Live-Birth Rate Associated With Repeat In Vitro Fertilization Treatment Cycles

Importance  The likelihood of achieving a live birth with repeat in vitro fertilization (IVF) is unclear, yet treatment is commonly limited to 3 or 4 embryo transfers.

Objective  To determine the live-birth rate per initiated ovarian stimulation IVF cycle and with repeated cycles.

Design, Setting, and Participants  Prospective study of 156 947 UK women who received 257 398 IVF ovarian stimulation cycles between 2003 and 2010 and were followed up until June 2012.

Exposures  In vitro fertilization, with a cycle defined as an episode of ovarian stimulation and all subsequent separate fresh and frozen embryo transfers.

Main Outcomes and Measures  Live-birth rate per IVF cycle and the cumulative live-birth rates across all cycles in all women and by age and treatment type. Optimal, prognosis-adjusted, and conservative cumulative live-birth rates were estimated, reflecting 0%, 30%, and 100%, respectively, of women who discontinued due to poor prognosis and having a live-birth rate of 0 had they continued.

Results  Among the 156 947 women, the median age at start of treatment was 35 years (interquartile range, 32-38; range, 18-55), and the median duration of infertility for all 257 398 cycles was 4 years (interquartile range, 2-6; range, <1-29). In all women, the live-birth rate for the first cycle was 29.5% (95% CI, 29.3%-29.7%). This remained above 20% up to and including the fourth cycle. The cumulative prognosis-adjusted live-birth rate across all cycles continued to increase up to the ninth cycle, with 65.3% (95% CI, 64.8%-65.8%) of women achieving a live birth by the sixth cycle. In women younger than 40 years using their own oocytes, the live-birth rate for the first cycle was 32.3% (95% CI, 32.0%-32.5%) and remained above 20% up to and including the fourth cycle. Six cycles achieved a cumulative prognosis-adjusted live-birth rate of 68.4% (95% CI, 67.8%-68.9%). For women aged 40 to 42 years, the live-birth rate for the first cycle was 12.3% (95% CI, 11.8%-12.8%), with 6 cycles achieving a cumulative prognosis-adjusted live-birth rate of 31.5% (95% CI, 29.7%-33.3%). For women older than 42 years, all rates within each cycle were less than 4%. No age differential was observed among women using donor oocytes. Rates were lower for women with untreated male partner–related infertility compared with those with any other cause, but treatment with either intracytoplasmic sperm injection or sperm donation removed this difference.

Conclusions and Relevance  Among women in the United Kingdom undergoing IVF, the cumulative prognosis-adjusted live-birth rate after 6 cycles was 65.3%, with variations by age and treatment type. These findings support the efficacy of extending the number of IVF cycles beyond 3 or 4.

In vitro fertilization (IVF) is commonly stopped after 3 or 4 unsuccessful embryo transfers,^1,2 with 3 unsuccessful transfers labeled “repeat implantation failure.”³ This practice has been influenced by a study of 1328 embryo transfers undertaken 20 years ago, without use of intracytoplasmic sperm injection (ICSI), which reported a decline in live-birth rates after the fourth cycle.⁴ With 1 exception,⁵ previous studies of cumulative pregnancy or live-birth rates have been relatively small, with limited ability to precisely estimate cumulative success beyond 4 transfers.^4,6-9 Previous studies have defined a cycle of IVF as an embryo transfer.^5-9 Thus, each initiation of IVF with ovarian stimulation has been treated as several separate cycles whenever there has been a series of repeated embryo transfers. Because both the promotion of single embryo transfer and the effectiveness of freezing embryos have increased markedly over the last 10 to 15 years,^10-15 it has been suggested that IVF success should be calculated as the live-birth rate per initiated ovarian stimulation, including all subsequent separate fresh and frozen embryo transfers.^5,10-13

Quiz Ref IDThe aim of this study was to determine the extent to which repeat IVF cycles continue to increase the likelihood of a live birth, defining an IVF cycle as the initiation of treatment with ovarian stimulation and all resulting separate fresh or frozen embryo transfers; hereafter, we use the term cycle for this. Specific objectives were to determine (1) the live-birth rate within each cycle and the cumulative rate across all cycles, (2) how these varied by age and treatment type (use of donor oocyte, ICSI, or sperm donation), and (3) the association between oocyte yield in 1 cycle and live-birth rate in subsequent cycles.

Ethical approval for this study was provided by the UK Human Fertilisation and Embryology Authority (HFEA), which has statutory obligations to prospectively collect information on all assisted reproductive treatment (ART) in the United Kingdom. Women provided written consent for this information to be used in analyses, audits, and publications. The HFEA provided us with data on all ART events occurring in the United Kingdom between January 1, 2003, and June 30, 2012, with linkage of cycles to individual women and data on birth outcomes (Figure 1). Because all UK clinics, whether private or public, must provide information on any patients treated with ART, together with the outcomes of that treatment, to the HFEA, they are able to link cycles to individual women for all UK ART. We chose the 2003 start date to obtain a large cohort representative of contemporary treatment, and June 2012 was the latest date for which the HFEA could provide validated data. Because the live-birth outcome data were incomplete for cycles commencing between January 2011 and June 2012 (as many of these cycles were still continuing and births from them could occur after June 2012), we limited our potentially eligible cohort to ovarian stimulation cycles initiated between January 1, 2003, and December 31, 2010, with live-birth outcome data collected up to June 2012.

We excluded ART that was not IVF or was undertaken for the purpose of storage, donation, or surrogacy. We excluded women who had started IVF before 2003. As in other studies,^5-9 once a live birth occurred, women were censored from further analysis. To reflect clinical practice and allow comparisons with other studies,^4,5,7,9 we included all embryo transfers, whether the individual transfer was of 1 or more embryos.

Live birth was defined as an infant born alive after 24 weeks’ gestation who survived longer than 1 month. The World Health Organization defines live birth as a birth showing any sign of life irrespective of gestational age. As in other studies,^5,15,16 we modified this definition to capture births that were likely to be viable. We defined an IVF cycle as the initiation of ovarian stimulation and all resulting separate fresh or frozen embryo transfers. Quiz Ref IDThe live-birth rate within a cycle was defined as the probability of a live birth from an ovarian stimulation encompassing all subsequent fresh and frozen embryo transfers from that stimulation. Thus, for those embarking on IVF, the live-birth rate within 1 cycle answers the question, “What is my chance of a live birth with 1 stimulation and retrieval of oocytes followed by as many subsequent separate embryo transfers as possible from that retrieval?” The cumulative live-birth rate at a given cycle was defined as the probability of a live birth from all cycles up to and including that cycle. This answers the question, “What is my total chance of a live birth with repeat ovarian stimulation and oocyte retrievals, together with the subsequent embryo transfers from each cycle, up to a given cycle number?”

Information on age, types of treatment (oocyte donation, sperm donation, and ICSI), oocyte yield, and other couple characteristics were obtained from the HFEA data set.

We calculated the live-birth rates within the first and subsequent cycles up to the ninth as the proportion of cycles resulting in a live birth, using a normal approximation to construct confidence intervals. We calculated estimates of cumulative live-birth rates up to the ninth cycle using the Kaplan-Meier method with Greenwood’s approximation to calculate confidence intervals (eMethods in the Supplement).^17,18 We calculated 3 different estimates using different assumptions for women who discontinued IVF without a live birth (see below). We used a log-rank test¹⁹ to compare the live-birth rate within each cycle and cumulatively across all cycles. The first set of comparisons was between woman’s age and oocyte source category, and the second was between no male cause of infertility and male cause of infertility with and without treatment by ICSI or sperm donation. We assessed the relationship of oocyte yield in 1 cycle to live-birth rates in subsequent cycles in women younger than 40 years using their own oocytes, by calculating the within-cycle live-birth rates in the first, second, and third cycles by oocytes retrieved in the first cycle, and also calculating the within-cycle live-birth rates up to the fifth cycle by oocytes retrieved in the immediately preceding cycle.

Infertile couples discontinue IVF for a number of reasons, with a systematic review of patient perceptions concluding that the most common reasons were the physical or psychological burden of treatment, relationship problems, and personal problems.²⁰ In any study estimating cumulative live-birth rates, assumptions must be made about what the rate in those who discontinue would have been had they continued. To account for this, we calculated “optimal” and “conservative” estimates, which have been assessed in previous studies. In addition, we calculated a prognostic-adjusted estimate. The optimal estimate is based on the observed data, and while not always explicit in previous publications, this assumes that the cumulative live-birth rate in women who discontinue IVF without a live birth if they had continued would be equal to the rate in those who continue to have further cycles.⁵ The conservative estimate assumes those who discontinue IVF would have had a subsequent live-birth rate of 0.⁵ The true rate is thought to lie between these two.⁷ The prognostic-adjusted estimate aims to obtain this more realistic value. It assumes that a fixed proportion of those who discontinue do so because of poor prognosis and that the live-birth rate in that proportion would have been 0, whereas for those who discontinue for other reasons, such as inability to pay, emotional distress, or (in our data set) emigration from the United Kingdom, it would have been similar to those who continue with treatment.

For the prognosis-adjusted estimate, we considered the woman’s age at her first cycle and oocyte yield in the previous cycle to be the strongest prognostic factors, because these have been shown to be strongly related to live-birth success.^5,7,9,21,22 We checked that these were indicators of live birth and of discontinuation of treatment in our own data and compared other available characteristics between those who discontinued and continued treatment after 1 unsuccessful cycle. To obtain age-adjusted and oocyte yield–adjusted estimates, we calculated results for each age strata (18-34, 35-37, 38-39, 40-42, 43-44, 45-50, and ≥50 years) and for each possible oocyte yield in the previous cycle and then obtained an average, weighted by the numbers within each category in the first cycle. It was not possible to calculate age-adjusted estimates for the age-stratified analyses because there was too little age variation within the age strata. For any analyses that included women using donor oocytes, it was not possible to calculate rates adjusted for oocyte yield in the previous cycle as women using donor oocytes will not have an oocyte yield.

The age and previous oocyte yield–adjusted results suggested that 3% of those who discontinued IVF did so because of poor prognosis. However, to calculate a prognostic-adjusted cumulative live-birth rate, we assumed 30% of those who discontinued did so because of poor prognosis. We chose a value of 10× that suggested by our data to obtain a conservative prognostic-adjusted estimate. (Full details of how these estimates were calculated are provided in the eMethods in the Supplement.)

Because the average population live-birth success rate for a single embryo transfer is between 20% and 30% in high-income countries,^10-13 we considered 20% to be a benchmark for a good live-birth rate within a cycle. All analyses were undertaken in Stata version 13 MP2. Two-sided P values <.05 were considered to provide evidence against the null hypothesis.

We used data on pregnancy and pregnancy loss rates from published literature to estimate live-birth rates for women who conceive naturally.^23-25 Two prospective cohort studies of couples actively trying to conceive provided age-specific pregnancy rates attained within 12 menstrual cycles.^23,24 Live-birth rates were calculated assuming 20% of natural conceptions result in a pregnancy loss.²⁵

Following planned exclusions, the eligible cohort included 257 665 cycles in 157 475 women. For all analyses, we excluded women with missing linkage information or implausible linkage (ie, first IVF transfer being a frozen embryo transfer without preceding ovarian stimulation). This resulted in an analysis cohort of 257 398 cycles by 156 947 women (more than 99% of the eligible cohort; Figure 1). Table 1 shows the characteristics of the cohort. Among the 156 947 women, the median age at start of treatment was 35 years (interquartile range, 32-38; range, 18-55), and the median duration of infertility for all 257 398 cycles was 4 years (interquartile range, 2-6; range, <1-29). eTable 1 in the Supplement shows characteristics by year of treatment. Because of the large sample size, there was statistical evidence of differences in all characteristics, but for most these were small and unlikely to be clinically important. For example, median age of the women differed by 1 year, and median oocyte retrieval differed by 1 across the study period. Use of ICSI increased by 11% and transfer of single embryos by 17%, although the live-birth rate increased by just 2% across the study period.

Table 2 shows the live-birth rate within each cycle for the whole cohort. In all women, the live-birth rate for the first cycle was 29.5% (95% CI, 29.3%-29.7%). The live-birth rate within cycles remained above 20% for each cycle up to and including the fourth. After their first cycle, there were 110 614 women (70.5% of the analysis cohort) who did not have a live birth. Of these, 37 704 (34.1%) discontinued treatment and 72 910 (65.9%) had at least 1 more cycle. eTable 2 in the Supplement compares characteristics between these 2 groups. Although there was statistical evidence of differences for all characteristics, the actual differences were small.

The cumulative live-birth rate continued to increase up to the ninth cycle, with a cumulative prognosis-adjusted live-birth rate of 65.3% (95% CI, 64.8%-65.8%) by the sixth cycle (Table 2). The equivalent optimal (78.0% [95% CI, 77.3%-78.8%]) and age-adjusted (76.7% [95% CI, 76.0%-77.5%]) estimates for 6 cycles were similar, while the conservative estimate was 46.8% (95% CI, 46.5%-47.0%) (Table 2 and eFigure 1 in the Supplement).

Results varied by age and oocyte source (Figure 2, Table 3, and eTables 3 and 4 in the Supplement). In women who were younger than 40 years and using their own oocytes (133 379 women, 85% of the cohort), the live-birth rate for the first cycle was 32.3% (95% CI, 32.0%-32.5%). This remained above 20% up to and including the fourth cycle. The previous cycle oocyte yield–adjusted and optimal estimates were similar. Six cycles achieved cumulative live-birth rates of 68.4% (95% CI, 67.8%-68.9%), 80.3% (95% CI, 79.5%-81.0%), and 50.7% (95% CI, 50.5%-51.0%), for the prognostic-adjusted, optimal, and conservative estimates, respectively. For women aged 40 to 42 years, the live-birth rate for the first cycle was 12.3% (95% CI, 11.8%-12.8%), with 6 cycles achieving cumulative live-birth rates of 31.5% (95% CI, 29.7%-33.3%), 41.5% (95% CI, 38.0%-44.9%), and 19.2% (95% CI, 18.5%-19.8%) for prognostic-adjusted, optimal, and conservative estimates, respectively. For women older than 42 years, all rates within each cycle were less than 4% or based on too few live births to calculate confidence intervals.

Use of donor oocytes removed this age differential, as the log-rank test showed no evidence for different cumulative live-birth rates between age categories (eTable 3 in the Supplement). Irrespective of age, women using donor oocytes achieved live-birth rates within each cycle of 29.6% or greater for all cycles up to and including the ninth and a cumulative live-birth rate after 6 cycles of 86.7% (95% CI, 85.2%-88.3%), 91.7% (95% CI, 90.3%-93.1%), and 75.5% (95% CI, 74.0%-77.1%) for the prognostic-adjusted, optimal, and conservative estimates, respectively (eTable 4 in the Supplement).

Live-birth rates varied by male-partner cause infertility and its treatment (Figure 3 and eTables 5 to 7 in the Supplement). Women whose infertility was due to a male-related cause and who were not treated with either ICSI or donor sperm had lower live-birth rates than those with a nonmale cause of infertility (eTables 3 and 5). Women with a male-partner cause of infertility who were treated with ICSI had cumulative live-birth rates, after 6 cycles, of 71.3% (95% CI, 70.5%-72.1%), 82.2% (95% CI, 81.1%-83.4%), and 54.7% (95% CI, 54.3%-55.2%) using the prognostic-adjusted, optimal, and conservative estimates, respectively (eTable 6 in the Supplement). Equivalent results for those with male infertility treated with donor sperm were 81.2% (95% CI, 78.6%-83.9%), 90.2% (95% CI, 87.2%-93.1%), and 65.9% (95% CI, 63.9%-67.9%), respectively (eTable 7). Live-birth rates in both of these groups were greater than for women with a nonmale cause of infertility (eTables 3 and 8 in the Supplement).

Figure 4 shows the live-birth rate within the first, second, and third cycles plotted against the number of oocytes retrieved in the first cycle in women younger than 40 years using their own oocytes. For those in whom no oocytes were retrieved in the first cycle, the live-birth rates in the second and third cycles were greater than 20%. The live-birth rates in the first, second, and third cycles continued to increase with increasing oocytes retrieved in the first cycle up to around 15 oocytes; thereafter, the curves flatten. Plotting the live-birth rate within any cycle against the number of oocytes retrieved in the previous cycle gave a similar pattern (eFigure 2 in the Supplement).

Using published data,^23-25 we estimated the live-birth rate for women conceiving naturally who had been trying for 12 menstrual cycles. This rate varied between 58% and 74%, depending on the woman’s age and frequency of intercourse (eTable 9 in the Supplement). These estimates are based on studies that only included women younger than 40 years. Similar cumulative live-birth rates were achieved by the fifth or sixth cycle of IVF treatment in women of this age (Table 3); however, in these women, 5 cycles took a median of 2 years (interquartile range, 2-3).

Quiz Ref IDTo our knowledge, this is the first study to have linked fresh and frozen embryo transfers to obtain estimates of live-birth rate within each IVF ovarian stimulation cycle and cumulative live-birth rates across repeated stimulation cycles. Despite a decline in the success rate within each cycle as the number of these increased, the cumulative rate across cycles increased up to the ninth in the whole cohort, women younger than 40 years (using their own oocytes), and women using donor oocytes (irrespective of age). They also increased up to the eighth or ninth cycle for women aged 40 to 42 years, although for women older than 42 years (using their own oocytes), the likelihood of success was low and the cumulative live-birth rate did not appear to clearly increase beyond the fourth or fifth cycle. For those women able to use donor oocytes, age was unrelated to success. Quiz Ref IDIn those for whom the cause of infertility was related to a male partner problem, treatment with ICSI or donor sperm made a marked difference in the likelihood of success, with cumulative rates increasing up to the eighth or ninth cycle, whereas without treatment rates were lower than for women with other causes of infertility. For women younger than 40 years with a low oocyte yield in a previous cycle, there was benefit in continuing with further cycles. We also found that women younger than 40 years could achieve cumulative live-birth rates after 5 or 6 cycles that were similar to published live-birth rates achieved naturally within 12 menstrual cycles.^23-25 It should be noted, however, that, in these women, 5 cycles took a median of 2 years.

Quiz Ref IDWidespread adoption of single embryo transfer has reduced multiple pregnancies and adverse perinatal outcomes but has meant that the chance of a live birth from a single ovarian stimulation cycle is spread across multiple embryo transfers, which we have assessed here. Because this method of assessing IVF success combines all embryo transfer events after an ovulation stimulation into 1 analysis unit, we were unable to examine the effect of the number of embryos transferred per event. However, this method of assessing IVF success is increasingly recommended.^5,10-13 Our results show how success rates per embryo transfer event are misleadingly lower, compared with the rate within each ovarian stimulation cycle. We have previously shown, using unlinked data from the same population, that the number of embryos transferred in an embryo transfer event is associated with live-birth rate, with a difference of 9% in women younger than 40 years and 16% in those 40 years or older, comparing double with single embryo transfer.¹⁵

Despite the differences in the definition of cumulative success between our study and the previous largest study (from the United States), in which cumulative live-birth rates were estimated on the basis of each embryo transfer,⁵ and differences in health systems between the United States and United Kingdom, both studies found age differences in rates and that the differences were removed with the use of donor oocytes. In the US study, women with a male partner cause of infertility had one of the highest cumulative live-birth rates per embryo transfer, but that study did not examine the effect of different treatments (ICSI or sperm donation), and it may be that all of those patients with male partner–cause infertility in the United States receive one of these treatments.

The key limitation of all studies looking at cumulative outcomes with repeat IVF is how patients who discontinue treatment are treated. As seen in our data and in previous studies,^5,7 the extremes of the optimal and conservative estimates often vary markedly; for example, in our data, the optimal and conservative estimates were 78.0% and 46.8%, respectively, for the whole cohort. This is because of the differences between these 2 estimates in assumptions made regarding what would have been the live-birth rate in those who discontinued IVF had they continued; for the optimal estimate, this is assumed to be the same as those who did continue, whereas for the conservative estimate it is assumed to be 0. We examined the likelihood that such discontinuation was due to poor prognosis based on age and previous cycle oocyte retrieval. These analyses suggested approximately 3% of those who discontinued did so because of poor prognosis. This small proportion was because although these 2 factors are important predictors of live birth, few women receiving IVF are older than 40 years (only 15% in our national population cohort), and most women have a high oocyte yield (median, 9 per cycle in our cohort). However, to account for other factors, such as pretreatment reproductive hormone levels, smoking, and body mass index, which have been linked to live-birth rates^7,22 but were not available in this study, we assumed a 30% discontinuation due to poor prognosis. Because of the legal requirement for all UK clinicians to provide data on all ART patients, the HFEA was able to link cycles to individual women even if they moved between clinics within the United Kingdom. However, treatment abroad would be absent from our data. A European study conducted 6 years ago found very few UK couples traveled for ART to 49 clinics in 6 (non-UK) European countries with high rates of cross-border patients.²⁶ We were only able to assess live birth as an outcome: future studies should also consider potential adverse effects of continued treatment, including ovarian hyperstimulation syndrome and possible increased risk of preterm birth, low birth weight, or congenital anomalies.^16,27,28

For some couples, the emotional stress of repeat treatments may be undesirable, and the cost of a prolonged treatment course, with several repeat oocyte stimulation cycles, may be unsustainable for health services, insurers, or couples. However, we think the potential for success with further cycles should be discussed with couples. A cost-effectiveness analysis is beyond the scope of this study, and the difficulties of undertaking such analyses for IVF, in which decisions related to how one values a new life and whether “benefits” and “costs” for both parents and the child should be included, are well documented.²⁹ The costs of IVF treatment vary between countries, whether publicly or privately funded, and the treatment type used but are approximately $14 000 (equivalent of approximately £9000, €12 000) to $17 000 (equivalent of approximately £11 000, €15 000) per cycle.^1,29,30 These costs exclude assessment prior to starting IVF and are based on transfer of 1 fresh embryo. Assuming each additional frozen embryo transfer costs $4000 to $5000,³⁰ the cost per couple of continuing to 6 cycles, rather than having just 3, could be as much as $132 000 compared with $66 000 (assuming 1 fresh and 1 frozen transfer per cycle).

Among women in the United Kingdom undergoing IVF, the cumulative prognosis-adjusted live-birth rate after 6 cycles was 65.3%, with variations by age and treatment type. These findings support the efficacy of extending the number of IVF cycles beyond 3 or 4.

Corresponding Author: Debbie A. Lawlor, PhD, MRC Integrative Epidemiology Unit at the University of Bristol, Oakfield House, Oakfield Grove, Clifton, Bristol BS8 1BN, United Kingdom (d.a.lawlor@bristol.ac.uk).

Author Contributions: Drs Smith and Lawlor had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Nelson and Lawlor contributed equally to this work.

Study concept and design: Smith, Nelson, Lawlor.

Acquisition, analysis, or interpretation of data: Smith, Tilling, Nelson, Lawlor.

Drafting of the manuscript: Smith, Lawlor.

Critical revision of the manuscript for important intellectual content: Tilling, Nelson.

Statistical analysis: Smith.

Obtained funding: Nelson, Lawlor.

Study supervision: Tilling, Nelson, Lawlor.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Nelson reported having participated in advisory boards for and received speakers’ fees from Beckman Coulter, Besins, Ferring, Merck Serono, Merck Sharp & Dohme, and Roche Diagnostic. No other disclosures were reported.

Funding/Support: Drs Smith, Tilling, and Lawlor work in a unit that receives funding from the UK Medical Research Council (http://www.mrc.ac.uk) (MC_UU_12013/5 and MC_UU_12013/9), and Dr Lawlor receives support from a National Institute for Health Research Senior Investigator award (NF-SI-0611-10196). This work is also supported by a grant from the Wellcome Trust (http://www.wellcome.ac.uk), which pays Dr Smith’s’ salary (WT094311MA).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: All views are those of the authors and not necessarily those of the Human Fertilisation and Embryology Authority (HFEA).

Additional Contributions: We thank the staff of the HFEA and the contributing UK clinics for these data.