Although population-based estimates of the incidence of ectopic pregnancy among assisted reproductive technology (ART) patients have not been reported in the United States, results of small clinical studies published from 1991 through 1995 suggest that the ectopic rate associated with ART conceptions may be elevated, between 2.2 and 8.6 per 100 pregnancies, compared with the estimated rate of 2.0 per 100 pregnancies for the general U.S. population in 1990–1992.1–8
Theoretically, differences between conception via ART and natural conception may affect the risk of ectopic pregnancy. As currently performed, the vast majority of ART procedures involve in vitro fertilization and transcervical embryo transfer (IVF-ET) of multiple embryos.9 Because neither fertilization nor embryo transfer in IVF-ET involves the fallopian tubes directly, ART might reduce the risk of ectopic pregnancy. Nonetheless, ectopic pregnancies have been documented in IVF-ET cycles, raising questions about the etiology of these ectopic implantations, including whether the transfer of multiple embryos plays a role.
The majority of information on ectopic pregnancies among ART conceptions stems from case reports or case series. The few studies with denominator data were small and thus unable to evaluate sufficiently the risk of ectopic pregnancy by important patient subgroups. Limited data from previous studies suggest that tubal factor infertility and prior ectopic pregnancy are associated with an increase in the risk for ectopic pregnancy after ART.1–3,10,11
We assessed the incidence of ectopic pregnancy among ART conceptions using data from the large, population-based, U.S. registry of ART procedures. These data also allowed us to perform a more detailed analysis of risk factors for ectopic pregnancy among ART conceptions than has previously been reported.
MATERIALS AND METHODS
Providers of ART in the United States are required to report data on every procedure to the Centers for Disease Control and Prevention (CDC) annually.12 Assisted reproductive technology is defined as any treatment where oocytes and sperm are handled outside the body for the purpose of establishing a pregnancy. The Society for Reproductive Technology (SART) compiled a data file from participating ART clinics and shared these data with the CDC (the U.S. ART Registry). Data were abstracted from medical records and included patient demographic characteristics, medical history, and clinical information on the ART procedure and resultant pregnancies. From 92% to 95% of U.S. clinics reported data annually to the U.S. ART Registry.13
For this study, we selected procedures performed between January 1, 1999, and December 31, 2001, that resulted in a reported clinical intrauterine, ectopic, or heterotopic pregnancy. In the ART Registry, a clinical intrauterine pregnancy was defined as documentation of one or more gestational sacs visible by ultrasound examination; ectopic pregnancy was defined as documentation of one or more gestational sacs outside of the uterus; and heterotopic pregnancy was defined as a pregnancy that met the criteria for both ectopic and clinical intrauterine pregnancy. Pregnancies reported as biochemical pregnancies only (defined in the registry as having an elevated human chorionic gonadotropin [hCG] without a visible gestational sac and no clinical diagnosis of pregnancy) were excluded from the analysis.
From 1999 to 2001, 294,852 ART procedures were reported to the ART Registry. Of these procedures, 94,700 resulted in either an ectopic, heterotopic, or intrauterine pregnancy. From this group of all ART pregnancies, we excluded a small number of pregnancies that resulted from less common treatment options (< 1%). These uncommon options included the following: procedures in which any combination of IVF-ET, gamete intrafallopian transfer (GIFT), or zygote intrafallopian transfer (ZIFT) were used for transfer (n = 176), procedures in which both frozen-thawed and freshly fertilized embryos were transferred (n = 120), procedures in which embryos from both donor and patient oocytes were transferred (n = 109), and GIFT and ZIFT procedures that involved either donor oocytes or frozen-thawed embryos (n = 170). We also excluded pregnancies for which the improbable transfer of 15 or more embryos was reported (n = 7). The resulting study population thus consisted of 94,118 pregnancies.
We classified a pregnancy as ectopic if it was reported as either ectopic only or heterotopic. We calculated ectopic incidence by dividing ectopic pregnancies by the total number of pregnancies (ectopic + clinical intrauterine pregnancies). Rates were presented as per 100 pregnancies. We also calculated heterotopic incidence rates by dividing the number of pregnancies classified as heterotopic by the total number of pregnancies (ectopic + clinical intrauterine pregnancies).
In our initial analyses, we calculated ectopic rates for pregnancies achieved through various ART procedure types, which we classified according to whether the embryos were fertilized during the current procedure (fresh) or had been fertilized during a previous procedure and frozen until the current procedure (frozen) and by whether the source of the oocyte was the patient (nondonor) or an oocyte donor. We calculated rates for procedures in which a surrogate was used to gestate the pregnancy and for GIFT and ZIFT procedures separately.
Because significant variation in ectopic rates was found across procedure types, these types were not combined for further analysis of risk factors. Instead, more detailed analyses were restricted to pregnancies that resulted from the most common type of ART: IVF with transcervical transfer of fresh nondonor embryos. In all, 69,366 clinical intrauterine and ectopic pregnancies were selected for these analyses (74% of our original study population). Risk factors evaluated included patient age, race/ethnicity, previous births, previous spontaneous abortion, and infertility diagnosis; year of ART procedure; use of intracytoplasmic sperm injection (ICSI) (in which a single sperm is injected directly into an oocyte); use of assisted hatching (in which lasers, chemicals, or other means are used to create an opening in the zona pellucida of the embryo before transfer); number of days in embryo culture; number of embryos transferred; and whether extra embryos were available and cryopreserved for future use.
Although the U.S. ART Registry does not contain data on specific embryo quality assessments, 2 factors have been associated with increased implantation potential (higher estimated embryo potential): extra embryos available for cryopreservation and the number of days in embryo culture.14,15 Although our data are observational and not randomized, and thus both of the aforementioned factors are limited by provider and patient choices of specific ART treatment practices, these 2 variables nonetheless have been predictive of success rates in past analyses, independent of patient age or the number of embryos transferred, and thus are likely to be correlated with a higher estimation of embryo implantation potential.14,15
During preliminary unadjusted analyses, we initially divided infertility diagnoses into 14 distinct categories. For stratified and multivariable analyses, sample size constraints necessitated that we collapse these categories into 5: male factor, tubal pathology with or without a hydrosalpinx, tubal ligation, endometriosis, and nontubal female factors (which include ovulatory dysfunction, uterine factor, diminished ovarian reserve, immunologic factors, infertility related to chemotherapy or other chronic disease, and unexplained infertility). Pregnancies that met the criteria for only one infertility diagnosis group were assigned to that group. Pregnancies among women meeting the criteria for more than 1 of the 5 diagnosis categories were assigned to the group that had the highest ectopic rate in our initial analyses. That is, the ectopic rate for tubal pathology was greater than that for nontubal female factors, which was greater than that for endometriosis, which was greater than that for male factor, which was greater than that for tubal ligation. As a result of this algorithm, only those in the tubal ligation group had a single diagnosis of infertility.
We assessed simple (unadjusted) associations between ectopic pregnancy risk and each factor of interest with χ2 tests. On the basis of epidemiologic studies and/or biologic plausibility, we identified 3 a priori risk factors for ectopic pregnancy: type of infertility diagnosis, number of embryos transferred, and use of assisted hatching. Through our analyses, we determined that type of infertility diagnosis and the number of embryos transferred were potentially important confounders or effect modifiers. We further assessed associations between ectopic pregnancy risk and each of the other factors of interest after stratification on infertility diagnosis and stratification on number of embryos transferred. We also performed multivariable logistic regression to independently assess associations between maternal and ART treatment factors. The model included patient age, prior spontaneous abortions, prior births, infertility diagnosis, use of assisted hatching, use of ICSI, and one interaction term: estimated embryo implantation potential by number of embryos transferred. Estimated implantation potential was a 4-level categorical variable derived from 2 variables included in the initial (unadjusted) analysis, day of embryo transfer (day 3 versus day 5), and availability of extra embryos that were cryopreserved (yes versus no). Number of embryos transferred was classified as 1–2 versus 3+. The interaction term of implantation potential by number of embryos transferred included 8 levels; the referent group was 3+ embryos transferred with no indication of high implantation potential (day 3 transfer and no extra embryos cryopreserved). The data were analyzed in SAS 9.1 (SAS Institute Inc, Cary, NC). This study was approved by the CDC’s Institutional Review Board.
Of the 94,118 pregnancies included in our final study population, 2,009 were reported as ectopic, 143 of which were heterotopic. Thus, the overall rate of ectopic pregnancy for women undergoing ART procedures in the United States from 1999 to 2001 was 2.1%, and the heterotopic pregnancy rate was 0.15%.
The ectopic rate varied significantly by ART procedure type (Table 1). The ectopic rate among fresh nondonor IVF-ET treatment cycles, the most widely used type of ART, accounting for 74% of the pregnancies, was 2.2%. In comparison, the ectopic rate among fresh nondonor ZIFT procedures was significantly increased, at 3.6%. The ectopic rate was significantly decreased among fresh donor IVF-ET (1.4%) and gestational surrogate procedures (0.9%).
More than half of the pregnancies conceived with fresh nondonor IVF-ET procedures were among women less than 35 years of age (Table 2). Although a large proportion (37.2%) of pregnancies were missing data for race/ethnicity, the most common racial-ethnic group reported was white, non-Hispanic. For a majority of pregnancies, no spontaneous abortions had been reported and no previous births were reported. The 2 most common infertility diagnoses were tubal factor (excluding tubal ligation) (24.6%, n = 17,087) and male factor (38.0%, n = 26,355) (alone, or in combination with other factors). Assisted hatching was used in 40.8% of the treatments, and ICSI was used in 52.7%. Nearly three fourths of the pregnancies resulted from embryos cultured for 3 days and 17.5% from embryos cultured for 5 days (corresponding with the blastocyst stage). Although the number of embryos transferred varied widely, 70.5% of pregnancies involved the transfer of 3 or more embryos. More than one third of the women who became pregnant after IVF-ET had extra embryos available for cryopreservation.
In unadjusted analyses, we observed significant variation in ectopic rate by prior births, infertility diagnosis, use of ICSI, and number of embryos transferred (Table 3). Compared with ectopic rates among couples with male factor infertility, ectopic rates among women with tubal pathology (with/without hydrosalpinx), uterine factor, endometriosis, diminished ovarian reserve, infertility relating to immunologic factors, chemotherapy or other chronic disease, unexplained infertility, multiple female and male factors with tubal pathology, or multiple female factors (with or without tubal pathology) were each significantly increased. For example, women diagnosed with tubal pathology with hydrosalpinx had a 4.2% ectopic rate, more than 2.5 times higher than the rate among couples diagnosed with male factor infertility only. Women treated with ART because of a tubal ligation only had the lowest ectopic rate of all infertility diagnoses (1.0%). Use of ICSI was associated with a decreased ectopic rate. The transfer of 3 embryos or 4 or more embryos was associated with an increased ectopic rate.
Because a disproportionate number of pregnancies conceived using ICSI were among couples with male factor infertility (a group at relatively low risk for ectopic pregnancy), the apparent protective effect initially observed with use of ICSI disappeared after stratification and adjustment for infertility diagnosis (odds ratio [OR] 0.94, 95% confidence interval [CI] 0.8–1.0). There was no difference in the risk for ectopic pregnancy among ICSI and non-ICSI pregnancies within specific infertility diagnosis strata.
Our initial analyses indicated that women with a race/ethnicity other than white non-Hispanic were at a greater risk for ectopic pregnancy than white (non-Hispanic) women, but this association varied according to infertility diagnosis (Table 4). Race/ethnicity other than non-Hispanic white was associated with increased ectopic rates among women seeking ART for either male factor infertility or endometriosis. In contrast, there was no variation by race/ethnicity among women diagnosed with tubal pathology or nontubal female factors. The results for the tubal ligation group are based on a very small sample and thus were not appropriately powered to assess statistical significance. Also, it is important to note that 37% of the data for race/ethnicity were missing and thus were not included in these analyses.
The association between ectopic pregnancy risk and embryo implantation potential, based on the 2 indicators we were able to assess, number of days in embryo culture and availability of extra embryos for cryopreservation, varied according to the number of embryos transferred. The ectopic risk among pregnancies in which 3 or more embryos had been transferred was 2.4–2.5%, whether or not there was an indication of higher embryo implantation potential. However, when only 1–2 embryos were transferred, ectopic rates varied according to embryo implantation potential indicators: 2.2% with neither indicator present, 1.6% when extra embryos had been available and cryopreserved, 1.4% when embryos were cultured for 5 rather than 3 days, and 1.4% when both of these conditions were met. These latter 3 rates were significantly (P < .05) different from the referent group, 3+ embryos transferred with neither indication for higher implantation potential.
The unadjusted results (Table 3) for the risk of ectopic pregnancy according to prior births, prior spontaneous abortions, year of current ART procedure, and use of assisted hatching, were not appreciably altered by stratification and adjustment for infertility diagnosis and number of embryos transferred (data not shown).
The final logistic regression model included maternal age, prior spontaneous abortions, prior births, infertility diagnosis, use of assisted hatching, and use of ICSI as independent variables. Additionally, we included an interaction term constructed by using number of embryos transferred and estimated embryo implantation potential (we combined 2 factors into a single estimate of embryo implantation potential: days in embryo culture and extra embryos cryopreserved). Race/ethnicity was excluded from the final logistic regression model because of the large amount of missing data. The results of the logistic regression analysis were consistent with the results of the stratified analyses (Table 5). Women with tubal pathology (with or without a hydrosalpinx) were 2.0 times more likely to have an ectopic pregnancy than women treated with ART because of male factor infertility (95% CI 1.7–2.4). Endometriosis conferred a modest risk (OR 1.3, 95% CI 1.0–1.6). Nontubal female factors also conferred a modest risk (OR 1.4, 95% CI 1.2–1.6). Having a previous birth was protective (OR 0.6, 95% CI 0.5–0.7). The transfer of 2 embryos or fewer was protective among 3 subgroups of women with at least one indication of higher embryo implantation potential (ORs 0.6–0.7). Maternal age, prior spontaneous abortions, use of assisted hatching, and ICSI were not significant predictors of ectopic pregnancy.
The ectopic pregnancy rate reported here (2.1%) for women who conceived with ART in the United States from 1999 to 2001 was below the range of rates reported from previous clinical studies of ART patients (2.2–8.6%) and comparable with the estimated ectopic rate among conceptions in the general U.S. population, (2.0%).1–8 The rate of heterotopic pregnancy among ART users was 0.15%.
Our results suggest that 3 main factors are important in assessing the risk of ectopic pregnancy among ART users: the specific type of ART procedure, the reproductive health characteristics of the woman carrying the pregnancy, and the estimated embryo implantation potential.
Our observation of a significant increase in the risk of ectopic pregnancy after ZIFT, in comparison with fresh nondonor IVF-ET procedures, is perhaps somewhat intuitive because embryos are transferred into the fallopian tubes in ZIFT. However, we did not observe this effect with GIFT procedures. Additionally, we lacked the data to explore the ZIFT association by the location of ectopic pregnancy and the side of the embryo transfer. Limited sample size also precluded analysis of specific patient or treatment risk factors among ZIFT pregnancies.
Ectopic pregnancy risk was associated with several measures of women’s reproductive health status. Women with tubal factor infertility had a 2-fold increase in risk, and women with nontubal female factors of infertility or endometriosis had a 30–40% increased risk of ectopic pregnancy. It is possible that women classified as only having infertility related to nontubal female factors might nonetheless have had tubal pathology that was not diagnosed or reported. Dubuisson et al1 reported tubal damage among women with ectopic pregnancies who had used IVF because of endometriosis or unexplained infertility. Likewise, our finding of an increased ectopic pregnancy risk among women with race/ethnicity other than non-Hispanic white (with a diagnosis of endometriosis or male factor infertility) might be related to tubal pathology if tubal pathology in these women were less likely to be diagnosed or reported.
Women who had a prior birth, a group that has demonstrated capacity to have an intrauterine pregnancy, were less likely to have an ectopic implantation in the index pregnancy for this study. Further, the decreased risk in ectopic pregnancy among gestational carriers supports the contribution of host characteristics to the risk of ectopic pregnancy during ART.
In this study, women who had embryos with at least one indication of higher estimated embryo implantation potential were at decreased risk of ectopic pregnancy if 2 or fewer embryos were transferred. The apparent benefit, however, was not evident when 3 or more embryos were transferred. The protective effect we observed among procedures using donor oocytes further supports the hypothesis that embryo implantation potential is associated with risk because, in most clinics, oocyte donors are young women without an indication of infertility. The underlying mechanism for these findings related to embryo characteristics requires further study. Although prior studies suggested that chromosomal abnormalities may play a role in the etiology of ectopic pregnancy, results across studies are inconsistent, and both positive and negative studies were limited by small sample sizes.16–18
A recent study suggested that the assisted hatching procedure was associated with an increased risk of ectopic pregnancy.19 Our findings do not support that conclusion, but the ART Registry does not include data to separately assess specific types of assisted hatching procedures.
The primary strength of our study was the ability to investigate potential ectopic pregnancy risk factors with a large, population-based sample. Most previous studies of ectopic pregnancy were clinic based, and the results may not be generalizable. Our large sample of ART pregnancies also allowed us to assess the risk of ectopic pregnancy among several ART subgroups. Although the sample size for individual subgroups varied, for most, our results were stable, as evidenced by the narrow confidence limits around our measures of associations.
Even though we report a lower ectopic pregnancy rate than previous ART studies, our results are nonetheless consistent with these prior reports. All suggest that the condition of the fallopian tubes is central to a woman’s risk for ectopic pregnancy.1–7,11 Many previous studies were published in the early 1990s, when tubal factor infertility was the major indication for the use of IVF-ET.20 In these studies, 43–74% of clinical pregnancies were among patients with a diagnosis of tubal factor infertility, whereas in our study, only 24.6% of clinical pregnancies were among patients with a tubal factor diagnosis (alone or in combination with other factors, excluding tubal ligation), as ART is now more commonly used among couples diagnosed with nontubal infertility disorders.1,6,7 Indeed, in our population, the ectopic rates of pregnancies among patients with a sole diagnosis of tubal pathology (4.2% and 3.0% for tubal pathology with and without a hydrosalpinx, respectively) fall within the range of ectopic rates observed in earlier ART studies (2.2–8.6%).1–7
The major limitation of this study is that the ART Registry lacks specific clinical data pertaining to embryo factors, transfer techniques (including ultrasound-guided embryo transfer), or history of ectopic pregnancy. Therefore, we were unable to explore underlying mechanisms for the associations reported here.
Another potential concern was the accuracy and completeness of ectopic pregnancy reporting in the ART Registry. However, we found that no single clinic accounted for a disproportionate number of ectopic pregnancies and that ectopic pregnancies were diagnosed and reported by most clinics (data not shown). In addition, ART treatment outcome data for 1999–2001 were validated by trained ART clinicians during on-site medical record reviews at 5–10% of ART clinics included in the ART Registry and were found to be highly concordant with the data in the medical record (< 1% misreporting rate). Finally, because women undergoing ART are routinely monitored by the ART provider with pregnancy testing and ultrasonography, we assume that this surveillance system, based on ART providers, is sensitive and specific at capturing ectopic pregnancies. Even though ectopic pregnancy was narrowly defined in the ART Registry as the documentation of one or more gestational sacs outside the uterus, we assume that ART providers would likely have recorded an ectopic pregnancy even if it was identified by a means other than ultrasound examination (eg, the absence of an intrauterine sac with a hCG level above the discriminatory zone with a clinical history consistent with ectopic pregnancy). Our analysis of the distribution of pregnancies across clinics indicated that, not only was the reporting of ectopic pregnancies evenly distributed, but reported biochemical pregnancies were also distributed evenly (most clinics reporting ectopic pregnancy were also reporting biochemical pregnancies). Thus, the data suggest clinics were distinguishing these 2 types of pregnancy, rather than select clinics conservatively reporting their ectopic pregnancies as biochemical. However, because we do not have data on whether some women may have been diagnosed with an ectopic pregnancy after leaving the care of their ART provider, we cannot discount the possibility that some ectopic pregnancies were missed.
This study demonstrated that the transfer of 2 or fewer embryos with higher estimated implantation potential was protective against ectopic pregnancy. Further research into the relationship between embryo implantation potential and ectopic pregnancy might be able to identify the mechanism behind this reduction in risk. Also, as new technologies in ART become available, their potential impact on ectopic pregnancy should be investigated.
1. Dubuisson JB, Aubriot FX, Mathieu L, Foulot H, Mandelbrot L, de Joliere JB. Risk factors for ectopic pregnancy in 556 pregnancies after in vitro fertilization: implications for preventive management. Fertil Steril 1991;56:686–90.
2. Karande VC, Flood JT, Heard N, Veeck L, Muasher SJ. Analysis of ectopic pregnancies resulting from in-vitro fertilization and embryo transfer. Hum Reprod 1991;6:446–9.
3. Marcus SE, Brinsden PR. Analysis of the incidence and risk factors associated with ectopic pregnancy following in-vitro fertilization and embryo transfer. Hum Reprod 1995;10:199–203.
4. Nazari A, Askari HA, Check JH, O’Shaughnessy, A. Embryo transfer technique as a cause of ectopic pregnancy in in-vitro fertilization. Fertil Steril 1993;60:919–21.
5. Pyrgiotis E, Sultan KM, Neal GS, Liu HC, Grifo JA, Rosenwaks Z. Ectopic pregnancies after in vitro fertilization and embryo transfer. J Assist Reprod Genet 1994;11:79–84.
6. Ribic-Pucelj M, Tomazevic T, Vogler A, Meden-Vrtovec H. Risk factors for ectopic pregnancy after in vitro fertilization and embryo transfer. J Assist Reprod Genet 1995;12:594–8.
7. Verhulst G, Camus M, Bollen N, Van Steirteghem, A, Devroey P. Analysis of the risk factors with regard to the occurrence of ectopic pregnancy after medically assisted procreation. Hum Reprod 1993;8:1284–7.
8. Centers for Disease Control and Prevention. Current trends in ectopic pregnancy-United States, 1990–1992. MMWR Surveill Summ 1995;44(03):46–8.
9. Wright VC, Schieve LA, Reynolds MA, Jeng G. Assisted reproductive technology surveillance–United States, 2002. MMWR Surveill Summ 2005;54:1–24.
10. Lesny P, Killick SR, Robinson J, Maguiness SD. Transcervical embryo transfer as a risk factor for ectopic pregnancy. Fertil Steril 1999;72:305–9.
11. Strandell A, Thorburn J, Hamberger L. Risk factors for ectopic pregnancy in assisted reproduction. Fertil Steril 1999;71:282–6.
12. Fertility Clinic Success Rate and Certification Act of 1992 (FCSRCA). Pub L No. 102-493. October 24, 1992.
13. Schieve LA, Wilcox L, Zeitz J, Jeng G, Hoffman D, Brzyski R, et al. Assessment of outcomes for ART: overview of issues and the U.S. experience in establishing a surveillance system. In: Vayena, E, Rowe PJ, Griffin PD, editors. Current practices and controversies in assisted reproduction: report of a meeting on “Medical, Ethical and Social Aspects of Assisted Reproduction” held at WHO Headquarters in Geneva, Switzerland, September 2001. Geneva, Switzerland: World Health Organization; 2002. p. 361–76.
14. Schieve L, Peterson HB, Meikle SF, Jeng G, Danel I, Burnett NM, Wilcox LS. Live- birth rates and multiple-birth risk using in vitro fertilization. JAMA 1999;282:1832–8.
15. Kissin DM, Schieve LA, Reynolds MA. Multiple-birth risk associated with IVF and extended embryo culture: USA, 2001. Hum Reprod 2005;20:2215–23.
16. Karikoski R, Aine R, Heinonen PK. Abnormal embryogenesis in the etiology of ectopic pregnancy. Gynecol Obstet Invest 1993;36:158–62.
17. Toikkanen S, Joensuu H, Erkkola R. DNA aneuploidy in ectopic pregnancy and spontaneous abortions. Eur J Obstet Gynecol Reprod Biol 1993;51:9–13.
18. Goddijn M, van der Veen, F, Schuring-Blom GH, Ankum WM, Leschot NJ. Cytogenetic characteristics of ectopic pregnancy. Hum Reprod 1996;11:2769–71.
19. Jun SH, Milki AA. Assisted hatching is associated with a higher ectopic pregnancy rate. Fertil Steril 2004;81:1701–3.
20. Kodaman PH, Arici A, Seli E. Evidence-based diagnosis and management of tubal infertility. Curr Opin Obstet Gynecol 2004;16:221–9.