Endometriosis and Assisted Reproductive Technology: United States Trends and Outcomes 2000–2011

The absolute number and percentage of ART cycles with an endometriosis diagnosis showed a significant quadratic trend ( Fig. 1A ). The number rose from 16,751 in 2000 to 17,902 in 2003 and then fell to 15,311 in 2011. The percentage of ART cycles with any diagnosis of endometriosis fell over time, from 17.0% in 2000 to 9.6% in 2011. A similar significant decreasing quadratic trend was also noted among endometriosis-associated first ART cycles. The average number of oocytes retrieved among endometriosis-associated cycles also had a significant quadratic trend, rising to a peak of 12.3 in 2006 (ranging from 11.9 to 12.3). The percentage of cancelled ART cycles among patients with endometriosis significantly decreased over time, from 14.5% in 2000 to 10.0% in 2011 ( Fig. 1B ), whereas the pregnancy rate and live birth rates per transfer showed a significant quadratic trend ( Fig. 1C ). The pregnancy rate rose from 40.0% (2000) to 48.3% (2008) and then fell to 45% (2011). The live birth rate rose from 33.6% (2000) to 40.0% (2008) and then fell to 37.3% (2011). Between 2000 and 2011, NASS included 488,032 fresh autologous IVF cycles for which either endometriosis or male factor was among the diagnoses for infertility, excluding those with both endometriosis-associated and male factor infertility ( Table 1 ). A higher percentage of women in the endometriosis-associated group as compared with male factor reported race as non-Hispanic white, held a concomitant diagnosis of tubal factor infertility, and had to be cancelled before retrieval. A lower percentage of women in the endometriosis-associated group held a concomitant diagnosis of ovulatory dysfunction or diminished ovarian reserve, and reported a complication of ovarian hyperstimulation. Between 2000 and 2011, NASS included 406,255 noncancelled fresh autologous IVF cycles resulting in transfer among women with endometriosis or male factor infertility ( Table 2 ); 81,777 cycles were cancelled either before retrieval or before transfer. Statistically significant ( P <.0001 for all comparisons) differences were detected between male factor- and endometriosis-associated cycles in the distribution of the following cycle characteristics: number of oocytes retrieved (endometriosis-associated with 50.5% vs. male factor with 42.5% of cycles with 0–10 oocytes retrieved), use of ICSI (endometriosis-associated 52.3% vs. male factor 92.3%), day of ET (endometriosis-associated 67.4% day 3 vs. male factor 63.4% day 3), number of embryos transferred (endometriosis-associated 48.3% transfer of three or more embryos vs. male factor 45.0%), and elective single embryo transfer (endometriosis-associated 1.9% vs. male factor 2.6%). For both groups, the majority of cycles (67.0% endometriosis and 65.6% male factor) had no embryos cryopreserved. Of the cycles resulting in pregnancy, the majority of women in both groups delivered a singleton live-born infant (55.5% endometriosis and 56.1% male factor); the incidence of twins was also comparable (25.8% endometriosis and 25.0% male factor). As compared with male factor infertility, endometriosis was associated with a higher mean total FSH dose (3,274.3 IU vs. 3,106.6 IU, adjusted estimate 224.53, 95% CI 179.1–270.0), and with a significantly lower risk of hyperstimulation (1.1% vs. 1.3%, adjusted RR [aRR] 0.82, 95% CI 0.74–0.91) and an increased risk of cancellation (12.9% vs. 10.1%, aRR 1.30, 95% CI 1.25–1.35) ( Table 3 ). Hospitalization rates were not significantly different between the two groups (0.3% for both, aRR 0.97, 95% CI 0.84–1.11). E 1Among noncancelled cycles for which either a day-3 or day-5 transfer was performed, the average implantation rate in the endometriosis group was lower than in the male factor group (25.3% vs. 26.3%, aRR 0.96, 95% CI 0.94–0.97). Also for cycles resulting in either a day-3 or day-5 transfer, endometriosis was associated with a decreased chance of pregnancy (43.7% vs. 44.8%, aRR 0.96, 95% CI 0.95–0.98), a decreased chance of miscarriage (5.8% vs. 6.3%, aRR 0.93, 95% CI 0.89–0.97), and, among couples who did not have tubal factor infertility, a decreased chance of live birth (37.2% vs. 37.6%, aRR 0.96, 95% CI 0.94–0.98).

We assessed endometriosis trends among all ART cycles, including fresh, frozen, and banking cycles, performed in the United States between 2000 and 2011 (n = 1,589,079) using NASS, a nationally mandated reporting system that includes approximately 97% of all ART cycles performed in the United States (Fertility Clinic Success Rate and Certification Act of 1992 [ 15 , 16 ], Public Law No. 102–493, October 24, 1992). The NASS data, which are ART cycle-based, include patient demographics, obstetric and medical history, infertility diagnoses, detailed parameters of each ART treatment cycle and, if applicable, the resultant pregnancy outcome. Regression analysis was used to assess linear and quadratic trends over time in absolute number and percentage of ART cycles including any diagnosis of endometriosis and cancellation, pregnancy, and live birth rates among endometriosis-associated cycles for which a transfer was performed. The primary outcomes of interest were treatment complications (hyperstimulation, hospitalization, and cancellation) and pregnancy outcomes (pregnancy, live birth, and miscarriage). Cycles were considered to result in pregnancy if they had an outcome of clinical intrauterine gestation, defined as ultrasound confirmation of gestational sac(s) within the uterus, regardless of whether a heartbeat(s) was observed or fetal pole established. When ultrasound data were not available, confirmation was achieved through documented birth, spontaneous miscarriage, or induced abortion. A cycle was considered to result in live birth when at least one live-born infant was delivered at ≥20 weeks’ gestation and as a miscarriage if the pregnancy outcome occurred at <20 weeks’ gestation. Secondary characteristics of interest included total FSH medication dose and implantation rate, defined as the number of fetal heartbeats at 6-week ultrasound per number of embryos transferred. Bivariate analyses were conducted to explore the relationship between infertility diagnosis and other maternal and cycle characteristics. Infertility diagnosis was classified as endometriosis or male factor, excluding those with both endometriosis and male factor infertility, but allowing for other concomitant diagnoses. Male factor infertility was chosen as the comparison group because it suggests a female without known infertility. We did not include women with unexplained infertility because women with either endometriosis or male factor infertility were, by definition, not “unexplained.” Chi-square tests were used to test for statistical significance between infertility type and categorical characteristics, whereas the Wilcoxon rank sum test was used for associations between infertility type and continuous characteristics. Log binomial regression was used to generate unadjusted and adjusted risk ratios (RRs), 95% confidence intervals (CI), and P values to compare complication rates per cycle start and success rates per ET among fresh, nondonor ART cycles performed for couples with endometriosis-associated infertility (n = 112,475) as compared with male factor infertility (n = 375,557). Poisson regression was used to model the number of embryos implanted per number of embryos transferred, and linear regression was used to model FSH levels. Generalized estimating equations with an independent correlation matrix were used to account for clustering by clinic. Characteristics controlled for in the multivariable models varied by outcome, as determined by their statistical significance in bivariate analysis. For complication outcomes, the possible list of covariates included maternal age, obstetric history (number of prior pregnancies, spontaneous miscarriages, preterm births, and full-term births), number of prior ART cycles, concomitant infertility diagnoses, and year. For pregnancy outcomes, the possible list of covariates included the same variables previously listed as well as use of intracytoplasmic sperm injection (ICSI), use of assisted hatching, the embryo stage at transfer, the number of embryos transferred, the number of supernumerary embryos cryopreserved, and the number of oocytes retrieved. Race or ethnicity was included in the bivariate tables but not in the multivariable analysis because this factor was missing for approximately 40% of the observations. We explored the possibility of an interaction between type of infertility (endometriosis or male factor) and tubal factor because women with endometriosis-induced tubal factor may have a noninfectious cause as compared with those with tubal factor that resulted from history of prior pelvic infection. The interaction was significant when modeling live birth; therefore, RRs are reported by the presence or absence of tubal factor. Statistical significance was determined using an α of 0.05. All analyses were conducted using SAS v. 9.3 (SAS Institute). This study was approved by the institutional review board of the Centers for Disease Control and Prevention.

The percentage of endometriosis-associated ART cycles decreased over the 12-year study period. The decreasing trend in endometriosis-associated cycles may reflect not only a decrease in diagnosis of endometriosis among women using ART but also an increase in alternate indications for ART. The decrease in diagnosis may reflect the tendency to perform IVF immediately rather than after diagnostic laparoscopy among women with unexplained infertility. As compared with male factor infertility, endometriosis is associated with increased total medication dose and cancellation risk and with decreased oocyte yield and hyperstimulation risk. An initial assumption may be that the lower oocyte yield, increased medication requirement, increased cancellation, and decreased ovarian hyperstimulation rates may reflect diminished ovarian reserve among patients with endometriosis ( 17 ). However, the fact that the diagnosis of diminished ovarian reserve was more prevalent among the male factor group and the fact that the association with cancellation persists after adjusting for age and concomitant diagnoses including diminished ovarian reserve suggests that endometriosis itself may confer an altered response to high-dose gonadotropins. Among cycles for which transfer was performed, the likelihood of live birth is statistically decreased among cycles associated with endometriosis in the absence of tubal factor as compared with male factor infertility but is likely of limited clinical significance. Although statistically significant owing to the large number of cycles in this study, the clinical impact of endometriosis on chance of pregnancy and live birth is likely minimal, with the percentage of cycles leading to pregnancy or live birth differing only by 1% or less between the two groups, and point estimates and confidence intervals approaching 1. Previous studies investigating ART pregnancy outcomes among women with endometriosis have yielded conflicting results: several have found no association between endometriosis and pregnancy outcomes ( 8 , 9 , 13 , 14 ), whereas a prior systematic review and meta-analysis found an association between severe stage III/IV endometriosis and decreased implantation and clinical pregnancy rates as compared with women without endometriosis ( 6 , 12 , 18 ). This study supports prior findings suggesting limited clinically significant impact of endometriosis on pregnancy outcomes among women undergoing ART. The lack endometriosis stage within the collected variables prevents us from drawing a conclusion about the impact of severe-stage endometriosis. The study is limited by the lack of patient surgical history; however, the mean serum FSH and presence or absence of a diagnosis of diminished ovarian reserve indirectly reflect the potential deleterious impact of prior ovarian surgery on ovarian reserve. The study is also limited by the fact that data are cycle-, rather than patient-based, and that embryo quality is not a collected variable. We were also not able to account for time between endometriosis diagnosis and treatment, method of diagnosis, and treatment before initiation of IVF. Moreover, this is a retrospective cohort study that uses surveillance system data that rely on the input accuracy of individual clinics. The study is among the first of its size to investigate ART outcomes specifically among those with an endometriosis diagnosis; it also offers additional analysis of medication requirements and cancellation and ovarian hyperstimulation syndrome risks. Additionally, we were able to control for potential confounders, including age, obstetric history, ART history, concomitant infertility diagnoses, stimulation type, laboratory procedures, number and stage of embryo transfer, and time. In conclusion, when counseling infertility patients with endometriosis who are considering ART, clinicians now have additional information regarding slightly increased medication requirements and chance of cancellation and slightly decreased risk of hyperstimulation compared with those without endometriosis. Although statistically significant, the magnitude of effect detected for the decrease in implantation, live birth, and pregnancy rates was small, and confidence intervals for these measures approached the null value, suggesting that any difference in these outcomes between women who do and who do not have endometriosis is likely to be minimal.