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 Tur1,3, Pedro N. Barri1, Buenaventura Coroleu1, Rosario Buxaderas1, Francisca Martínez1 and Juan Balasch2

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 Clínic-Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Barcelona, Spain


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
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 specific 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. Stratification of the number of follicles into three categories (1 to 3, 4 to 5, and >5 follicles respectively), peak serum oestradiol and woman's age according to the ROC curves, showed that the risk of high-order multiple implantation correlated significantly with increasing total number of follicles and was significantly 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
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
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, 2000Go; The ESHRE Capri Workshop Group, 2000Go; Jones and Schnorr, 2001Go). Interestingly, according to population-based studies, as much as two thirds of iatrogenic multiple pregnancies, mainly high-order gestations (>=triplets), may be attributable to ovulation-inducing drugs without IVF or related techniques (Levene et al., 1992Go; Derom et al., 1993Go; Evans et al., 1995Go; Corchia et al., 1996Go; Wilcox et al., 1996Go), a figure 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, 2000Go; Jones and Schnorr, 2001Go). In fact, there has been greater control of IVF than ovulation stimulation (Evans et al., 1995Go; Jones and Schnorr, 2001Go). Therefore, identification 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 development, i.e. ovarian ultrasonography and determination of serum oestradiol concentration. However, data regarding the predictive 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., 1996Go; Valbuena et al., 1996Go; Pasqualotto et al., 1999Go; Gleicher et al., 2000Go; Dickey et al., 2001Go) but others disagree (Kurachi et al., 1985Go; Shelden et al., 1988Go; Dodson et al., 1988Go; Navot et al., 1991bGo; Yu et al., 1991Go; Ben-Nun et al., 1993Go; Goldenberg et al., 1994Go; Goldfarb et al., 1997Go). Similarly, the usefulness of ovarian ultrasonography in identifying cycles with high-risk for multiple pregnancy is stressed in some reports (Dickey et al., 1991Go, 2001Go; Navot et al., 1991bGo; Farhi et al., 1996Go; Valbuena et al., 1996Go; Gleicher et al., 2000Go) but not all reports (Kurachi et al., 1985Go; Dodson et al., 1988Go; Shelden et al., 1988Go; Yu et al., 1991Go; Ben-Nun et al., 1993Go; Godenberg et al., 1994; Goldfarb et al., 1997Go; Pasqualotto et al., 1999Go).

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., 1990Go; Seoud et al., 1992Go; Gleicher et al., 1995Go; Berkowitz, 1996Go; Evans et al., 1998Go). However, women with infertility first and foremost demand high rates of pregnancy, which may entail a relatively high risk of low-order multiple pregnancy (Gleicher et al., 1995Go). Finally, it has been stressed that an attempt to eliminate most twin pregnancies after ovulation induction would imply a significant decline in the overall pregnancy rate (Goldenberg et al., 1994Go).

Therefore, on the above evidence, the present investigation was undertaken to determine whether there were specific 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
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Patients
Between January 1988 and December 1998, 1878 intrauterine pregnancies 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. Women were either anovulatory or were undergoing ovarian stimulation on an empirical basis, usually in conjunction with intrauterine insemination (IUI). Thus, of the 1878 conceptional cycles, 1493 resulted from IUI with husband's (n = 1041) or donor frozen–thawed (n = 452) spermatozoa and the remaining 385 followed ovulation induction and timed intercourse.

Treatment protocols
All pregnancy cycles were treated with gonadotrophins. The preparation used was human menopausal gonadotrophin (HMG) (Pergonal; Serono S.A., Madrid, Spain) in 1201 (64%) cycles and highly purified 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 3–5 days until an ovarian response occurs (Wang and Gemzell, 1980Go), 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., 1996Go). 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 36–40 h after HCG administration using husband or donor sperm samples processed with Percoll gradients as reported previously (Pardo et al., 1988Go).

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 transvaginal ultrasonography at 6 weeks gestation. The order of the multiple pregnancy was classified 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 specific 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 coefficients 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 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 3–5 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 significant associated with high-order multiple pregnancy were retained for testing in a multivariate logistic regression model (Hosmer and Lemeshow, 1989Go). Receiver operating characteristic (ROC) curves (Hanley and McNeil, 1982Go; Zweig and Campbell, 1993Go) were used to determine the cut-off point which best discriminated between high-order multiple pregnancy (>=3 gestational sacs) and low-order gestation (1–2 gestational sacs) for those variables remaining statistically significant 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 (AUCROC) 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
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
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 IGo. Of them, nine were independently associated with high-order pregnancy. When these nine variables were entered simultaneously into the logistic regression model, only three of them remained statistically significant: the age of the woman; the peak serum oestradiol; and the total number of growing follicles (>10 mm) on the day of HCG administration. Their ß-coefficients 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 AUCROC curves determined with ROC analysis. AUCROC curves (95% CI) for age, number of growing follicles, and peak serum oestradiol were 0.60 (0.57–0.63), 0.65 (0.63–0.67), and 0.70 (0.67–0.72) 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.


View this table:
[in this window]
[in a new window]
 
Table I. Results of univariate analysis of the association of the variables investigated and type of pregnancy
 
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 IIGo. Thus, for example, when the total number of follicles is 4–5 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 concentrations >862 pg/ml, the risk increases 3-fold to as much as 13% (Table IIGo). Overall, Table IIGo clearly shows that the risk increases with increasing serum oestradiol concentrations, with younger age of the woman and increasing total number of growing follicles.


View this table:
[in this window]
[in a new window]
 
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*
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
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 modification in the original philosophy of their use. Initially, these agents were used exclusively for the treatment of anovulation, oligoovulation 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, 1998Go; Hecht and Magoon, 1998Go). 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 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, reflected 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 of hyperstimulation has become commonplace (Blazar and Seifer, 1998Go; Hecht and Magoon, 1998Go).

As recently shown in long-term follow-up studies (Yudin et al., 2001Go) 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, 2000Go). 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. 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, 1994Go; Gleicher et al., 2000Go). 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, 1999Go). This notwithstanding, no official or unofficial 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., 2001Go). 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, 2001Go).

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, flexible 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 woman's age and peak serum oestradiol concentrations to discriminate between low-order and high-order pregnancies. In ROC curve analysis many efficiencies of all decision levels can be calculated, resulting in an overall quantification of accuracy which is not affected by the prevalence of a condition (Hanley and McNeil, 1982Go; Zweig and Campbell, 1993Go). Thus, our analysis shows clearly that the possibility of high-order multiple gestation in gonadotrophin treatment cycles is dependent on age, serum oestradiol concentrations 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., 2000Go).

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 efficacy and safety of ovulation induction. First, like all previous studies on the subject, this is a retrospective analysis of data from a heterogeneous 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 assessment (Navot et al., 1991aGo). Fourth, significantly 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., 1992Go).

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 woman's 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 specific parameters of sperm function (e.g. amplitude of lateral head movement of spermatozoa) in multiple gestation, as recently suggested (Pasqualotto et al., 1999Go), also deserves further study.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
This work was performed under the auspices of `Càtedra d'Investigació en Obstetrícia i Gineclologia' of the Department of Obstetrics and Gynecology, Institut Universitari Dexeus, Universitat Autònoma de Barcelona.


    Notes
 
3 To whom correspondence should be addressed at: Service of Reproductive Medicine, Institut Universitari Dexeus. Pso Bonanova 67, 08017-Barcelona, Spain. E-mail: rostur{at}iudexeus.uab.es Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 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, 551–556.[ISI][Medline]

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, 781–785.[ISI][Medline]

Berkowitz, R.L. (1996) From twin to singleton. Br. Med. J., 313, 373–374.[Free Full Text]

Blazar, A.S. and Seifer, D.B. (1998) Reproductive technology and multiple births. Infertil. Reprod. Med. Clin. N. Am., 9, 371–377.

CDC (2000) Contribution of assisted reproduction technology and ovulation-inducing drugs to triplet and higher-order multiple births—United States, 1980–1997. J.A.M.A., 284, 299–300.[Free Full Text]

Collins, J.A. (1994) Reproductive technology—The price of progress. N. Engl. J. Med., 331, 270–271.[Free Full Text]

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, 851–854.[Abstract]

Derom, C., Derom, R., Vlietinck, R. et al. (1993) Iatrogenic multiple pregnancies in East Flanders, Belgium. Fertil. Steril., 60, 493–496.[ISI][Medline]

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 human menopausal gonadotrophin-induced intrauterine insemination cycles. Fertil. Steril., 56, 89–92.[ISI][Medline]

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, 69–78.[ISI][Medline]

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, 382–385.[ISI][Medline]

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, 1750–1755.[ISI][Medline]

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, 47–54.[ISI]

Farhi, J., West, C., Patel, A. et al. (1996) Treatment of anovulatory infertility: the problem of multiple pregnancy. Hum. Reprod., 11, 429–434.[Abstract]

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, 1079–1084.[Abstract]

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, 2–7.[Abstract/Free Full Text]

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, 83–87.[ISI][Medline]

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, 88–91.[ISI][Medline]

Hanley, J.A. and McNeil, B.J. (1982) The meaning and use of the area under receiving operating characteristic (ROC) curve. Radiology, 143, 29–36.[Abstract]

Hecht, B.R.and Magoon, M.W. (1998) Can the epidemic of iatrogenic multiples be conquered? Clin. Obstet. Gynecol., 41, 127–137.[ISI]

Hershlag, A., Lesser, M., Montefusco, D. et al. (1992) Interinstitutional variability of follicle-stimulating hormone and estradiol levels. Fertil. Steril., 58, 1123–1126.[ISI][Medline]

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. 11–17.[ISI][Medline]

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, 43–51.[ISI][Medline]

Levene, M.I., Wild, J. and Steer, P. (1992) Higher multiple births and the modern management of infertility in Britain. Br. J. Obstet. Gynaecol., 99, 607–613.[ISI][Medline]

Navot, D., Bergh, P.A., Drews, M. et al. (1991a) The role of ultrasound in ovulation induction. Infertil. Reprod. Med. Clin. N. Am., 2, 741–752.

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, 1152–1155.[Abstract]

Pardo, M., Barri, P.N., Bancells, J. et al. (1988) Spermatozoa selection in discontinuous Percoll gradients for use in artifical insemination. Fertil. Steril., 49, 505–509.[ISI][Medline]

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, 613–618.[ISI][Medline]

Sassoon, D.A., Castro, L.C., Davis, J.L. et al. (1990) Perinatal outcome in triplet versus twin gestations. Obstet. Gynecol., 75, 817–820.[Abstract]

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, 825–834.[ISI][Medline]

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, 607–610.[ISI][Medline]

Soules, M.R., Chang, R.J., Lipshultz, L.I. et al. (2001) Multiple pregnancies : action is taking place. Fertil. Steril., 75, 15–16.[ISI][Medline]

te Velde, E.R. and Cohlen, B.J. (1999) The management of infertility. N. Engl. J. Med., 340, 224–226.[Free Full Text]

The ESHRE Capri Workshop Group (2000) Multiple gestation pregnancy. Hum. Reprod., 15, 1856–1864.[Abstract/Free Full Text]

Valbuena, D., Simón, 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, 663–668.[ISI][Medline]

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, 479–486.[ISI][Medline]

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, 361–366.[ISI][Medline]

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, 58–62.[ISI][Medline]

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, 4–11.[Medline]

Zweig, M.H. and Campbell, G. (1993) Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin. Chem., 39, 561–577.[Abstract/Free Full Text]

Submitted on March 26, 2001; accepted on July 13, 2001.