P-755 Conventional outcome reporting per embryo-transfer cycle systematically underestimates the chance of success of women undergoing ART: a source of bias in registries, trials and guidelines. (30th June 2022)

• Record Type: Journal Article
• Authors: Griesinger, G; Larsson, P; Mannaerts, B
• Is Part Of: Human reproduction. Volume 37(2022)Supplement 1
• Publication Date: 2022-06-30
• Subjects: Human reproduction -- Periodicals; 618
• DOI: 10.1093/humrep/deac107.696
• Languages: English
• ISSNs: 0268-1161

Abstract

Abstract: Study question: How should ART treatment success be estimated from data consisting of several treatment attempts per subject (e.g., fresh transfer followed by cryotransfer[s] and/or repetitive attempts)? Summary answer: Live birth rate estimates per transfer cycle will be substantially higher when accounting for differences in the number of embryo transfers between subjects. What is known already: Preconception counselling and expectation management is an integral part of fertility care. In the course of IVF-treatment, the majority of women undergo several embryo transfers (ET). The live birth rate per ET cycle may be severely underestimated since women with a poor prognosis will be overrepresented in the population compared to women with a good prognosis. This is because women achieving a live birth early will contribute few ETs, whereas women without live births will contribute several ETs to the pool of observations. This phenomenon does also add bias when comparing the live birth rates between fresh- transfers and cryo-transfers. Study design, size, duration: The complete data set from the RAINBOW trial (NCT03564509) was applied to exemplify the underestimation of live birth chance of individual women in cryo-cycles. In total 557 women underwent treatment in a fixed, individualized daily- dose of follitropin delta in a long GnRH agonist protocol, IVF/ICSI and blastocyst transfer. Surplus blastocysts were cryopreserved on day 5/6 and subsequently used for cryo-transfers up to one year after start of the stimulation. Participants/materials, setting, methods: Live birth rate was analyzed using a mixed effect logistic regression model where the log-odds of achieving a live birth was assumed to be normally distributed in the trial population. Type of transfer (fresh or cryo) was included as a factor in the model. The delta method was used to transform estimated mean log-odds from the model into estimated live birth rates per transfer, and to estimate the difference between fresh- and cryo-transfers. Main results and the role of chance: Of the 619 women starting stimulation, 557 had at least one fresh- or cryo-transfer. A total of 927 embryo transfers were performed whereof 520 fresh- and 407 cryo-transfers. In total, 252 subjects had at least one cryotransfer, 102 had a second cryotransfer, 37 a third, 12 a fourth, three a fifth, and one had a sixth cryotransfer, respectively. Of the 520 fresh transfers, 212 (41%) resulted in a live birth, whereas for the 407 cryo-transfers 100 (25%) resulted in a live birth (unadjusted difference 16%). When using the mixed effect logistic regression model to adjust for repeated transfers in the same women, the live birth rates were estimated to be 41% for fresh transfers and 36% per transfer for cryo-transfers. The difference between rates for fresh transfer and per cryo-transfer was estimated to be 5.3% with 95% confidence interval ranging from -6.6% to 17.1% and with no statistically significant difference (p = 0.3827). Limitations, reasons for caution: Comparison between live birth rate after fresh transfer and per cryotransfer in the same group of women will already be biased in favour of the fresh transfer. This is because the fresh transfer will normally use the embryo of highest quality, leaving embryos of lower quality for cryo-transfers. Wider implications of the findings: Average chance of success reported per cycle or per embryo-transfer (e.g. registry reports) do not apply to individual couples, especially not at the outset of treatment. Live birth rates per transfer from datasets encompassing multiple transfers from single individuals can be more accurately reported using mixed effect logistic regression models. Trial registration number: NCT03564509