IN VITRO FERTILIZATION

Transcription

1 IN VITRO FERTILIZATION FERTILITY AND STERILITY VOL. 81, NO. 6, JUNE 2004 Copyright 2004 American Society for Reproductive Medicine Published by Elsevier Inc. Printed on acid-free paper in U.S.A. Use of outcomes-based data in reducing high-order multiple pregnancies: the role of age, diagnosis, and embryo score C. Matthew Peterson, M.D., a James C. Reading, Ph.D., a Harry H. Hatasaka, M.D., a Kirtly Parker Jones, M.D., a Laurence C. Udoff, M.D., a Eli Y. Adashi, M.D., a Paul H. Kuneck, M.D., b Lisa D. Erickson, M.D., b John W. Malo, M.D., b Bruce F. Campbell, M.D., b and Douglas T. Carrell, Ph.D. a,b Received March 20, 2003; revised and accepted January 7, Portions of this manuscript were presented at the First Global Conference on Reducing the Incidence of High Order Multiple Pregnancies, sponsored by The National Institutes of Health and the Bertarelli Foundation, November 22, 1999, Bethesda, Maryland; and at the American Society for Reproductive Medicine 58th Annual Meeting, October 12 17, 2002, Seattle, Washington. Reprint requests: C. Matthew Peterson, M.D., Professor and Director, Division of Reproductive Endocrinology and Infertility, University of Utah Health Sciences Center and Intermountain Health Care, 30 North 1900 East, Suite 2-B 200 SOM, Salt Lake City, Utah (FAX: ; c.matthew. peterson@hsc.utah.edu). a Utah Center for Reproductive Medicine, Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, University of Utah Health Sciences Center. b Center for Reproductive Medicine, Abbott Northwestern Hospital /04/$30.00 doi: /j.fertnstert Utah Center for Reproductive Medicine, Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, University of Utah Health Sciences Center, Salt Lake City, Utah; and Center for Reproductive Medicine, Abbott Northwestern Hospital, Minneapolis, Minnesota Objective: To identify high-risk categories for high-order multiple pregnancy (HMP) in in vitro fertilization (IVF), establish clinic-specific HMP risk data for counseling use, and verify their utility in reducing HMP. Design: Before and after intervention study. Setting: Two IVF programs using the same embryology laboratory and IVF protocols. Patient(s): All IVF patients undergoing fresh embryo transfers. Intervention(s): Use of clinic-specific age, diagnosis, and embryo score (ES) risk data in assessing individual HMP risk during informed consent. Main Outcome Measure(s): HMP and pregnancy outcomes. Result(s): In determining clinic-specific high risk categories and developing outcomes-based HMP risk data for counseling, the good outcome rate (GR) was defined as the percentage of singleton or twin deliveries per cycle and the bad outcome rate included no pregnancy or nondelivered pregnancies (miscarriages, multifetal reduction) and HMP per cycle. During 1995 to 1999, age 35 years, calculated morphologic ES, and donor egg (DE) cycles were factors shown by logistic regression to statistically significantly affect the GR. The optimal GRs for DE 35 and 35 years (donor age), and non-de cycles 35 years were achieved with two (57.7%), three (43.2%), and three (43.2%) embryos transferred, respectively. A DE 35 years with 3 embryos transferred had the highest risk for HMP. The GR correlated (0.91) with the ES according to the formula: GR ES, when ES range was between 4 and 26. Clinic-specific risks for HMP based on age, diagnosis, and ES were developed and considered while counseling for ET during The clinicspecific HMP risk data made for a reduction in the HMP rate of 90.9% for DE-IVF (11.8% to 1%) and 53.8% for all IVF (9.1% to 4.2%), without decreases in clinical pregnancy or delivery rates. Physicians showing the greatest decline (64%) in HMP had no reduction in pregnancy or delivery rates. Conclusion(s): The use of clinic-specific HMP risk data in counseling based on age, diagnosis, and ES provided a 53% to 64% reduction in HMP without affecting rates of pregnancy or delivery. The clinic-specific ES system correlated closely with good outcomes. A standardized ES system may provide useful information for counseling during ET informed consent. (Fertil Steril 2004;81: by American Society for Reproductive Medicine.) Key Words: High-order multiple pregnancy, IVF, triplets, quadruplets High-order multiple pregnancies ( 3) now constitute 0.1% to 0.3% of all pregnancies, nearly a 100-fold increase since the first IVF procedure in 1978 (1). Present estimates attribute approximately 50% of twin pregnancies to assisted reproductive technologies (ART, in the form of ovulation induction, superovulation, IVF, gamete intrafallopian transfer, or zygote intrafallopian transfer). Approximately 90% of triplet and 93% of quadruplet pregnancies are considered to be the result of ART (2, 3). The economic impact of high-order multiple pregnancies (HMP) is significant. Charges associated with a singleton birth in 1994 were estimated at U.S.$9,845 as compared with $37,947 and $109,765 for twins and triplets, 1534

2 FIGURE 1 Embryo scoring system. Embryos were assigned a level (1 to 3) based on the uniformity of size of the blastomeres and the presence of cellular fragmentation. (A) Level 1 embryos contained no fragmentation and uniform blastomeres. (B) Level 2 embryos had unequal blastomeres and/or fragmentation of 30% of the embryo volume. (C) Level 3 embryos had unequal blastomeres and/or fragmentation of 30% of the embryo volume. Each individual embryo score was determined by subtracting the embryo level (1 to 3) from the total number of blastomeres on day 3 after retrieval. The final embryo score (ES) was then calculated by adding together the individual scores for each embryo transferred. Further information on embryo scoring can be found in Carrell et al. (8). respectively (4). In many countries with socialized health care systems, legislation limits the number of embryos that can be transferred during an IVF cycle in an effort to eliminate HMP (5). Recent studies suggest that limiting the number of embryos transferred in specific high-risk categories may reduce the incidence of HMP. However, such efforts may also result in a concurrent reduction in pregnancy rate, particularly in women over 35 years of age (6). The American Society for Reproductive Medicine (ASRM) has suggested that individual programs determine clinic-specific data to guide them in determining the number of embryos to be transferred during IVF (7). The purpose of our study was to determine whether the presentation of HMP-specific data regarding age, diagnosis, and embryo quality while counseling patients regarding the number of embryos for transfer affected our clinics rates of HMP, pregnancy, and/or delivery. MATERIALS AND METHODS The respective institutional review boards of the participating institutions approved this study. We retrospectively analyzed all of the fresh non donor egg (non-de) and donor egg (DE) IVF cycles (n 1,757 retrievals, 1,740 embryo transfers, 777 clinical pregnancies) of the Utah Center for Reproductive Medicine (330 of the 777 clinical pregnancies) and Abbott Northwestern Hospital s Center for Reproductive Medicine (447 of the 777 clinical pregnancies) from January 1995 through December The two clinics programs are related through common IVF laboratory procedures, embryo scoring systems, personnel, and IVF protocols. Both programs adhere to all standards and embryo transfer (ET) guidelines established by ASRM and the Society for Assisted Reproductive Technology (SART). In this study, an IVF cycle refers to a case in which oocytes (eggs) were removed from a woman s ovaries, were combined with sperm (microdroplet insemination or intracytoplasmic sperm injection), and resulting embryos were transferred into a uterus. A clinical pregnancy was defined as one or more gestational sacs with a fetal pole possessing cardiac activity documented by ultrasound; this specifically excluded chemical and ectopic pregnancies. A multiple pregnancy was defined as two or more fetal poles with cardiac activity on early ultrasound and/or the delivery of two or more infants. The outcome of each clinical pregnancy was recorded as a live birth, still birth, spontaneous abortion, therapeutic abortion, or multifetal reduction. Each live birth delivery was counted as a single live birth. The good outcome rate (GR) was defined as singleton or twin deliveries per cycle. The bad outcome rate included no pregnancy or delivery (including all miscarriages, stillbirths, abortions, and multifetal reductions) or a HMP per cycle. Embryo scores (ES) based on morphologic features at the time of ET were calculated as described in Figure 1 (8). Table 1 details demographic and outcome data for the period 1995 to Logistic regression was performed to determine the variables related to good outcomes. The IVF variables analyzed by logistic regression included age, diagno- FERTILITY & STERILITY 1535

3 TABLE 1 Demographics and outcomes from IVF cases Category Pregnant cohort (n 777) Pregnancy rates by diagnosis (n 761) Nonpregnant cohort (n 963) Age (y) (SD) (SD) Average number of embryos transferred Advanced maternal age 1.7% (13/761) 31% (13/42) 3% (29/953) Donor egg IVF 20.1% (153/761) 58.4% (153/262) 11.4% (109/953) Endometriosis 15.6% (119/761) 47% (119/253) 14.2% (135/953) Male factor infertility 22.2% (169/761) 37.1% (169/455) 30% (286/953) Ovulatory dysfunction 5.3% (40/761) 50% (40/80) 4.2% (40/953) Tubal factor infertility 24.4% (186/761) 45.8% (186/406) 23.1% (220/953) Unexplained infertility 10.5% (80/761) 38.1% (80/210) 13.6% (130/953) Uterine factor infertility 0.2% (1/761) 16.7% (1/6) 0.5% (5/953) Note: There were no statistically significant differences between the pregnant and nonpregnant cohorts, except outcomes. Diagnoses were recorded for 761 of 777 (98%) of the pregnant and 953 of 963 (99%) of the nonpregnant cohorts. The diagnostic categories for IVF patients were: advanced maternal age (female 40 years of age); donor egg IVF (requiring the use of donated eggs); endometriosis (laparoscopically documented disease); male factor (requiring intracytoplasmic sperm injection on some or all of embryos transferred); ovulatory dysfunction (absence of spontaneous menses but demonstrating bleeding after a progestin challenge); tubal factor (one or both fallopian tubes occluded); unexplained infertility (no known infertility factors); and uterine factor (DES exposure or significant uterine anomaly [iatrogenic or congenital], excluding excised uterine septi). The age of the egg donor was used for the age in the analysis of donor egg IVF cycles. TABLE 2 Logistic regression of factors affecting the good/bad outcome of IVF. Variable P value Age 35 y a.0001 Donor egg a.0262 Embryo score a.0001 % Fertilization.7914 Institution (state).6698 Endometriosis/ovulatory dysfunction.1382 Male factor infertility.4543 Tubal factor infertility.2145 Unexplained infertility.6443 Note: To determine the factors that influenced the good/bad outcome, logistic regression was performed using the following variables: age 35 or 35 years old; embryo score; fertilization rates; institution (state); and the diagnostic categories tubal factor, male factor, donor egg IVF, unexplained infertility, and the combination of endometriosis and ovulatory dysfunction. This approach yielded 1,679 evaluable cases. a Age, donor egg, and embryo score were statistically significant: P.0001,.0001, and.0262, respectively. sis, embryo transfer scores, fertilization rates, institution, number of embryos transferred, and outcome (Table 2). Logistic regression, including all variables in the model, was performed initially, followed by forward and backward, stepwise logistic regression for confirmation. Although cycle number may affect the number of embryos for transfer, to achieve statistical significance in developing clinic-specific HMP risk data for counseling, all cycles were weighted equally. Because of this analysis strategy, the guidelines developed are applicable to the first four cycles in our programs. In addition to descriptive statistics and logistic regression, tests were performed to test the differences in proportions. Linear regression was employed to determine the linear relationship of the good outcome rate per cycle (GR) to the ES. The planned intervention in our population was to have clinic-specific HMP risk data (Table 3 data specific to age, diagnosis, and embryo score; Fig. 2 for examples) available for the patient counseling and informed consent processes before embryo transfer from January 2000 through September The discussion with patients at the time of ET included the ET guidelines, both ASRM/SART and later (January 2000 to September 2001 intervention), clinic-specific HMP risk data derived from 1995 to 1999 data (e.g., see Table 3, Fig. 2), and the consequences of HMP. Informed consent regarding the number of embryos to transfer was completed before ET by the physician directing the cycle. The follow-up evaluation included an analysis of all IVF cycles during the intervention period (1,288 embryo transfers with 597 clinical pregnancies). The IVF procedure was performed using a standard protocol. Briefly, on day 21 of the cycle before gonadotropin stimulation, mg/day s.c. leuprolide acetate (LA) was given in the evening. At the onset of menses, the dose of LA was reduced to mg/day. After a screening ultrasound evaluation was performed, IU of gonadotropin was given i.m. for 3 days. Thereafter, gonadotropin 1536 Peterson et al. Reducing high-order multiple pregnancies Vol. 81, No. 6, June 2004

4 FIGURE 2 Outcomes in patients undergoing IVF embryo transfer with three embryos for the diagnostic category male factor infertility 35 years old (n 215). The embryo score for the three embryos (shown on the x-axis) in five-point increments (0 5, 6 10, 11 15, 16 20, and 21 30) is compared with the percentage of singleton (blue), twin (yellow), and triplet (red) live births per embryo transfer on the y-axis. The columns represent the total cumulative live-birth rate for each embryo score group in this diagnostic category (n for embryo score 0 5 6, , , , and ). administration was maintained at the original dose or reduced to IU until adequate follicular development and estradiol concentrations were achieved. A minimum of three follicles 1.7 cm and an estradiol level of 500 pg/ml (conversion factor to IS units, 3.671) was required before administering i.m. human chorionic gonadotropin (hcg) (10,000 IU). Thirty-six hours after the hcg injection, eggs were retrieved transvaginally, concurrent with 3 to 4 days of prophylactic doxycycline (100 mg p.o., b.i.d.) or tetracycline (250 mg p.o., every 6 hours). Patients received 50 mg/day i.m. of progesterone on the day of retrieval; progesterone supplementation was continued for 8 weeks if a clinical pregnancy occurred. Resulting embryos were evaluated regularly; at 72 hours after egg retrieval, two to five cleavage-stage embryos were transferred into the uterus transcervically. The DE IVF patients were also placed on LA (unless they were determined to be postmenopausal by FSH level 40 IU/L), and micronized estradiol was administered in a stepwise fashion from 2 to 6 mg/day until the endometrial lining achieved 8 mm in thickness. Progesterone was initiated at mg/day i.m. beginning the day after donor hcg injection. The age of the egg donor was used for the age in the analysis of DE IVF cycles. A pregnancy test was performed 14 days after ET. Ultrasound examination to document a clinical pregnancy was performed at 6 weeks gestational age. All patients were contacted for pregnancy outcome information. RESULTS From 1995 to 1999, 777 clinical pregnancies resulted from 1,740 ETs (777 out of 1,740, 44.7%), all with complete pregnancy follow-up. The demographics and pregnancy rates by diagnosis are detailed in Table 1. The multiple pregnancy rate of the pregnant cohort was 38% (295 out of 777): twins 30.2% (235 out of 777), triplets 7.2% (56 out of 777), and quadruplets 0.6% (4 out of 777). This gave a HMP rate of 7.7% (60 out of 777) or 20.3% (60 out of 295) of all multiple pregnancies. However, there were 11 multifetal reduction procedures for maternal/fetal indications: four quadruplets resulting in three sets of twins and one singleton; and seven triplets resulting in five sets of twins, and two with no delivery. This resulted in a total pregnancy loss rate attributed to multifetal reduction of 18% (2 out of 11 procedures). When HMP treated by multifetal reduction was taken into account, the corrected HMP rate for all clinical pregnancies was triplets 8.1% (63 out of 777) and quadruplets 1% (8 out of 777). To determine the factors that influenced the good/bad outcome, logistic regression was performed on multiple variables (see Table 2). Age 35 years old (P.0001), ES (P.0001), and DE cycles (P.0262) were found to be statistically significant. Thereafter, forward and backward, stepwise logistic regression was performed on the variables, confirming the statistical significance of age 35 years (P.0001), ES (P.0001), and DE cycles (P.0008). The ES correlated with the GR in a linear fashion, defined by the equation GR ES within the range 4 ES 26, with a correlation coefficient of 0.91 (Fig. 3). To determine the influence of DE cycles and age on good or bad outcomes, we tabulated the average number of embryos transferred, clinical pregnancy, implantation rate, delivery rate, number of multifetal reductions, triplet/quadruplet number, good rate, bad rate with its subcategories, and number of cases for [1] DE cycles 35 years old (egg donor 35 years old); [2] DE cycles 35 years old (egg donor 35 years old); [3] all non-de cycles 35 years old; [4] all non-de cycles 35 years old, and [5] total cases in women receiving one, two, three, four, or more embryos (see Table 3). The only statistically significant difference in the good/ bad rate was between all non-de 35 years old with the transfer of two embryos, compared with all non-de 35 years old with three or four embryos transferred (GR 19.7% vs. 30.4% and 33.1%, P.025 and.004, respectively). Because of the linear relationship between ES and GR (see Fig. 3), HMP risk data was developed for counseling at the time of ET that included the singleton, twin, and triplet FERTILITY & STERILITY 1537

5 TABLE 3 Outcomes for the transfer of two, three, and four or more embryos Age and classification Donor egg IVF 35 years Donor egg IVF 35 years All non donor egg IVF 35 years All non donor egg IVF 35 years Total for IVF cases Outcomes for one or two embryos transferred (twins: 9.8% [8/82], > triplets: 0) Avg. number transferred Clinical pregnancy rate 57.7 (15/26) 33.3 (5/15) 33.3 (37/111) 20.5 (25/122) 29.5 (82/278) Implantation rate Delivery rate 57.7 (15/26) 33.3 (5/15) 32.4 (36/111) 19.7 (24/122) 28.8 (80/278) Multifetal reduction Number of triplets/quadruplets Good rate a 57.7 (15/26) 33.3 (5/15) 32.4 (36/111) 19.7 (24/122) 28.8 (80/278) Bad rate a 42.3 (11/26) 66.7 (10/15) 67.6 (75/111) 80.3 (98/122) 71.2 (198/278) HMP/clinical pregnancy SAB/Stillbirths/pregnancy (1/37) 4 (1/25) 2.4 (2/82) No pregnancy/et 42.3 (11/26) 66.7 (10/15) 66.7 (74/111) 79.5 (97/122) 70.5 (196/278) Number of cycles b Outcomes for three embryos transferred (twins: 30.5% [146/478], > triplets: 9.8% [47/478]) Avg. number transferred Clinical pregnancy rate 65.8 (96/146) 52.3 (23/44) 53.5 (239/447) 36.5 (108/296) 50.3 (478/951) Implantation rate Delivery rate 59.6 (87/146) 47.7 (21/44) 49.4 (221/447) 31.1 (92/296) 45 (428/951) Multifetal reduction Number of triplets/quadruplets 10/0 2/0 28/0 2/0 42/0 Good rate a 49.3 (72/146) 43.2 (19/44) 43.2 (193/447) 30.4 (90/296) 40.1 (381/951) Bad rate a 50.7 (74/146) 56.8 (25/44) 56.8 (254/447) 69.6 (206/296) 59.9 (570/951) HMP/clinical pregnancy 15.6 (15/96) 8.7 (2/23) 11.7 (28/239) 1.9 (2/108) 9.8 (47/478) SAB/stillbirths/pregnancy 9.4 (9/96) 8.7 (2/23) 7.5 (18/239) 14.8 (16/108) 10.5 (50/478) No pregnancy/et 34.2 (50/146) 47.7 (21/44) 46.5 (208/447) 63.5 (188/296) 49.7 (473/951) Number of cycles b Outcomes for four or more embryos transferred (twins: 36.4% [77/211], > triplets: 11.4% [24/211]) Avg. number transferred Clinical pregnancy rate 55.6 (5/9) 44.4 (8/18) 46.1 (83/180) 41.0 (109/266) 43.7 (211/483) Implantation rate Delivery rate 55.6 (5/9) 38.9 (7/18) 43.9 (79/180) 36.8 (98/266) 40.2 (194/483) Multifetal reduction Number of triplets/quadruplets 0 0 8/3 6/1 14/4 Good rate a 44.4 (4/9) 38.9 (7/18) 36.6 (66/180) 33.1 (88/266) 35.1 (170/483) Bad rate a 55.6 (5/9) 61.1 (11/18) 63.3 (114/180) 66.9 (178/266) 64.8 (311/483) HMP/clinical pregnancy 20 (1/5) (13/83) 9.2 (10/109) 11.4 (24/211) SAB/stillbirths/pregnancy 0 (0/5) 12.5 (1/8) 4.8 (4/83) 10.1 (11/109) 8.1 (17/211) No pregnancy/et 44.4 (4/9) 55.6 (10/18) 53.9 (97/180) 59 (157/266) 56.3 (272/483) Number of cycles b a The good outcome rate (GR) was defined as singleton or twin deliveries per cycle, and the bad outcome rate included no pregnancy, or delivery, miscarriages (SAB, spontaneous abortion), need for multifetal reduction, triplet, or quadruplet pregnancies per cycle. b Total cases are greater than the sum of each subcategory evaluated because of missing data points (age or diagnosis) in subcategories, which eliminated the case for analysis in the subcategory but not in the total cases category. The average number transferred refers to the average number of embryos transferred in the group. The clinical pregnancy rate (PR) is the number of ultrasound-documented gestational sacs with a fetal pole demonstrating cardiac activity per embryo transfer (ET). The implantation rate is the number of ultrasound-documented gestational sacs with a fetal pole demonstrating cardiac activity divided by the number of embryos transferred. The delivery rate (DR) is the delivery of a viable infant(s) per ET. The multifetal reduction procedures are given for each group (11 total). The delivered triplets/quadruplets are tabulated for each group. The bad rate is further subdivided into HMP/clinical pregnancy, miscarriage (SAB, spontaneous abortion) or stillbirths/clinical pregnancy, and non-pregnant categories. The only statistically significant difference in the good/bad rate was between all non-de 35 years old with the transfer of two embryos, compared with all non-de 35 years old with three or four embryos transferred (19.7 vs and 33.1%, P.025 and.004, respectively). conception rates stratified according to ES (0 5, 6 10, 11 15, 16 20, 21 25, 26 30), age ( 35 and 35 years old, respectively), diagnosis, and the number of embryos transferred (three or four). There were 1,209 cases evaluated in this analysis. For example, Figure 2 shows the data for the diagnosis of male factor infertility with three embryos transferred at 35 years of age. The combination of the data depicted in Table 3 and the age, diagnosis, and ES-specific 1538 Peterson et al. Reducing high-order multiple pregnancies Vol. 81, No. 6, June 2004

6 FIGURE 3 Good outcome rate as a linear function of embryo score. The good outcome rate, defined as the percentage of singletons and twins delivered per cycle, shown on the y-axis (blue), demonstrates a linear function with embryo score (ES) shown on the x-axis (yellow), and is defined by the equation GR ES (green) within the range 4 ES 26. The correlation co-efficient is Red dots represent multiple data points. TABLE 4 Before and after pregnancy and delivery rates/embryo transfer for the total study groups and in the diagnostic categories of donor egg and non donor egg IVF. Outcome category Before After Total clinical pregnancy rate/et 44.7% (777/1,740) 46.3% (597/1,288) Total HMP rate 9.1% (71/777) 4.2% (25/597) a Donor egg HMP rate 11.8% (18/153) 1% (1/99) b Pregnancy rate/et 58.4% (153/262) 59.3% (99/167) Delivery rate/et 53.8% (141/262) 58.7% (98/167) Non donor egg HMP rate 8.7% (54/624) 4.8% (24/498) c Pregnancy rate/et 43% (624/1,452) 44.5% (498/1,121) Delivery rate/et 38.4% (557/1,452) 40.2% (451/1,121) Note: The clinical pregnancy and delivery rate/embryo transfer (ET) and high-order multiple pregnancy (HMP) rate/clinical pregnancy for all IVF cycles in and are tabulated. The clinic-specific HMP risk data resulted in a statistically significant reduction in the HMP rates for the total, donor egg, and non-donor egg IVF cycles without a reduction in the pregnancy or delivery rates in compared with a Statistically significant reduction, P.001. b Statistically significant reduction, P.003. c Statistically significant reduction, P.02. data (e.g., Fig. 2) were available for patient counseling during IVF cycles in 2000 to To assess the effect of the information provided by the 1995 to 1999 data on practice patterns, the outcomes of the ETs for January 2000 through September 2001 were also analyzed (n 597 clinical pregnancies of 1,288 ETs, 46.3%; and 549 deliveries of 1,288 ETs, 42.6%) and tabulated in Table 4. During this study interval, the total HMP rate fell from 9.1% (71 out of 777) in 1995 to 1999 to 4.2% (25 out of 597) in 2000 to 2001 (P.001). A subanalysis of individual physicians who achieved the greatest reduction in HMP revealed a 64% reduction: 7.6% (25 out of 330) in 1995 to 1999, to 2.7% (4 out of 155) in 2000 to 2001 (P.02), respectively. The clinical pregnancy rate for these physicians was 39.9% (330 out of 827) in 1995 to 1999 compared with 40.3% (155 out of 385) in 2000 to 2001 (P NS). The delivery rate for these physicians was 37.6% (311 out of 827) in 1995 to 1999 compared with 38.4% (148 out of 385) in 2000 to 2001 (P NS). DISCUSSION Large, multicenter studies of IVF programs have noted that three factors patient age, number of embryos transferred, and the ability to select embryos for transfer played significant roles in the success of an IVF procedure, as well as in the risk for multiple birth (9 19). A recent study using the data from 300 U.S. clinics reporting IVF transfer procedures to the Centers for Disease Control and Prevention further confirmed that multiple births from IVF are associated with maternal age and with the number of embryos transferred (6). Interpreting our historical data (1995 to 1999) and using an outcomes-based management approach (good vs. bad) to determine the number of embryos to transfer, we found the highest GR for DE cycles 35 years old was achieved with the transfer of two embryos (57.7%) (see Table 3). The highest risk for HMP in the study population was associated with the transfer of three or more embryos in DE cycles 35 years old; 17 DE IVF patients 35 years old receiving three or more embryos had triplets or quadruplets, which accounted for 25.3% (18 out of 71) of all HMP. For DE cycles 35 years old, the highest GR was achieved with the transfer of three embryos (43.2%) (see Table 3). The addition of a third embryo increased the GR by nearly 33% (43.2% vs. 33.3% for three and two embryos transferred, respectively). In all non-de cycles 35 years old, the maximum GR was attained with the transfer of three embryos (43.2%) (see Table 3). For all non-de cycles 35 years old, the GR was 30.4% and 33.1% for three and four embryos transferred, respectively (NS). The only statistically significant difference in the good/bad rates, illustrated in Table 3, was between all non-de cycles 35 years old with the transfer of two embryos, compared with all non-de cycles 35 years FERTILITY & STERILITY 1539

7 old with three or four embryos transferred (19.7% vs. 30.4% and 33.1%, P.025 and.003, respectively). The transfer of three or four embryos, however, resulted in a0to20% risk of HMP, depending on the diagnosis, highlighting the need for additional selection criteria such as ES. For example, Figure 3 demonstrates the linear relationship between the ES system and GR, confirming the potential utility of ES as an integral part of any protocol designed to reduce HMP in our programs (9 19). Analysis of singleton, multiple, and HMP rates by age ( 35 and 35 years old), diagnosis, and stratified according to ES and the number of embryos transferred (three or four), revealed that HMP occurred only in ETs with scores greater than 15 (e.g., see Fig. 2), except for instances when the diagnosis was endometriosis or tubal factor infertility. In these categories, HMP occurred when the ES was over 10. When considering any diagnosis or age, triplets occurred only with ES over 10 and fertilization rate greater than 50%, and all quadruplets occurred only with ES over 15 and fertilization rate over 70%. If fertilization rate and ES were used as the major decision points for the transfer of more than two embryos at ET, a third and/or fourth embryo could be transferred if the final ES was 10 and the fertilization rate was 50% (no risk of HMP based on pregnant cohort data 1995 to 1999). Only 32 cycles with ES 10 and fertilization rate 50% were encountered during the study period, suggesting this clinical situation would likely be uncommon; therefore, application of this strategy would not be expected to result in a statistically significant improvement in the GR, primarily because it would likely result in a reduction in the pregnancy and delivery rates. The estimated overall IVF pregnancy rate for such a strategy would be 30.7% (516 out of 1,680) based on the 1995 to 1999 data. In contrast, the use of HMP-specific risk data regarding age, diagnosis, and ES during informed consent for ET was applicable to the majority of cases, and assisted patients in understanding their personal risk of HMP (e.g., see Fig. 2). Previous studies have suggested a potential reduction in pregnancy and delivery rates using strict ET guidelines such as we have described (6). This study, using Table 3 supplemented by data specific to age, diagnosis, and ES (e.g., see Fig. 2), accepted, on a case-by-case basis, a marginal risk of HMP (approximately 5%). The use of this clinic-specific HMP risk data during informed consent resulted in a 53.8% reduction in HMP (9.1% to 4.2%) with no change in the overall pregnancy and delivery rates (see Table 4). Furthermore, among physicians experiencing the greatest reduction in HMP (7.6% to 2.7%), no statistically significant reduction was noted in the clinical pregnancy or delivery rates from 1995 to 1999 or 2000 to 2001 (pregnancy rate: 39.9% [330 out of 827] vs. 40.3% [155 out of 385]; delivery rate: 37.6% [311 out of 827] vs. 38.4% [148 out of 385], respectively). Cycle number (initial or repeat attempt) is likely an important factor in the GR from IVF; however, our sample size precluded using this variable in our analysis. Thus, all cycles were weighted equally, and the results of this study apply to the first four cycles in our programs. Despite the equal weighting of each cycle and regardless of cycle number, a reduction in HMP was accomplished without a reduction in pregnancy or delivery rates. Larger study populations, perhaps through SART data collection, may provide useful data specific to cycle numbers in the future. Because predetermination of those who will achieve a clinical pregnancy is impossible, a balance between a pregnancy/delivery-rate and outcomes-based management approach can result in excellent pregnancy and delivery rates while still reducing bad outcomes through consideration of the following points. First, transferring only two embryos in DE cycles with egg donors 35 years old eliminates over 25% of HMP without affecting the pregnancy or delivery rates. Triplets and quadruplets may occur in IVF cycles when the ES is over 10 and 15, respectively; clinic-specific HMP risk data assist in quantifying the specific risk. For DE cycles 35 years old, the highest GR is achieved with the transfer of three embryos (43.2%); however, this increase in good outcomes (33.3% vs. 43.2%) comes with over a 10% risk of a HMP. The addition of a third or fourth embryo in the diagnostic categories of endometriosis and tubal factor infertility results in a HMP rate of 2% to 10%, which depends on the ES. For male factor infertility and unexplained infertility, three embryos provide acceptable delivery rates ( 35 and 25%, respectively), with HMP occurring in those with ES greater than 15 at a rate of 5% to 20%, which increases with increasing ES (see Fig. 2). Attention to highrisk categories and adjustment for specific clinical situations using specific data for age, diagnosis, and ES can result in a 53% to 64% reduction in HMP rates (to 5%) without compromising the pregnancy or delivery rates. Recent data suggest non-de cycles (any age) with a previous IVF pregnancy may also have a higher risk for multiple implantations and thus for HMP with the transfer of three embryos. Finally, when patients are presented with clinicspecific data on HMP risk by age, ES, and diagnosis, the individual decision-making process regarding the number of embryos to transfer is greatly facilitated. This study cannot comment on outcomes of IVF cycles with three or four embryos transferred that had been preceded by a cycle or cycles with two embryos transferred; however, we have documented that equal weighting of all cycles in our programs resulted in clinically useful HMP risk data for counseling that resulted in a statistically significant reduction in HMP. Multicenter studies with large sample sizes in the future will allow analysis by cycle number. Further limitations of this study include the possibility that improved IVF techniques accounted for the absence of change in the pregnancy or delivery rates despite the reduction in embryos transferred, or that a notable change in the patient populations occurred between the study periods. Al Peterson et al. Reducing high-order multiple pregnancies Vol. 81, No. 6, June 2004

8 ternative strategies to considering clinic-specific HMP risk data before cleavage-stage ET (as used in this study) would be the transfer of only one or two cleavage-stage or blastocyst-stage embryos. This study documents the utility of identifying clinicspecific HMP risk data for ET counseling. A 53.8% reduction (P.001) in HMP was accomplished by a group of caregivers through a heightened awareness of program-specific risk factors that included age, diagnosis (particularly DE cycles), and ES. This was accomplished without a reduction in DE or non-de cycle pregnancy or delivery rates. The reduction in HMP was primarily (11.8% to 1.0%, P.003), but not exclusively due to DE cycles, with non-de cycles also showing a statistically significant reduction (8.7% to 4.8%, P.02). The physicians experiencing the greatest reduction in HMP (64% reduction, P.02) had no reduction in pregnancy or delivery rates; this suggests that with appropriate HMP risk data the decision-making process associated with ET can be improved without sacrificing the pregnancy or delivery rates, even among the most conservative physicians. National standards for ES reported with SART data (20), in addition to other markers (6), would assist in the establishment of specific guidelines for age, diagnosis, and embryo quality to help reduce HMP. Additional multicenter studies of factors associated with HMP from IVF will assist in avoiding the moral, medical, psychosocial, and monetary consequences of HMP. References 1. Kurinczuk JJ, Pemberton RJ, Binns SC, Parsons DE, Stanley FJ. Singleton and twin confinements associated with infertility treatments. Aust NZ J Obstet Gynecol 1995;357: Collins MS, Bleyl JA. Seventy-one quadruplet pregnancies. Am J Obstet Gynecol 1990;162: Ho ML, Chen JY, Ling UP, Chen JH, Huang CM, Chang CC, et al. Changing epidemiology of triplet pregnancy: etiology and outcome over 12 years. Am J Pernatol 1996;12: Callahan TL, Hall JE, Ettner SL, Christiansen CL, Greene MF, Crowley WF Jr. The economic impact of multiple gestation pregnancies and their contribution of assisted reproductive techniques to their incidence. N Engl J Med 1994;331: Lipitz S, Seidman DS, Alcalay M, Achiron R, Mashiach S, Reichman B. The effect of fertility drugs and in vitro methods on the outcome of 106 triplet pregnancies. Fertil Steril 1993;60: Schieve LA, Peterson HB, Meikle SF, Jeng G, Danel I, Burnett NM, et al. Live-birth rates and multiple-birth risks using in vitro fertilization. JAMA 1999;282: A Committee Opinion, Practice Committee Report of the American Society for Reproductive Medicine. Guidelines on number of embryos transferred. Birmingham, AL: ASRM, November Carrell DT, Peterson CM, Jones KP, Hatasaka HH, Udoff LC, Cornwell CE, et al. A simplified coculture system using autologous, attached cumulus tissue results in improved embryo morphology and pregnancy rates during in vitro fertilization. J Assist Reprod Genet 1999;16: Society for Assisted Reproductive Technology and the American Society for Reproductive Medicine. Assisted reproductive technology in the United States: 1998 results generated from the American Society for Reproductive Medicine/Society for Assisted Reproductive Technology Registry. Fertil Steril 2002;77: Templeton A, Morris JK. Reducing the risk of multiple births by transfer of two embryos after in vitro fertilization. N Engl J Med 1998;339: Hu Y, Maxson WS, Hoffman DI, Ory SJ, Eager S, Dupre J, et al. Maximizing pregnancy rates and limiting higher-order multiple conceptions by determining the optimal number of embryos to transfer based on quality. Fertil Steril 1998;69: Senoz S, Ben-Chetrit A, Casper RF. An IVF fallacy: multiple pregnancy risk is lower for older women. J Assist Reprod Genet 1997;14: Tasdemir M, Tasdemir I, Kodama H, Fukuda J, Tanaka T. Two instead of three embryo transfer in in-vitro fertilization. Hum Reprod 1995;10: Staessen C, Janssenswillen C, Van Den Abbeel E, Devroey P, Van Steirteghem AC. Avoidance of triplet pregnancies by elective transfer of two good quality embryos. Hum Reprod 1993;8: Staessen C, Camus M, Bollen N, Devroey P, Van Steirteghem AC. The relationship between embryo quality and the occurrence of multiple pregnancies. Fertil Steril 1992;57: Waterstone J, Parsons J, Bolton V. Elective transfer of two embryos. Lancet 1991;337: Elsner CW, Tucker M, Sweitzer CL, Brockman WD, Morton PC, Wright G, et al. Multiple pregnancy rate and embryo number transferred during in vitro fertilization. Am J Obstet Gynecol 1997;177: Svendsen TO, Jones D, Butler L, Muasher SJ. The incidence of multiple gestations after in vitro fertilization is dependent on the number of embryos transferred and maternal age. Fertil Steril 1996;65: Widra EA, Gindoff PR, Smotrich DB, Stillman RJ. Achieving multipleorder embryo transfer identifies women over 40 years of age with improved in vitro fertilization outcome. Fertil Steril 1996;65: Fertility Clinic Success Rate and Certification Act of 1992 (FCSRCA) (Pub. L , 42 U.S.C. 263a-1 et seq.), October 24, FERTILITY & STERILITY 1541