New Statistical Model for Predicting IVF Success

Imputation 'promising' for gauging IVF success
Source: Human Reproduction 2007; Advance online publication
Assessing the accuracy of a new statistical method for estimating IVF success after four cycles.

MedWire News: A statistical method known as multiple imputation offers the best estimate of the success rate of IVF programs, French researchers believe.

The simplest approach to evaluating IVF outcomes is based on pregnancy rate per oocyte retrieval but this has various limitations.

The present study used an alternative imputation method that takes into account treatment interruptions to estimate the cumulative birth rate after four IVF aspirations.

The analysis, by N. Soullier (Institut Fédératif de Recherche Santé Publique, Paris, France) and team, used data from 3037 couples beginning IVF treatment in two French IVF units.

Using multiple imputation, they estimated that cumulative live birth rates were 21 percent at the first aspiration, 35 percent at the second, 41 percent at the third, and 46 percent at the fourth.

The actual percentage of live births after four aspirations was 37 percent, whereas life-table analysis (Kaplan-Meier) gave a 52 percent success rate.

Soullier et al say their innovative approach is a "pertinent estimation" of the probability that a couple entering an IVF unit will have a child if they do not interrupt treatment until four aspirations or a delivery.

They conclude that multiple imputation is "a promising method" and call for it to be more thoroughly explored.