Standardized Calculation and Reporting of Live Birth Rate in IVF

Definition and Counting Principles

Live birth in IVF is defined as the expulsion or extraction of a product of fertilization after 20 weeks gestation that shows any sign of life (e.g., heartbeat, cord pulsation, or voluntary movement) 1.
Live births must be counted as birth events, not as individual infants; a twin delivery counts as one live‑birth event 1.

The success rate should be expressed per participant randomized or per cycle started to preserve intention‑to‑treat integrity and reflect the true patient‑centered outcome 1.
Using “per embryo transfer” as the denominator inflates success rates because it excludes patients who never reach transfer (used in 15 % of studies) 2.
Using “per oocyte retrieval” excludes patients who fail stimulation, also biasing results 2.
Restricted denominators typically omit a median of 8 % of participants (range 2–38 %) and introduce selection bias 3.

For cumulative rates across multiple fresh and frozen transfers, apply life‑table analysis with all patients who initiated a treatment course as the denominator, including those who discontinue treatment 3.
Only 3–5 % of published studies correctly report cumulative live‑birth rates using this approach 3.
Patients who never reach embryo transfer must be counted as treatment failures to avoid overestimation 3.

Report effect‑size estimates with 95 % confidence intervals for live‑birth events 1.
Present outcomes for singleton, twin, and higher‑order multiple births separately 1.
Provide gestational age at birth as median and interquartile range 1.
Record birthweight for each newborn (measured within 24 h with a calibrated electronic scale) and do not adjust birthweight for gestational age 1.
Never calculate live‑birth rate per embryo transferred, as embryos are not statistically independent 3.

51 % of studies fail to define “live birth” at all, contributing to heterogeneity 4.
Over 800 different numerator/denominator combinations have been reported in the literature 3.
Only 42 % of studies use the point of randomization as the denominator for live‑birth reporting 3.

Cumulative pregnancy rate in IVF should be calculated with life‑table analysis, using the total number of pregnancies (clinical, ongoing, or live birth) as the numerator and all patients who initiated a treatment course as the denominator, thereby capturing outcomes across multiple fresh and frozen embryo transfers from one or more stimulation cycles.  5
The denominator “per course of treatment started” provides the most realistic assessment for patients deciding whether to undergo IVF, compared with denominators based on individual transfers or oocyte retrievals. [5][6]

Using sub‑group denominators such as “per transfer” or “per oocyte retrieval” excludes a median 8 % of participants (range 2‑38 %) and introduces bias, undermining the balance of confounding factors established by randomisation in RCTs. [7][8]
Patients who never reach embryo transfer must be counted as treatment failures (unsuccessful response) to preserve intention‑to‑treat integrity. 8

Cumulative clinical pregnancy should be reported per course of treatment started; only 1 % of published IVF RCTs adhered to this recommendation. 5
Cumulative ongoing pregnancy should also be reported per course of treatment started; only 2 % of studies complied. 5
Cumulative live birth should be reported per cycle started, with only 3‑5 % of studies presenting cumulative live‑birth rates across multiple fresh and frozen transfers. [9][6]

Life‑table analysis is the standard method for estimating cumulative pregnancy rates because it accommodates censoring of patients who discontinue treatment at different time points. 5 (systematic review)

Do not report cumulative pregnancy rates per embryo transferred; embryos are not statistically independent, and outcomes must be defined at the patient level. 8
Do not exclude patients who fail to reach a treatment stage (e.g., oocyte retrieval or embryo transfer) from the denominator, as this violates intention‑to‑treat principles and creates selection bias. [7][6]8
Do not use a single outcome measure without pre‑specification; reporting multiple un‑pre‑specified outcomes increases the risk of false‑positive findings. 8