Assisted reproductive technology with donor sperm: national trends and perinatal outcomes.

ART donor sperm use has increased since 1996 and accounted for 6.2% of all banking and fresh ART cycles in 2014. Average maternal age was higher in cycles using donor vs nondonor sperm. Use of donor sperm was associated with an increased rate of pregnancy and live birth compared with nondonor sperm cycles when calculated per retrieval, but no significant differences were noted when calculated per transfer. There were no significant differences in rates of preterm births when comparing donor and nondonor sperm among singleton births. Among singletons, a decrease in low birthweight risk was noted in donor sperm cycles; however, these statistical findings are likely of limited clinical significance. Studies have shown a general trend in the United States toward increased maternal age at time of first birth. 29 From 2000 to 2014, the proportion of first births to women aged 30–34 years rose 28% and first births to women aged 35 years and older rose 23% 33 In the present study, we found that between 2010 and 2014, 73.4% of ART cycles using donor sperm were conducted among women 35 years old and older as compared with 55.2% of cycles using nondonor sperm. Similarly, increased use of third-party ART (including donor sperm) with advancing maternal age has been reported. 3 With increased social acceptance of the use of donor gametes, single women may be more inclined to use donor sperm. 34 The primary indications for use of donor sperm are male factor infertilityor lack of a male partner. In NASS, male factor infertility is defined as infertility caused by abnormal semen parameters or abnormal sperm function. 35 Because cycles among patients with no male partner do not satisfy a defined reason for using ART in NASS, we would expect these to be classified as other. Data from the United Kingdom demonstrates a significantly increasing trend in female couples pursuing treatment with donor sperm (4% to 11.5% from 2000 to 2010) and a similar rise in single women (2% to 8%). 31 The Human Fertilisation and Embryology Authority reports that almost two thirds of patients accessing fertility treatments registered with a male partner, 15% registered with no partner, and 21% registered with a female partner. 5 In their 2014 report, they note a 20% increase in both donor inseminations and donor sperm IVF in female couples. 6 Other studies note similar trends. 36 , 37 As a result, we anticipated a higher percentage of cycles with a male factor diagnosis as well as a higher percentage of cycles classified as other reason for ART among cycles using donor sperm than among nondonor sperm cycles. While we did see a higher percentage of other reason for donor sperm cycles, the percentage of cycles using donor sperm that received a male factor diagnosis was lower (33.7%) than those using nondonor sperm (36.8%). The unexpected findings for cycles with male factor infertility may be due to limitations in the definition; severity of male factor is not quantified and the reasoning behind use of donor sperm for male factor is often unknown. Among cycles with male factor, only 4.7% used donor sperm, whereas 5.4% of cycles without male factor used donor sperm. This may reflect a lack of clarity in diagnosis categorization once donor sperm is used; one possibility is that clinics may perceive the cycle to no longer be characterized by a male-related factor because the donor sperm source does not have infertility. Because donor sperm use in ART cycles increases, national reporting can be improved by clarifying the male factor infertility diagnosis, especially in relation to use of donor sperm, to ensure proper cycle and outcome monitoring. Following male factor, diminished ovarian reserve was the next most common diagnosis. Decreasing prevalence of male factor infertility with increased maternal age, particularly in women older than age 40 years, who were more likely to have the diagnosis of diminished ovarian reserve than male factor when undergoing IVF cycles with donor sperm has been reported. 3 We found that ICSI use between 2010 and 2014 was higher than expected for donor sperm cycles at 71.4%, only slightly lower than ICSI use in nondonor sperm cycles (76.3%). 32 , 38 This may reflect the application of ICSI for reasons other than male factor infertility (eg, unexplained infertility, poor-quality oocytes, low oocyte yield, or prior failed conventional fertilization), which may reflect a decision made based on semen parameters on the day of fertilization. 39 Whereas some studies have found when postprocessing donor spermatozoa motility is low, fertilization by ICSI is associated with significantly higher pregnancy rates compared with those of conventional fertilization. 16 The increase in donor sperm ART may be partially explained by increased ART use among single women and female couples as well as increased social acceptance of donor gametes; however, the reason is likely multifactorial. After adjustment, use of donor sperm was associated with an increased rate of intrauterine pregnancy and live birth per retrieval but not per transfer; although statistically significant, many estimates and confidence intervals were close to 1.00 and are likely of limited clinical significance. Maternal age was the confounding variable driving this shift in association. Other studies have found no difference in clinical pregnancy or LBRs when comparing IVF with autologous oocytes and donor vs nondonor sperm. 3 , 40 Similar LBRs per pregnancy (76.2% as compared with 78.4% in our study) were found in a French cohort study of 3381 pregnancies following IVF with donor spermbetween1989and1994. 9 Inastudy of patients with male factor infertility, higher LBRs were found in IVF/ICSI donor sperm cycles compared with nondonor sperm cycles; however, the differences were not statistically significant. 41 Pregnancy and obstetric outcome data in our study were also reassuring, with no difference in preterm birth or miscarriage rates seen when comparing outcomes for donor vs nondonor sperm. Findings from other studies suggest no difference in outcomes such as spontaneous abortion and stillborn infants 3 as well as low birthweight, preterm delivery, or birth defects following IVF cycles and intrauterine inseminations using donor sperm vs nondonor sperm. 9 , 42 In the present study, a subanalysis controlling for maternal race/ethnicity and BMI revealed similar pregnancy outcomes between women using donor and nondonor sperm. To our knowledge, our study is the largest of recent trends in donor sperm use among US ART cycles and resulting treatment and perinatal outcomes. The study is strengthened by the large sample size and the 98% compliance of fertility clinics reporting ART data. Furthermore, because the majority of previously published donor sperm outcome data has come from insemination studies, a strength of our study is that we specifically describe ART outcomes, namely pregnancy rates, LBRs, and perinatal outcomes including preterm and low birthweight deliveries. Our study was limited by the accuracy of input from individual clinics, the amount of missing data for some covariates, and the absence of certain variables such as the fertilization rate or state of the sperm (fresh or frozen) from NASS. While all anonymous donor sperm was likely frozen in accordance with current practice patterns, this field was added to NASS and will be available starting with reporting year 2016. Additionally, because data collection is cycle based, women who underwent more than 1 ART cycle with oocyte retrieval were included more than once in the outcome data. While generally a strength, the large sample size can also serve as a limitation in that it powers a study to detect small, statistically significant but potentially clinically unimportant differences. Furthermore, the indication for use of donor sperm was not available for analysis. This highlights an area for potential improvement in ART reporting strategies; particularly, if the use of donor gametes continues to rise. These results must be interpreted with caution because they involve only ART cycles and exclude those women who conceived with donor sperm insemination and may have a better prognosis. In conclusion, use of donor sperm in ART cycles and the number of resultant births has increased over time. Reassuringly, we found that perinatal outcomes were clinically similar between ART cycles that use donor and nondonor sperm.

In total, 74,892 fresh cycles (with oocyte retrieval) using donor sperm were performed between 1996 and 2014; 54,900 of these cycles used autologous oocytes and 13,542 used donor oocytes. The number of fresh cycles using donor sperm with autologous oocytes increased over time, peaking in 2011 with 5161 cycles ( Figure 1A ). The number of fresh cycles using donor oocytes followed a similar pattern except for a decline in recent years, peaking in 2012 with 1195 cycles. Although there was a slight initial decline from 3.8% to 2.6% between 1996 and 1999, the percentage of the banking and fresh cycles using donor sperm has since continued to increase over time, accounting for 4.9–6.2% of all ART cycles between 2010 and 2014 (n = 7351) ( Figure 1B ). The total LBR across all years was 28.5% and 31.9% in fresh autologous cycles using donor and nondonor sperm, respectively, with a range of 24.5–30.6% for donor sperm and 26.2–34.0% for nondonor sperm cycles ( Figure 2A ). For both groups, the LBR increased and then decreased over time. The number of infants born from fresh autologous cycles using donor sperm increased over time before falling to 1570 in 2014, while the number of infants born from fresh autologous cycles using nondonor sperm increased from 15,452 in 1996 to a peak of 40,386 in 2008 before falling to 29,615 in 2014, possibly reflecting a trend toward transfer of fewer embryos ( Figure 2 , B and C ). The average female age of patients undergoing fresh autologous cycles using donor and nondonor sperm both increased over time with a decrease in recent years but with cycles using donor sperm exhibiting a higher average female age overall ( Figure 3 ). The female age distribution at cycle start differed by sperm source, with a higher percentage of females 38–40 years (24.6% vs 19.3%) and ≥41 years old (30.0% vs 14.6%) among donor sperm cycles compared with nondonor sperm cycles ( Table 2 ). The distribution of female BMI among donor sperm cycles showed a higher percentage of women with BMI ≥25 kg.m 2 (22.3%) and BMI ≥30 kg.m 2 (20.4%), compared with nondonor sperm cycles (19.3% and 15.1%). The distributions of prior pregnancies, spontaneous abortions, and births showed these outcomes to occur in lower percentages of cycles using donor vs nondonor sperm. Approximately 27.6% of donor sperm cycles were conducted among patients having undergone ≥2 prior ART cycles compared with 23.0% among nondonor sperm cycles. The most common infertility diagnosis for both groups was male factor infertility, accounting for 33.7% and 36.8% of donor and nondonor sperm cycles, respectively. The second most common diagnosis was diminished ovarian reserve. A diagnosis of other accounted for 27.2% and 12.2% of donor and nondonor sperm cycles, respectively. Gonadotropin-releasing hormone antagonist stimulation protocols were most common in both donor and nondonor sperm cycles. Differences in the number of oocytes retrieved were noted, with greater than 10 oocytes retrieved in 50.8% and 57.0% of donor and nondonor sperm cycles, respectively. Regarding mode of fertilization, intracytoplasmic sperm injection (ICSI) was performed in 71.4% and 76.3% of cycles in donor and nondonor sperm groups. Cycles using donor sperm were more likely to result in fewer embryos cryopreserved. Transfers using donor sperm were more likely to involve 3 or more embryos and occur on day 2 or 3 compared with nondonor sperm cycles. All associations in Table 2 were statistically significant ( P < .001). The percentage of noncancelled cycles resulting in a live birth was higher for nondonor sperm cycles compared with donor sperm cycles (32.5% vs 27.9%, P < .0001) ( Table 1 ), as was the percentage of noncancelled cycles resulting in a singleton live birth (23.5% vs 20.5%, P < .0001) or a multiple live birth (9.0% vs 7.5%, P < .0001). However, among those cycles resulting in a live birth, the percentage of multiple live births was not significantly different for nondonor sperm cycles compared with donor sperm cycles (27.6% vs 26.8%, P = .2049; results not shown). Intrauterine pregnancy rates per cycle were 35.7% with donor sperm and 39.8% with nondonor sperm, with an adjusted RR of 1.05 (95% CI, 1.01–1.10). LBRs per cycle were 27.9% and 32.5% with donor and nondonor sperm (aRR, 1.06, 95% CI, 1.01–1.10); adjustment for maternal age explained most of the difference between the unadjusted and adjusted RRs for this association. The percentage of transfers resulting in intrauterine pregnancy and live birth were similarly lower for donor sperm cycles but not significantly different after adjustment. Per pregnancy, LBRs were 78.4% and 81.7% (aRR, 1.00, 95% CI, 0.99–1.01) and full-term LBRs were 59.8% and 61.3% (aRR, 1.01, 95% CI, 0.99–1.03) in donor and nondonor sperm cycles. Per singleton live birth, preterm LBRs were 11.5% and 11.8% (aRR, 0.98, 95% CI, 0.90–1.06) with low birthweight LBRs of 8.8% and 9.4% (aRR, 0.91, 95% CI, 0.83–0.99) in donor and nondonor sperm cycles. Results from a subanalysis controlling for maternal race/ethnicity and BMI were similar (results not shown).

Data were obtained from the Centers for Disease Control and Prevention’s National ART Surveillance System (NASS), a reporting system that captures data from approximately 98% of all ART cycles performed in the United States. 32 Our analysis included all ART procedures: IVF (>99% of all ART procedures), gamete intrafallopian transfer, and zygote intrafallopian transfer. 32 We reported trends in the absolute number and percentages of all donor and autologous oocyte, banking and fresh ART cycles using donor (n = 74,892) and nondonor (n = 1,635,142) sperm and trends in the number of live births, live birth rate (LBR), and average maternal age for fresh, autologous oocyte cycles using donor (n = 54,900) and nondonor sperm (n = 1,398,732) initiated between 1996 and 2014. Banking cycles were defined as cycles started with the intent to cryopreserve (freeze) all resulting oocytes or embryos (ie, no intent to transfer). We defined fresh ART cycles as cycles started with the intent to retrieve and fertilize oocytes and then transfer resulting embryos all within the same cycle. We excluded frozen oocyte cycles, frozen embryo cycles, and cycles canceled prior to retrieval because a sperm source was not reported for these cycle types as well as gestational carrier cycles. Cycles in which the semen source was mixed (ie, nondonor and donor, n = 5224) or unknown (n = 19,166) were also excluded. Patient demographics, infertility diagnoses, and ART cycle characteristics among all fresh autologous oocyte cycles (in which oocyte retrieval was performed) were compared between donor (n = 22,619) and nondonor sperm (n = 414,950) cycles from 2010 through 2014, using χ 2 tests. This part of the analysis was restricted to the most recent 5 years of data available to reflect current practice and improve clinical relevance. We compared ART treatment and perinatal outcomes of these same cycles, including rates of intrauterine pregnancy and live birth per cycle; intrauterine pregnancy and live birth per transfer; live birth and full-term live birth (≥37 weeks) per pregnancy; and preterm birth (<37 weeks) and low birthweight (<2500g) per singleton live birth. Clinical intrauterine pregnancy was defined as ultrasound confirmation of at least 1 gestational sac within the uterus, with or without fetal cardiac activity or fetal pole, or in the absence of ultrasound data, a documented birth, spontaneous miscarriage, or induced abortion. 4 A live birth was defined as the birth of at least 1 live-born infant at least 20 weeks’ gestational age. Polynomial trends in donor sperm use and outcomes were explored using linear regression with linear, quadratic, and cubic terms. We included in the fitted regression model the highest order term for time that was statistically significant as well as all lower-order terms, in which time was calculated as x = year – 1996. We reported the fitted regression equation and R 2 value (overlaid on figures). The outcomes were analyzed using log binomial regression with generalized estimating equations to adjust for clustering by clinic. For each outcome, we reported numbers, percentages, unadjusted and adjusted relative risks (RR), and 95% confidence intervals (CIs). The covariates included for statistical adjustment are listed in Table 1 . We considered interactions between use of donor sperm and maternal age and male factor infertility, but neither was statistically significant. In the event of a shift in the direction of a statistically significant effect before and after statistical adjustment (eg, RR 1), we explored the source of confounding by removing each covariate from the adjusted model one at a time and assessing the change in adjusted RR. Maternal race, ethnicity, and body mass index (BMI) were not included in the multivariable models because of a large percentage of missing values (>35% and >17%, respectively); however, a subanalysis was performed controlling for these variables. Statistical significance was assessed using α = 0.05. Statistical analysis was conducted using SAS version 9.3 (SAS Institute, Cary, NC). This study was approved by the Institutional Review Board of the Centers for Disease Control and Prevention.