*Correspondence address.

Transcription

1 Human Reproduction, Vol.31, No.1 pp , 2016 Advanced Access publication on November 3, 2015 doi: /humrep/dev270 ORIGINAL ARTICLE Infertility How do cumulative live birth rates and cumulative multiple live birth rates over complete courses of assisted reproductive technology treatment per woman compare among registries? D. De Neubourg 1, *, K.Bogaerts 2, C.Blockeel 3, T. Coetsier 4, A.Delvigne 5, F. Devreker 6,M.Dubois 7,N.Gillain 8, S. Gordts 9, and C. Wyns 10 1 Leuven University Fertility Centre, University Hospitals Leuven, 3000 Leuven, Belgium 2 I-BioStat, KU Leuven, 3000 Leuven and Universiteit Hasselt, 3500 Hasselt, Belgium 3 Centre for Reproductive Medicine, Universitair Ziekenhuis Brussel, 1090 Brussels, Belgium 4 Fertility Centre, AZ St Lucas, 9000 Ghent, Belgium 5 Centre de Procréation Médicalement Assistée, Clinique Saint-Vincent, 4000 Liège, Belgium 6 Laboratoire de Procréation Médicalement Assistée, Université Libre de Bruxelles, Hôpital Erasme, 1070 Brussels, Belgium 7 Centre de Procréation Médicalement Assistée, Université de Liège, 4000 Liège, Belgium 8 Nutrition, Environment and Health, Universityof Liège, 4000 Liège, Belgium 9 Leuven Institute for Fertility and Embryology, 3000 Leuven, Belgium 10 Department of Gynaecology-Andrology, Cliniques Universitaires Saint-Luc, Université Catholique de Louvain, 1200 Brussels, Belgium *Correspondence address. diane.deneubourg@telenet.be Submitted on June 12, 2015; resubmitted on August 24, 2015; accepted on September 29, 2015 study question: How do the national cumulative (multiple) live birth rates over complete assisted reproduction technology (ART) courses of treatment per woman in Belgium compare to those in other registries? summary answer: Cumulative live birth rates (CLBRs) remain high with a low cumulative multiple live birth rate when compared with other registries and publications. what is known already: In ART, a reduction in the multiple live birth rate could be achieved by reducing the number of embryos transferred. It has been shown that by doing so, live birth rates per cycle were maintained, particularly when the augmentation effect of attached frozen-thawed cycles was considered. study design, size, duration: A retrospective cohort study included all patients with a Belgian national insurance number who were registered in the national ART registry (Belrap) and who started a first fresh ART cycle between 1 July 2009 until 31 December 2011 with follow up until 31 December We analysed patients and cycles (both fresh and attached frozen cycles). participants, materials, settings, methods: CLBRs per patient who started a first ART cycle including fresh and consecutive frozen cycles leading to a live birth. Conservative estimates of cumulative live birth assumed that patients who did not return for treatment had no chance of achieving an ART-related live birth, whereas optimal estimates assumed that women discontinuing treatment would have the same chance of achieving a live birth as those continuing treatment. A maximum of six fresh ART cycles with corresponding frozen cycles was investigated and compared with other registries and publications. main results and role of chance: The CLBR was age dependent and declined from 62.9% for women,35 years, to 51.4% for women years, to 34.1% for women years and 17.7% for women years in the conservative analysis after six cycles. In the optimal estimate, the CLBR declined from 85.9% for women,35 years, to 72.0% for women years, to 50.4% for women years and 36.4% for women years. The cumulative multiple live birth rates for the whole population were 5.1 and 8.6% for the conservative and optimal estimate, respectively. limitations, reasons for caution: Conservative and optimal estimates use assumptions for the whole ART population and do not take the individual patient into account. & The Author Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. For Permissions, please journals.permissions@oup.com

2 94 De Neubourg et al. wider implications of the findings: These data reinforce the validity of the Belgian model of coupling reimbursement of ART costs to a restriction in the number of embryos transferred. Our data can improve decision-making in medical ART practice both on the patient level and for society at large and could provide health care takers and insurance companies with a valid model. study funding: competing interest(s): none. Key words: cumulative live birth rate / cumulative multiple live birth rate / assisted reproduction technology / conservative estimate / optimal estimate / registry Introduction From the patients perspective, a realistic presentation of the success of assisted reproductive technology (ART) treatment is mandatory (Adashi and Wyden, 2011). In mostcases, the reporting of success in ART is done by calculating outcome parameters such as pregnancy, delivery, (live) birth per started cycle, per (oocyte) aspiration or per (embryo) transfer. The reason for using these outcome measurements is obvious as these outcome parameters are easy to calculate. However, to fully inform the patients on the chances of success of an ART treatment, calculations need to be done on the patient s level rather than on the cycle level, because the former mode of reporting takes into account patients who discontinue treatment (Barnhart, 2014). Indeed, considering the potential psychological burden of the ART treatment, communication of success rates with the patient should be based on fair and realistic figures (Barnhart, 2014). A longitudinal reporting is thus preferred over a cross-sectional presentation of the outcome data. Presenting clinical outcome of ART in a patient-anchored approach is much more difficult from a logistic point of view as all cycles both fresh cycles and frozen-thawed embryo transfer cycles need to be attached to the patient entering the ART program. Cumulative birth rates after ART have been studied by Malizia et al. (2009) and Luke et al. (2012), who dealt with discontinuation of patients during treatment by using optimal and conservative estimates of cumulative live births. Conservative estimates assumed that women who did not return for treatment would not have a live birth, whereas optimal estimates assumed that these women would have live birth rates similar to those for women continuing treatment. The use of conservative estimates has been recommended by Daya (2005) in order to prevent overestimation of success rates and this was also applied in the study by Peeraer et al. (2014), where a further fine-tuning of estimates was done and a realistic estimate was proposed by taking embryo quality into account. It could have been anticipated that realistic estimates would lie between optimistic and conservative estimates, but it was shown that they tended more to the conservative estimate for all age categories. In Belgium, legislation was introduced in 2003 to reduce the number of embryos allowed for transfer, coupled to the reimbursement of the majority of the costs of six ART cycles in women up to the age of 43. Patients,36 years are allowed one embryo for transfer in the first cycle and one or two embryos in the second cycle, depending on the embryo quality. From the third until sixth cycle, two embryos are allowed. Patients 36 and,40 years can have two embryos for transfer in the first and second cycle and three as of the third cycle. Patients 40 and,43 years do not have a restraint on the number of embryos for transfer. In frozen-thawed cycles, two embryos are allowed for all ages and all cycles (De Neubourg et al., 2013). This has led to an increase in the number of single embryo transfer cycles from 15% in 2002 to 50% in Since then, the multiple delivery rate per cycle has been reduced by more than 50% from 27% in 2002 to 11% in 2010 (De Neubourg et al., 2013) and 10.5% in 2012 ( To evaluate the impact of the restriction of the number of embryos for transfer on cumulative delivery rates, a retrospective cohort study was performed in one fertility centre with a study group of patients undergoing ART after implementation of the new ART legislation and a control group of patients who received ART treatment before implementation of legislation. The study showed that there was no negative impact on the cumulative delivery rate per patient based on realistic estimates within six fresh cycles or 36 months after the start of ART treatment (Peeraer et al., 2014). ART treatment cycles are registered in Belgium online since 2001 and registration is compulsory for the centre in order to receive refunding of the majority of the costs of ART. Since 2009, a unique patient identification number was included in the registry that allows the coupling of all ART cycles performed and registered in Belgium on the patient level (De Neubourg et al., 2013). We wanted to investigate the cumulative live birth rates (CLBRs) and cumulative multiple live birth rates for the patients that receive up to six cycles with reimbursement of the majority of the treatment-related costs and compare these results to other CLBRs generated in other national registries. Materials and Methods Study population All patients with a Belgian national insurance number that was registered in the national ART registry (Belrap) who started a first fresh ART cycle between 1 July 2009 until 31 December 2011 with follow up until 31 December 2012 were included in the analysis. Registration of all ART cycles by the centres is mandatory in Belgium (De Neubourg et al., 2013) and consent for transmitting non-identifying data is signed by the patients in the respective fertility centres. After quality check of the data and resolving inconsistencies, data are anonymized prior to further analysis. On the basis of this registration, an annual report is made and supervised by the College of Physicians of ART and reimbursement of laboratory costs (for cycles complying with the legislation) to the fertility centres is calculated. This type of registration allows all subsequent cycles from a patient to be recorded even when she changes from one centre to another. Only autologous cycles were studied. For the division in age categories, the patient was allocated to the age category at the time of her first cycle. Following regulation, a consecutive use of cryopreserved embryos is obligatory before new embryos may be created (De Neubourg et al., 2013). We performed two sets of cycle counts. The first approach to measure the CLBR was to include six fresh cycles with each frozen-thawed embryo transfer cycle attached to the fresh cycle of origin. This calculation reflects the reimbursement system and embryo transfer policy in Belgium and calculates cumulative rates over complete courses of treatment per woman, i.e. all fresh cycles and the frozen cycles related to them. We analyzed up to six fresh cycles for a maximum duration of 24 months since the start of the

3 Cumulative live birth and multiple birth rates 95 first oocyte retrieval. The second way of measuring CLBR was to count fresh and frozen-thawed cycles consecutively as was done by the Society for Assisted Reproductive Technology (SART) and Australian New Zealand registry (the National Perinatal Epidemiology and Statistics Unit (NPESU) and up to seven cycles were analyzed. The latter cycle count was done to allow comparison with data from other registries. Patients were excluded from the analysis if they initiated their first cycles after 31 December 2011 or had no Belgian insurance number, if the women had a prior ART cycle (both fresh and frozen-thawed embryo transfer cycle) or underwent preimplantation genetic diagnosis (PGD) cycles, cycles after the first live birth after a previous ART cycle, cycles with donor oocytes or fresh cycles with a higher rank than six in the first type of cycle count. Outcome Live birth was defined as the live birth of a baby 500 g or 22 weeks of gestation if birthweight was unknown. Deliveries of multiple pregnancies are counted as one live birth. Cycles for which the live birth status was unknown due to the fact that the patient was lost-to-follow up were imputed as follows: cycles in which fetuses were still observed on a week ultrasound scan were assumed to have led to a live birth, other cycles were assumed not to have led to a live birth. Statistical methods Conservative estimates of the CLBR assumed that women who did not return for treatment would not have a live birth and thus potentially underestimates CLBR, whereas optimal estimates of CLBR assumed that these women would have live birth rates similar to those for women continuing treatment (Malizia et al., 2009; Luke et al., 2012). The conservative estimate of CLBR was calculated as the number of live births up to and including a specific cycle, divided by the number of women who started their first ART cycle during the study period. The optimal estimate of CLBR was based on the Kaplan Meier estimate when all cycles were included in the analysis. The standard errors (SE) for both these live birth rates were computed with the use of the binomial distribution. Differences in CLBR between age groups were assessed with the use of the log-rank test. Comparison of CLBR with other registries was done by means of a Z-test. The conditional live birth rate is the cycle-specific live birth rate and is calculated per course of treatment (first approach) or per embryo transfer cycle (second approach). The conditional live birth rate at a specific cycle was the probability of a live birth at that cycle and was equal to the number of live births (per course of treatment or per embryo transfer cycle) divided by the number of women who received ART treatment at that cycle. Withdrawal rates were calculated as the number of patients who were not pregnant and did not proceed to a next cycle divided by the number of women who were not pregnant. All analyses are performed with SAS software, version 9.3. No ethics approval was required for this research question. Results Characteristics of the study population The final data set for analysis included patients and cycles (both fresh and frozen cycles) (Table I). Causes of infertility were female factor only (n ¼ 2346, 18.2%), male factor only (n ¼ 5069, 39.4%), mixed (n ¼ 3623, 28.2%) or unexplained infertility (n ¼ 1831, 14.25%). Female causes of infertility included among others ovulation disorders (n ¼ 2547, 21%), tubal factor (n ¼ 2376, 20%), endometriosis (n ¼ 1585, 14%) and other (n ¼ 219, 3%). Age distribution was as follows: there were 8413 patients,35 years of age; 2082 aged between 35 and,38; 1568 aged between 38 and,41 and 806 aged between 41 and,43. ICSI was performed in 65.0% of cycles (n ¼ ), IVF in 26.5% (n ¼ 7122) and both IVF and ICSI in 8.5% of cycles (n ¼ 2294). Embryo transfer in fresh cycles consisted in 85.6% (n ¼ ) of cases of cleavage stage (D2 3) embryos and in 14.4% (n ¼ 3497) of blastocysts. Mean number of fresh cycles per patient until live birth was (Standard Deviation). Among all patients observed during the treatment period 92.2% of patients had ART treatment in the same fertility centre, whereas 7.6% consulted two centres and 0.2% three centres. Selective embryo reduction was performed in eight cycles (¼0.03% of the fresh cycles). Before the censoring of cycles after live birth, pregnancy outcome per cycle was missing in 1.8% of the cycles. Eighty percent of those cycles did also not have data on pregnancy evolution at weeks of gestation. They were presumed not to have had a live birth in the calculation of live birth estimates. The remaining 20% of these cycles (n ¼ 142) were imputed as having had a live birth. Live birth rates Table II gives an overview of the calculation of the conditional live birth rate, the conservative and optimal estimate of CLBR for the whole cohort up to six fresh cycles with frozen-thawed embryo transfer cycles attached to the cycle they originated from. The conservative estimate of CLBR after three and six cycles is (SE) and (SE), respectively, whereas this is (SE) and (SE) for the optimistic estimate of CLBR. The conditional live birth rate declines from 29.6 to 12.5% over six cycles and withdrawal increases from 23.7% after the first cycle to 46.9% after five cycles. Table I Cycles available for study before and after exclusions. # # Women Cycles... Study periodfrommid , female age ,43, non-cancelled fresh or frozen-thawed embryo transfer cycles with own oocytes National insurance number known Exclusions Start of first cycle after First cycle in study period was not the patients first fresh ART cycle First cycle was frozen-thawed embryo transfer cycle PGD cycles More than six fresh ART cycles 2247 Cycles after live birth (from fresh or frozen-thawed embryo transfer) Final N (fresh + frozen-thawed embryo transfer) Final N fresh cycles Final N frozen-thawed embryo transfer cycles

4 96 De Neubourg et al. Table II Cumulative birth rate: Results include fresh cycles and all frozen-thawed embryo transfer cycles resulting from the same oocyte recovery. Fresh (oocyte recovery) cycle Number of women Number of live births Conditional LBR (%) Conservative CLBR (%) SE conservative CLBR (%) Optimal CLBR (%) SE optimal CLBR (%) Withdrawal (%) LBR, live birth rate; CLBR, cumulative live birth rate; SE, standard error. Conditional LBR ¼ cycle-specific live birth rate calculated for all pregnancies resulting from a given oocyte recovery. Figure 1 shows the conservative and optimal estimates of CLBRs per age category. CLBRs rates decrease with increasing age for both conservative and optimal estimates. The conservative estimate of CLBR after six fresh and attached frozen-thawed embryo transfer cycles is 62.2% (SE) for women,35 years, 51.0% (SE) for women aged 35,38 years, 33.5% (SE) for women 38,41 and 17.7% (SE) for women years of age. The optimal estimate of CLBR after six fresh and attached frozen-thawed embryo transfer cycles is 85.6% (SE) for women,35 years, 71.7% (SE) for women aged 35,38 years, 49.5% (SE) for women 38,41 and 36.4% (SE) for women years of age. Multiple live birth rates CLBRs are 5.1% (SE) for the conservative estimate and 8.6% (SE) for the optimal estimate after six fresh and attached frozen-thawed embryo transfer cycles for the whole cohort. Comparison with other registries Table III shows the conditional live birth rate and the conservative and optimal estimates of CLBR for the whole cohort but with fresh and frozen-thawed embryo transfer cycles numbered consecutively up to seven cycles. The conservative estimate of CLBR after three, six and seven cycles is 40.3% (SE), 51.5% (SE) and 52.9% (SE), respectively, whereas for the optimal estimate, corresponding figures are 45.2% (SE), 68.8% (SE) and 75.1% (SE). Cumulative multiple live birth rates are 4.7% (SE) for the conservative estimate and 7.2% (SE) for the optimal estimate after six fresh and frozen-thawed embryo transfer cycles numbered consecutively for the whole cohort. The comparison of our results with those by Malizia et al. (2009) showed similar CLBR after six cycles (51.5 versus 53%; P ¼ 0.08) for the conservative estimate but lower CLBR after three and six cycles for the optimal estimate (45 versus 51%, P, and 69 versus 72%, P ¼ 0.004, respectively). Data from Table III were also compared with CLBRs of SART for 2008 (Luke et al., 2012). Both conservative and optimal estimates of CLBR are significantly higher (P, ) in the SART registry (53 versus 55% and 75 versus 88% for the CLBR after seven cycles, respectively). Comparison with the Australian New Zealand ART data as registered by the National Perinatal Epidemiology and Statistics Unit (NPESU, 2011) analyses patients that started their first autologous fresh ART treatment cycle during After seven consecutive cycles, the CLBR was 41.1% which was significantly lower (P, ). Discussion The CLBR after six fresh ART cycles and subsequent frozen-thawed embryo transfer cycles was 54.1% for the conservative estimate and 76.3% for the optimal estimate with a cumulative multiple live birth rate of 5.1 and 8.6% for the conservative and optimal estimate, respectively. The novel aspect of the study isthe reporting of the cumulative rates over complete courses of treatment per woman (i.e. all fresh and frozen cycles related to one episode of ovarian stimulation and oocyte retrieval) rather than after multiple fresh or frozen cycles per woman. No previous studies have reported large numbers of patients and cycles as in our reimbursement system, where frozen-thawed embryo transfer cycles are attached to the cycles they originate from. We aimed to compare our data with those available in the literature bearing in mind both patient populations and ART treatment may differ and change over time and is subject to legislation, policy, practice and technology factors. When we compared our data with those reported by Malizia et al. (2009) from a single centre where patients also received insurance coverage for infertility care, similar CLBRs for the conservative estimates after six cycles were observed. However, they reported a cumulative multiple birth rate of 28.6% after six cycles in the optimal estimate of cumulative live births (Malizia et al., 2013). This exceeds by far the 7.4% of cumulative multiple birth rate when analysing our data according to their cycle counting method. When we compare our data to the CLBRs in the SART registry calculated by Luke et al. (2012) for 2008 where up to seven consecutive fresh and frozen-thawed embryo transfer cycles were calculated, conservative and optimal estimates of CLBR are significantly higher in the SART registry. Unfortunately, their study did not contain information on (cumulative) multiple birth rates but the multiple live birth rate in

5 Cumulative live birth and multiple birth rates 97 Figure 1 Cumulative birth rate by age group: For each course of treatment (Rank) the results of fresh and frozen-thawed embryo transfer cycles following an oocyte recovery are included. (A) Conservative estimate (B) optimal estimate. the SART data of 2008 was 32% per embryo transfer cycle in 2008 and decreased to 28% in 2011 ( Corresponding figures from our data showed conditional multiple birth rates ranging from 0.7% for the first cycle to about 3% at the seventh cycle. We did report only 0.03% of the fresh cycles to have had a selective embryo reduction. This is another favorable evolution of reducing the number of embryos transferred. Comparison with the Australian New Zealand ART data as registered by the National Perinatal Epidemiology and Statistics Unit for 2011 show a significantly lower CLBR (41.1%; P, ) after seven consecutive cycles. An exact comparison is again difficult to make as the NPESU data include women of all ages, whereas in our database only women,43 are included. NPESU reports a reduction in the rate of multiple deliveries, with a decrease from 10.0% in 2007 to 6.9% in This was due to a shift to single embryo transfer, with the proportion increasing from 63.7% in 2007 to 73.2% in Clinical pregnancy rates remained stable at about 23.0% per initiated cycle. However, in this way of counting cycles, where fresh and frozenthawed embryo transfer cycles are counted consecutively, a different

6 98 De Neubourg et al. Table III Cumulative birth rate: All embryo transfers (fresh and frozen-thawed) in chronological order. Embryo transfer cycle Number of women Number of live births Conditional LBR (%) Conservative CLBR (%) SE conservative CLBR (%) Optimal CLBR (%) SE optimal CLBR (%) Withdrawal (%) Mean (SD) number of embryos transferred 1.2 (0.45) 1.6 (0.57) 1.7 (0.59) 1.8 (0.63) 1.8 (0.70) 1.8 (0.66) 1.8 (0.62) LBR, live birth rate; CLBR, cumulative live birth rate; SD, standard deviation. amount of fresh cycles per patient could have been included as the number of embryos transferred is higher in the paper by Malizia et al. (2009) allowing a faster passage to a new fresh cycle when compared with our data with fewer numbers of embryos for transfer. We could not compare our data before 1 July 2003 even if the reimbursement system was introduced with the same precision, as we did not register the unique patient identification number that allows the precise accumulation of fresh and frozen-thawed embryo transfer cycles per patient at that time. Our study shows that with a reimbursement system of the majority of ART-related costs up to six fresh cycles coupled to a reduction in the number of embryos for transfer, patients are given an excellent chance to achieve a live birth with a low risk of achieving a multiple pregnancy. The Belgian model, which contains a judicious application of single embryo transfer is offering a public health model for regulation and reimbursement of ART practice worldwide (Arie, 2014; De Neubourg et al., 2014). Our study shows that withdrawal of patients from further treatment increases from 23.7% after the first cycle to 46.9% after five cycles with only few women (3.2% given a mean follow-up of 1.6 years) returning to a fertility centre for six cycles. By contrast with other studies, these withdrawal rates did not include couples switching between centres (7.6%), because they remained included in the analysis due to the mandatory national character of our registration. The high withdrawal rates suggest that even in the absence of a substantial financial burden, ART treatment may bring along physical and emotional strain (Olivius et al., 2002; Malizia et al., 2009). It may also be a reflection of withdrawal of patients from further treatment after informative censoring in case of poor prognosis. The strength of our study is that all the patients in the national health insurance system who started ART treatment and all of their fresh and frozen-thawed cycles were included in the analysis. This was due to the use of a unique identification number that allowed the traceability of all the cycles a patient performed, even across centres, and also due to the mandatory registration of all ART activity performed in Belgium. The weakness of our study lies in the fact that we did not include naturally conceived pregnancies or pregnancies obtained after a switch to intrauterine insemination when the ART cycle was cancelled due to poor response. Limitations of our results may be a possible underestimation of the CLBR as not all patients had a follow-up long enough to allow patients to have been treated for up to six cycles. Neither could we make a comparison with cycles performed before the implementation of a reduction in the number of embryos transferred as the unique identification number was only introduced since the middle of 2009 and thus besides a reduction of more than 50% of multiple pregnancies, the exact impact of the strategy on CLBR could not be calculated. Future research could be directed to the investigation of the reasons for withdrawal of patients taking into account informative censoring issues (e.g. embryo quality) as well as patient centeredness quality of care aspects. Supplementary data Supplementary data are available at Acknowledgements The authors acknowledge the contribution of all gynaecologists, embryologists, lab technicians, nurses, midwives and secretaries in the data acquisition for Belrap. Authors roles D.D.N.: conception and study design, acquisition of data, analysis and interpretation of data, writing of manuscript. K.B.: acquisition of data, analysis and interpretation of data, critical review of manuscript. C.B., T.C., A.D., F.D., M.D., N.G. and S.G.: interpretation of data, critical review of manuscript. C.W.: conception and study design, interpretation of data, critical review of manuscript. Funding No external funding was used for this study. Conflict of Interest All authors declare they did not receive financial support from a company for the submitted work. Nor the authors nor their spouses, partners or

7 Cumulative live birth and multiple birth rates 99 children have financial relationships that may be relevant to the submitted work and none of the authors has non-financial intereststhat may be relevant to the submitted work. References Adashi EY, Wyden R. Public reporting of clinical outcomes of assisted reproductive technology programs: implications for other medical and surgical procedures. JAMA 2011;306: Arie S. Twin dilemma. BMJ 2014;348:f7603. Barnhart KT. Live birth is the correct outcome for clinical trials evaluating therapy for the infertile couple. Fertil Steril 2014; 101: Daya S. Life table (survival) analysis to generate cumulative pregnancy rates in assisted reproduction: are we overestimating our success rates? Hum Reprod 2005;20: De Neubourg D, Bogaerts K, Wyns C, Albert A, Camus M, Candeur M, Degueldre M, Delbaere A, Delvigne A, De Sutter P et al. The history of Belgian assisted reproduction technology cycle registration and control: a case study in reducing the incidence of multiple pregnancy. Hum Reprod 2013;28: De Neubourg D, Peeraer K, Debrock S, D Hooghe T. Belgium model of coupling reimbursement of ART costs to restriction in number of embryos transferred. BMJ 2014;348:g1559. Luke B, Brown MB, Wantman E, Lederman A, Gibbons W, Schattman GL, Lobo RA, Leach RE, Stern JE. Cumulative birth rates with linked assisted reproductive technology cycles. N Engl J Med 2012;366: Malizia BA, Hacker MR, Penzias AS. Cumulative live-birth rates after in vitro fertilization. N Engl J Med 2009;360: Malizia BA, Dodge LE, Penzias AS, Hacker MR. The cumulative probability of liveborn multiples after in vitro fertilization: a cohort study of more than 10,000 women. Fertil Steril 2013;99: National Perinatal Epidemiology and Statistics Unit. edu.au (14 August 2015, date last accessed). Olivius K, Friden B, Lundin K, Bergh C. Cumulative probability of live birth after three in vitro fertilization/intracytoplasmic sperm injection cycles. Fertil Steril 2002;77: Peeraer K, Debrock S, Laenen A, De Loecker P, Spiessens C, De Neubourg D, D Hooghe TM. The impact of legally restricted embryo transfer and reimbursement policy on cumulative delivery rate after treatment with assisted reproduction technology. Hum Reprod 2014; 29: