in gonadotrophin
treatment cycles is dependent on age, serum oestradiol concen-
trations and the number of growing follicles on the day of
HCG injection. This is in agreement with a recent study where
314 singleton, 88 twin and 39 triplet pregnancies were
included and similar statistical methods to ours were used to
identify values that predicted multiple conceptions (Gleicher
et al., 2000).
In summary, the three-variable model presented here and
based on a large series of 1878 intrauterine pregnancies
can identify patients at high-risk for high-order multiple
implantation. However, several facts should be considered
before the results of the present study are applied in clinical
practice and can help to improve the efcacy and safety of
ovulation induction. First, like all previous studies on the
2128
subject, this is a retrospective analysis of data from a hetero-
geneous group of patients with dissimilar diagnostic entities,
duration of infertility, prior therapy and age who, in addition,
were managed according to different criteria for ovarian
stimulation and monitoring because a number of changes in
both the clinic policy (e.g. conventional versus chronic low-
dose regimens of gonadotrophin administration) and available
tools (e.g. ultrasonography technology) were introduced after
1988. Second, our data are hospital-based and therefore open
to selection bias. Third, the technique for measuring follicles
is not fully standardized and lack of standardization in reporting
either mean or maximal diameters and intra- and inter-observer
variability further confound ultrasonographic follicular assess-
ment (Navot et al., 1991a). Fourth, signicantly different
values of oestradiol may be obtained depending on the
analytical method used, the reagent manufacturer and, most
importantly, even when the same kits are used at different
laboratories (Hershlag et al., 1992).
Therefore, recognizing the importance of certain variables
in predicting multiple pregnancies in cycles stimulated with
gonadotrophins, appropriate guidelines should be established
in each individual centre. Prospective, well-designed studies
including homogenous groups of patients are warranted to
determine whether withholding HCG on the basis of womans
age, oestradiol concentrations and follicle quantity may be
effective in reducing high-order multiple pregnancies without
negatively affecting overall pregnancy rates in patients who
require gonadotrophin ovarian stimulation to become pregnant.
The potential role of specic parameters of sperm function
(e.g. amplitude of lateral head movement of spermatozoa) in
multiple gestation, as recently suggested (Pasqualotto et al.,
1999), also deserves further study.
Acknowledgement
This work was performed under the auspices of Ca`tedra
dInvestigacio en Obstetr cia i Gineclologia of the Department of
Obstetrics and Gynecology, Institut Universitari Dexeus, Universitat
Auto`noma de Barcelona.
References
Balasch, J., Tur, R. and Peinado, J.A. (1996) The safety and effectiveness of
stepwise and low-dose administration of follicle stimulating hormone in
WHO group II anovulatory infertile women: evidence from a large
multicenter study in Spain. J. Assist. Reprod. Genet., 13, 551556.
Ben-Nun, I., Cohen, I., Shulman, A. et al. (1993) The inability of preovulatory
ovarian scan to predict multifetal pregnancy occurrence in a follow-up of
induction of ovulation with menotropins. Fertil. Steril., 60, 781785.
Berkowitz, R.L. (1996) From twin to singleton. Br. Med. J., 313, 373374.
Blazar, A.S. and Seifer, D.B. (1998) Reproductive technology and multiple
births. Infertil. Reprod. Med. Clin. N. Am., 9, 371377.
CDC (2000) Contribution of assisted reproduction technology and ovulation-
inducing drugs to triplet and higher-order multiple birthsUnited States,
19801997. J.A.M.A., 284, 299300.
Collins, J.A. (1994) Reproductive technologyThe price of progress. N. Engl.
J. Med., 331, 270271.
Corchia, C., Mastroiacovo, P., Lanni, R. et al. (1996) What proportion of
multiple births are due to ovulation induction? A register-based study in
Italy. Am. J. Pub. Health, 86, 851854.
Derom, C., Derom, R., Vlietinck, R. et al. (1993) Iatrogenic multiple
pregnancies in East Flanders, Belgium. Fertil. Steril., 60, 493496.
Dickey, R.P., Olar, T.T., Taylor, S.N. et al. (1991) Relationship of follicle
number, serum oestradiol, and other factors to birth rate and multiparity in
Multiple implantation after ovulation induction
human menopausal gonadotrophin-induced intrauterine insemination cycles.
Fertil. Steril., 56, 8992.
Dickey, R.P., Taylor, S.N., Lu, P.Y. et al. (2001) Relationship of follicle
numbers and estradiol levels to multiple implantation in 3608 intrauterine
insemination cycles. Fertil. Steril., 75, 6978.
Dodson, W.C., Hughes, C.L. and Haney, A.F. (1988) Multiple pregnancies
conceived with intrauterine insemination during superovulation: an
evaluation of clinical characteristics and monitored parameters of conception
cycles. Am. J. Obstet. Gynecol., 159, 382385.
Evans, M.I., Littman, L., St. Louis, L. et al. (1995) Evolving patterns of
iatrogenic multifetal pregnancy generation: implications for aggressiveness
of infertility treatments. Am. J. Obstet. Gynecol., 172, 17501755.
Evans, M.I., Kramer, R.L., Yaron, Y. et al. (1998) What are the ethical and
technical problems associated with multifetal pregnancy reduction? Clin.
Obstet. Gynecol., 41, 4754.
Farhi, J., West, C., Patel, A. et al. (1996) Treatment of anovulatory infertility:
the problem of multiple pregnancy. Hum. Reprod., 11, 429434.
Gleicher, N., Campbell, D.P., Chan, C.L. et al. (1995) The desire for multiple
births in couples with infertility problems contradicts present practice
patterns. Hum. Reprod., 10, 10791084.
Gleicher, N., Oleske, D.M., Tur-Kaspa, I. et al. (2000) Reducing the risk of
high-order multiple pregnancy after ovarian stimulation with gonadotropins.
N. Engl. J. Med., 343, 27.
Goldenberg, M., Rabinovici, J., Shalev, J. et al. (1994) Lack of association
between ovarian follicular size and number and the occurrence of multiple
pregnancies in menotropin cycles. Gynecol. Endocrinol., 8, 8387.
Goldfarb, J.M., Peskin, B., Austin, C. et al. (1997) Evaluation of predictive
factors for multiple pregnancies during gonadotropin/IUI treatment. J.
Assist. Reprod. Genet., 14, 8891.
Hanley, J.A. and McNeil, B.J. (1982) The meaning and use of the area under
receiving operating characteristic (ROC) curve. Radiology, 143, 2936.
Hecht, B.R.and Magoon, M.W. (1998) Can the epidemic of iatrogenic multiples
be conquered? Clin. Obstet. Gynecol., 41, 127137.
Hershlag, A., Lesser, M., Montefusco, D. et al. (1992) Interinstitutional
variability of follicle-stimulating hormone and estradiol levels. Fertil. Steril.,
58, 11231126.
Hosmer, D.W. and Lemeshow, S. (1989) Applied Logistic Regression. Wiley,
New York.
Jones, H.W. and Schnorr, J.A. (2001) Multiple pregnancies: a call for action.
Fertil. Steril., 75. 1117.
Kurachi, K., Aono, T., Suzuki, M. et al. (1985) Results of HMG (Humegon)-
HCG therapy in 6096 treatment cycles of 2166 Japanese women with
anovulatory infertility. Europ. J. Obstet. Gynecol. Reprod. Biol., 19,
4351.
Levene, M.I., Wild, J. and Steer, P. (1992) Higher multiple births and the
2129
modern management of infertility in Britain. Br. J. Obstet. Gynaecol., 99,
607613.
Navot, D., Bergh, P.A., Drews, M. et al. (1991a) The role of ultrasound in
ovulation induction. Infertil. Reprod. Med. Clin. N. Am., 2, 741752.
Navot, D., Goldstein, N., Mor-Josef, S. et al. (1991b) Multiple pregnancies:
risk factors and prognostic variables during induction of ovulation with
human menopausal gonadotrophins. Hum. Reprod., 6, 11521155.
Pardo, M., Barri, P.N., Bancells, J. et al. (1988) Spermatozoa selection in
discontinuous Percoll gradients for use in artical insemination. Fertil.
Steril., 49, 505509.
Pasqualotto, E.B., Falcone, T., Goldberg, J.M. et al. (1999) Risk factors for
multiple gestation in women undergoing intrauterine insemination with
ovarian stimulation. Fertil. Steril., 72, 613618.
Sassoon, D.A., Castro, L.C., Davis, J.L. et al. (1990) Perinatal outcome in
triplet versus twin gestations. Obstet. Gynecol., 75, 817820.
Seoud, M.A.F., Toner, J.P., Kruithoff, C. et al. (1992) Outcome of twin, triplet,
and quadruplet in-vitro fertilization pregnancies: the Norfolk experience.
Fertil. Steril., 57, 825834.
Shelden, R., Kemmann, E., Bohrer, M. et al. (1988) Multiple gestation is
associated with the use of high sperm numbers in the intrauterine
insemination specimen in women undergoing gonadotropin stimulation.
Fertil. Steril., 49, 607610.
Soules, M.R., Chang, R.J., Lipshultz, L.I. et al. (2001) Multiple pregnancies :
action is taking place. Fertil. Steril., 75, 1516.
te Velde, E.R. and Cohlen, B.J. (1999) The management of infertility. N.
Engl. J. Med., 340, 224226.
The ESHRE Capri Workshop Group (2000) Multiple gestation pregnancy.
Hum. Reprod., 15, 18561864.
Valbuena, D., Simon, C., Romero, J.L. et al. (1996) Factors responsible for
multiple pregnancies after ovarian stimulation and intrauterine insemination
with gonadotropins. J. Assist. Reprod. Genet., 13, 663668.
Wang, C.F. and Gemzell, C. (1980) The use of human gonadotropins for the
induction of ovulation in women with polycystic ovarian disease. Fertil.
Steril., 33, 479486.
Wilcox, L.S., Kiely, J.L., Melvin, C.L. and Martin, M.C. (1996) Assisted
reproductive technologies: estimates of their contribution to multiple births
and newborn hospital days in the United States. Fertil. Steril., 65, 361366.
Yu, S.L., Pepperell, R.J. and Evans, J.H. (1991) Can ultrasonography reliably
predict the occurrence of multiple pregnancies in gonadotrophin ovulation
induction? Aust. NZ. J. Obstet. Gynaecol., 31, 5862.
Yudin, M.H., Asztalos, E.V., Jefferies, A. et al. (2001) The management and
outcome of high order multifetal pregnancies: obstetric, neonatal and follow-
up data. Twin Research, 4, 411.
Zweig, M.H. and Campbell, G. (1993) Receiver-operating characteristic (ROC)
plots: a fundamental evaluation tool in clinical medicine. Clin. Chem., 39,
561577.
Received on March 26, 2001; accepted on July 13, 2001