Four Statistical Definitions Behind IVF Success Rates

FS FERTICARE｜Special topic on assisted reproduction data reading Data verification date: September 26, 2026

Summary

An IVF “success rate” depends on four choices: the outcome counted, the denominator, which cycles are included, and the length of follow-up. The same treatment cohort can therefore produce different percentages. In SART’s final 2023 national report, among intended retrievals with patients’ own eggs under age 35, the live-birth rate was 39.4% for the first embryo transfer and 53.2% cumulatively for all related transfers within one year. The corresponding figures in SART’s preliminary 2024 report were 34.8% and 41.7%. In ESHRE’s 24th European report, the 2020 clinical-pregnancy rate per IVF retrieval was 22.1%, rising to 26.4% when freeze-all cycles were excluded. This article uses CDC, SART, HFEA and ESHRE definitions and a SART registry study to show what each choice changes and what to request from a clinic.

Keywords: Assisted reproduction; live birth rate; intended oocyte retrieval; embryo transfer; cumulative live births; inclusion rules; gestational-carrier pregnancy

Methods: This explanatory review uses definitions from CDC and HFEA, numerical results from the specified SART, HFEA and ESHRE reports, and a JAMA retrospective study of SART transfer cycles. Because the sources differ by year, country, population and follow-up, their rates are not pooled. The hypothetical cohort used to explain the arithmetic is identified separately.

1. Endpoint: from positive pregnancy test to live birth

During a period of treatment, a positive pregnancy test, ultrasound confirmation of intrauterine pregnancy, continued pregnancy, and live birth occurred in sequence. The CDC glossary defines clinical pregnancy as the detection of an intrauterine gestational sac by ultrasound, and defines a live birth as at least one baby born alive in one delivery; twin births are counted as one [2] in the number of live births. So the numerator of "live birth rate" is the number of births, and "number of babies born" is another count.

The gap between the various endpoints has clinical origin. Biochemical pregnancy, early miscarriage, ectopic pregnancy, and second- and third-trimester pregnancy losses occur sequentially after a positive pregnancy test, with each step resulting in a lower count for subsequent endpoints than the previous step. In the 24th ESHRE report, in the fresh cycles of in vitro fertilization and intracytoplasmic sperm injection in 2020, the proportions of embryo transfering 1, 2, 3 and 4 or more embryos were 57.2%, 37.8%, 4.8% and 0.2% respectively, and the proportion of twin births was 10.9%[7]. Multiple births affect both the relationship between live births and number of babies born, as well as preterm birth and neonatal outcomes.

In its national summary report, SART listed delivery of a live baby as the primary outcome, and singleton delivery at term (more than 37 weeks) as the optimal outcome [4]. The main results below in this article uniformly use live-birth deliveries and indicate the denominator in the figures and tables. A document that simply states "pregnancy rate" or "success rate" lacks information to judge its comparability to the "live birth rate" in another report.

2. Denominator: start counting at which step of treatment

An intended retrieval is counted when treatment or monitoring begins with the aim of collecting oocytes. A completed retrieval requires the procedure to take place; a transfer cycle requires an embryo transfer. Some intended retrievals are cancelled, some completed retrievals produce no transferable embryo, and some embryos are frozen for later transfer. CDC reports results with all three denominators for cycles using patients’ own eggs [1].

The following cohort is hypothetical and illustrates the arithmetic; it is neither a clinic result nor an individual prognosis. Of 100 intended retrievals, 90 reach retrieval and yield 80 transfers, followed by 35 live-birth deliveries. Assume these deliveries arise from different retrievals. Eighty transfers could arise from fewer than 80 retrievals because one retrieval may produce several embryos.

The table holds the same live-birth events constant and changes only the denominator, so the three rates rise from 35.0% to 38.9% and 43.8%. The real report changes the year of observation, included population, number of embryo transfers, and follow-up time at the same time, and the values ​​are not necessarily arranged in this order.

Births per embryo transferred and live births per transfer are different measures. One transfer procedure can involve more than one embryo. HFEA’s clinic measure divides births by the number of embryos transferred [3]. As multiple-embryo transfers become more common, this rate can diverge further from a per-transfer rate.

3. Inclusion rules: which cycles enter the denominator

The denominators have the same name, but the periods included in them are not necessarily the same. The 24th ESHRE report provides a set of direct comparisons: In 2020, 1,440 institutions in 41 countries reported a total of 923,318 treatment cycles; among them, the clinical pregnancy rate for each oocyte retrieval was 22.1% for in vitro fertilization and 20.0% for intracytoplasmic sperm injection. The same report was recalculated after excluded the freeze-all cycle , and the two items were 26.4% and 25.6% [7] respectively.

These two sets of numbers come from the same registry, the same year, the same endpoint, and the same denominator name. The difference comes from inclusion rules. A freeze-all cycle has no fresh transfer after retrieval. Including it in a per-retrieval denominator lowers the contemporaneous pregnancy rate; excluding it raises the rate by 4.3 percentage points.

Exclusion rules are set in each system. The birth rate per embryo reported in HFEA's 2024 Trends and Data report excludes preimplantation genetic testing (PGT-A/PGT-M/PGT-SR), donated egg treatment, gestational-carrier pregnancy, and cycles with recorded pregnancy but no birth outcome, and includes fresh and frozen in vitro IVF and ICSI using patients’ own eggs [6]. SART's autologous egg forms and donated egg forms are separated. The CDC also displays [1] separately from autologous eggs and donated eggs.

Therefore, before comparing two percentages you need to confirm which cycles they exclude.Rates with different exclusions cannot be compared meaningfully even if their endpoint, denominator label and year match.

4. Observation time window: first embryo transfer, one-year accumulation and preliminary data

A cumulative live-birth rate links later transfers back to an earlier retrieval. SART attributes the first transfer and subsequent transfers within its specified window to the originating intended retrieval; its “all embryo transfers” measure follows outcomes within one year of cycle start [4]. CDC also links retrievals to transfers within a one-year window [1]. A later transfer of an embryo still in storage can lead to a live birth outside that window.

Taking SART ’s 2023 final national summary report as an example, the following tables are all sources of autologous eggs, and the denominators are all intended oocyte retrieval cycles [4].

The gap between cumulative and first-transfer rates narrows with age: 13.8 percentage points under 35 and 0.4 points over 42. This gradient reflects that the younger group generally has more embryos available for subsequent transfers after one oocyte retrieval. The two columns have similar but not identical denominators; subtracting them does not isolate a second transfer’s success rate.

The corresponding 2024 SART report is preliminary [8]:

SART explains that for cycles initiated within the reporting year, the first embryo transfer outcome will often not be determined until the next reporting year. Therefore, each annual report is marked as a preliminary version when it is released in the first year, and is converted to the final version [8] in the following year. In the 2024 preliminary version, the cumulative column is 6.9 percentage points higher than the first embryo transfer column (under 35 years old), and in the 2023 final version, it is 13.8 percentage points. The two reports have captured different amounts of follow-up and correspond to different years of treatment, and these two differences cannot be regarded as changes in outcomes between years.

The results of repeated transfers are correlated: they often use embryos from the same round of oocyte retrieval, and the underlying conditions of the recipients are the same. Repeated multiplication of a single embryo transfer rate to extrapolate an individual's chance after two or three times requires the additional assumption that each time is independent of each other; the cumulative rate of SART directly tracks oocyte retrieval cycles that comply with the rules and their subsequent results. During consultation, you can request data tracked by the institution based on the same oocyte retrieval cohort.

5. Treatment cycles outside a per-transfer rate

A per-transfer rate includes only cycles that reached transfer. Cancelled retrievals, cycles without a usable embryo and postponed transfers are absent from that denominator. An intended-retrieval rate retains those earlier steps. The former is more relevant once an embryo is available; the latter better describes the pathway from treatment start.

The way age is recorded is a second level option.Age should refer to the egg provider at retrieval. At a later frozen transfer, the age on transfer day does not describe the age of the egg. HFEA groups frozen transfers by age at egg collection [6]; SART also uses age at retrieval [4].

The proportion of cryoembryo transfer in recent years determines the scope of this distinction. The HFEA reports that the proportion of frozen embryo transfers in all IVF cycles in the UK has increased from 24% in 2014 to 48% in 2024. The higher the proportion of frozen embryo transfer, the greater the difference between the two recording methods of "age on the day of embryo transfer" and "age at the time of oocyte retrieval", and the further the difference between the results based on embryo transfer statistics and those based on oocyte retrieval statistics. The CDC also pointed out that the patient composition and treatment methods differ between institutions, and the aggregate success rate corresponds to the population of patients treated in the institution, not the probability of a particular patient.

6. Egg source and pregnancy burden are two independent pieces of information.

In this section, pregnancy without a gestational carrier is used to refer to the recipient carrying out the pregnancy, and gestational-carrier pregnancy is used to refer to the embryo being transferred and conceived by another recipient of the pregnancy.This identifies who carries the pregnancy, not where the egg came from. In either arrangement, an embryo may originate from the intended parent’s egg, a donor egg or a donated embryo.

“Live-birth rate with donor eggs” and “live-birth rate with a gestational carrier” refer to different dimensions. A clinic comparison should hold egg source, age at retrieval, embryo type and transfer definition as similar as possible, and report numerator, denominator and sample size for both groups.

A study published in the Journal of the American Medical Association (JAMA) in 2023 used registration data from the SART Clinical Outcome Reporting System and included who had undergone embryo transfer from 2014 to 2020. There were 1,008,205 cycles of implantation of , of which 40,177 (4.0%) used gestational-carrier pregnancy, and 968,028 (96.0%) did not use [5]. The pregnancy without a gestational carrier group in the study was a "embryo transfer cycle that did not use gestational-carrier pregnancy", not a cycle that "all used my own eggs."

The difference in unadjusted live birth rates per embryo transfer between the two groups was 10.7 percentage points. After adjusting for egg age, infertility diagnosis, egg donation, single sperm injection, assisted hatching, fresh or frozen embryos, embryo testing, transfer day and number of embryos transferred, the relative risk ratio of live birth in the gestational-carrier pregnancy group compared with the pregnancy without a gestational carrier group was 1.11 (95% confidence interval 1.10-1.12) [5]. 1.11 is the relative risk ratio after adjusting for covariates, which cannot be directly regarded as the causal gain of gestational-carrier pregnancy on an individual's chance of live birth; the unadjusted 10.7 percentage point difference also includes differences in the composition of patients and treatments between the two groups.

The authors of the study listed three limitations: the inability to examine pregnancy complications, the inability to handle the aggregation caused by the same pregnancy bearer appearing in multiple cycles, and the entry errors in the registration data [5]. The term aggregation has the same origin as the independence issue in Section 4 : Multiple cycles of the same pregnancy bearer are not independent observations, and calculating them as independent cycles will underestimate the uncertainty of the estimate. The study only included cycles that reached the transfer stage, and its results corresponded to the starting point of "embryos already available for transfer."

7. Reporting years and definitions in four registries

The four registries publish at different intervals and use different definitions. This table shows what was available when checked on 26 September 2026.

In the HFEA 2024 report, the birth rate of each embryo transferred after the combination of autologous egg in vitro fertilization and single sperm injection is 30%, including 38% for those aged 18-34 and 8% for those aged 43-44. The birth outcomes for this year are still marked as preliminary data [6] in the report. The 2023 numbers above SART are the one-year cumulative live birth rate for U.S. member clinics, autologous eggs, and each intended oocyte retrieval. The two groups of percentages have different countries, years, numerators, denominators, inclusion rules and follow-up methods, and the horizontal subtraction has no interpretable meaning.

ESHRE explained in its report that the data collection systems and reporting completeness of European countries are different, and some countries are unable to provide the number of initiation cycles or the number of deliveries [7]. CDC cautions that published rates reflect cycles performed more than a year earlier and may not describe current practice [1].

Comparisons between institutions suffer from the same problem. Patient age, ovarian reserve, number of previous treatments, whether patients with poor prognosis are admitted, and the clinic's rules for attributing canceled cycles to subsequent embryo transfers will all change aggregation rates. Annual fluctuations are particularly evident when sample sizes are small: a category with only 20 embryo transfers showing 50% for 10 live births turns into 55% or 45% for one more or fewer embryo transfers respectively. The CDC recommends interpreting institutional results within the context of patient composition and treatment methods, noting that clinics with fewer cycles have greater fluctuations in success rates from year to year.

When evaluating an institution, you can simultaneously check the single live birth ratio, laboratory quality, completeness of result disclosure, and suitability of treatment plans.

8. Institutional Data Checklist

These ten fields form a reviewable statistical statement.If a clinic cannot provide a subgroup rate, record it as “not stratified” or “not published” and discuss treatment using the information available. Filling the gap with published averages from another country and another group would merge numbers under different inclusion rules into the same judgment.

Conclusion

A success rate is interpretable only when the outcome, denominator, inclusion rules and follow-up window are stated together. An intended-retrieval rate includes early cancellations and cycles without transfer; a per-transfer rate describes only cycles that reached transfer. Inclusion rules determine who enters each denominator. A cumulative rate depends on later transfers and follow-up completeness, especially when a report is preliminary. Egg source and who carries the pregnancy require separate stratification.

The registry figures here help frame questions for a clinic; they are not an individual forecast. A clinical team must assess a person’s history, ovarian reserve and embryo information.

Source

[1] CDC. How to Interpret ART Success Rates. Page updated in 2024. https://www.cdc.gov/art/success-rates/interpret.html [2] CDC. Glossary of Terms, Assisted Reproductive Technology. 2024. https://www.cdc.gov/art/glossary/index.html [3] HFEA. Fertility treatment 2023: Quality and Methodology Report. https://www.hfea.gov.uk/about-us/publications/research-and-data/fertility-treatment-2023-trends-and-figures/quality-and-methodology-report/ [4] SART. Final National Summary Report for 2023. https://www.sartcorsonline.com/CSR/PublicSnapshotReport?reportingYear=2023 [5] Shandley LM, DeSantis CE, Lee JC, Kawwass JF, Hipp HS. Trends and Outcomes of Assisted Reproductive Technology Cycles Using a Gestational Carrier Between 2014 and 2020. JAMA. 2023;330(17):1691–1694. doi:10.1001/jama.2023.11023. Full text also available at PMC10585492. https://pmc.ncbi.nlm.nih.gov/articles/PMC10585492/ [6] HFEA. Fertility treatment 2024: trends and figures. Released on June 16, 2026. https://www.hfea.gov.uk/about-us/publications/research-and-data/fertility-treatment-2024-trends-and-figures [7] European IVF Monitoring Consortium (EIM) for ESHRE; Smeenk J, Wyns C, De Geyter C, et al. ART in Europe, 2020: results generated from European registries by ESHRE. Human Reproduction. 2025;40(11):2038–2055. doi:10.1093/humrep/deaf179. https://pubmed.ncbi.nlm.nih.gov/40985526/ [8] SART. Preliminary National Summary Report for 2024. https://www.sartcorsonline.com/CSR/PublicSnapshotReport?reportingYear=2024

Figure notes: Figure 1 uses a hypothetical teaching cohort. Figures 2 and 5 redraw SART’s 2023 final and 2024 preliminary data; Figure 4 redraws the JAMA study by Shandley and colleagues. Concept illustrations show statistical structure, not real embryos or treatment. This article is for public science and does not replace personal diagnosis and treatment advice.