Calculation of Cumulative Pregnancy Rate in IVF
Cumulative pregnancy rate in IVF is calculated using life-table analysis with the number of pregnancies achieved as the numerator and the number of patients who started treatment (per course of treatment started) as the denominator, tracking outcomes across multiple fresh and frozen embryo transfer cycles from a single or multiple stimulation cycles. 1
Standard Calculation Method
The most appropriate denominator is "per course of treatment started" rather than per transfer or per oocyte retrieval, as this provides the most realistic assessment of success for patients deciding whether to undertake IVF treatment. 1
Key Components of the Calculation:
Numerator: Total number of pregnancies (clinical, ongoing, or live births depending on the outcome measured) achieved across all cycles 1
Denominator: All patients who initiated treatment, regardless of whether they completed all planned cycles 1
Time frame: Cumulative rates track outcomes across multiple treatment cycles (typically 3-6 cycles) within a defined period 2, 3, 4
Methodological Considerations
Handling Patient Drop-Out
Three approaches exist for managing patients who discontinue treatment, with the third method providing the most realistic estimates: 2
Method 1 (underestimates): Assumes patients who stopped treatment have zero chance of pregnancy 2
Method 2 (overestimates): Assumes patients who stopped have the same pregnancy probability as those who continued—this is the most commonly used but inflates success rates 2, 5
Method 3 (most realistic): Assumes only patients who stopped for medical reasons have zero pregnancy chance, while others who stopped for non-medical reasons have the same probability as those who continued 2
Critical Pitfall: Denominator Selection
A major methodological error occurs when researchers use subgroups (per transfer or per oocyte retrieval) rather than the randomized cohort as the denominator. This undermines the validity of randomized trials because it disrupts the balance of confounding factors established by randomization. 1, 6
Studies using "per transfer" denominators exclude a median of 8% (range 2-38%) of participants, potentially biasing results 1
The correct approach defines patients who fail to reach embryo transfer as having an unsuccessful treatment response 1
Reporting Standards
Outcome-Specific Denominators
According to the 2016 Human Reproduction Update systematic review of IVF RCTs: 1
Cumulative clinical pregnancy: Should be reported per course of treatment started (only 1% of studies did this correctly) 1
Cumulative ongoing pregnancy: Should be reported per course of treatment started (only 2% of studies did this correctly) 1
Cumulative live birth: Should be reported per cycle started (only 3-5% of studies reported this cumulatively across multiple fresh and frozen transfers) 1
Proper Statistical Approach
Life-table analysis is the standard statistical method for calculating cumulative pregnancy rates, allowing for censoring of patients who discontinue treatment at different time points. 2, 5, 7, 4
The Cox proportional hazards model can estimate the influence of covariates (age, diagnosis, etc.) on cumulative success 7
Cumulative rates should account for both fresh and frozen embryo transfers from the same stimulation cycle 1
Expected Cumulative Rates
Research demonstrates that cumulative pregnancy rates increase substantially with multiple cycles: 7, 3, 4
After 3 cycles: Clinical pregnancy rates of 54-57%, live birth rates of 48-51% 3
Age-dependent decline: Cumulative rates decline significantly with increasing maternal age 3
Common Errors to Avoid
Never report cumulative pregnancy rates per embryo transferred—embryos are not statistically independent units, and outcomes are defined at the patient level. 1, 6
Never exclude patients who failed to reach certain treatment stages (oocyte retrieval, embryo transfer) from the denominator, as this violates intention-to-treat principles and creates selection bias. 1
Never use a single outcome measure without pre-specification, as reporting multiple outcomes increases the risk of false discoveries through multiple testing. 1, 6