What are my chances of getting pregnant? More specifically, if I am having trouble getting pregnant, will I be successful if I undergo the process of in vitro fertilization (IVF)? These are important questions especially since infertility treatments (especially IVF) is time consuming and costly.
There are tools to assess a woman's fertility status. For instance, Fertistat is a free on-line questionnaire that was developed at Cardiff University in the UK. Its primary function is as a fertility awareness tool.
A recent publication by Choi et al reported a personalized model for predicting the success rate of women undergoing their first cycle of IVF. It should be noted that this method is now patented and marketed by Univfy Inc. and this blog post is in not an endorsement for this product. Their model utilizes patient demographics and reproductive history, age at time of first IVF, BMI, smoking status, gravity, parity, pregnancy losses before 20 weeks, number of ectopic pregnancies, antral follicle count, day 3 serum FSH the year, patient diagnosis, male partner reproductive health including age, total motile sperm count, use of sperm extraction method and use of donor sperm. They developed their model using data from 13,076 first IVF treatment cycles performed at three clinics in Spain, Canada, and the United States. Then they created their PreIVF-D by combining the data from the three clinics and they used a training set of 1,061 independent cases. Finally they tested their model using another set of data from 1,058 patients. They found that the most important prognostic contributors to their model were patient age, total motile sperm count, BMI, day 3 serum FSH, and antral follicle count.
The authors conclude that age-based estimates of live birth probabilities in IVF treatment are not optimal and that their PreIVF-D model performs better with a 35.7% improvement in the ability to predict live birth. Notably, the area under the curve (AUC) for the age-based prediction was 0.614 and for Pre-IVF-D it was 0.634. This means that the two models were essentially equivalent. There are two limitations to this study that should be pointed out. First, the authors discuss that it is not possible for them to prove that PreIVF-D works for a clinic outside this study. In fact, there may be clinic-specific trends (such as referral types, regional BMI, and treatment approaches) that significantly affect their predictive model. Second, the authors do not mention the methods that each clinic used for FSH measurement. According to 2013 CAP surveys, there are up to 2 fold differences in FSH results depending on the method used. It would be interesting to know if the clinics utilized one FSH method or several as this may contribute to significant clinic-to clinic variations.
Personalized prognostic tools are likely the way of the future in reproductive medicine & infertility treatment. Although their utility now may be limited, they will undoubtedly continue to improve and evolve just as predictive models for assessing risk of downs syndrome has evolved over the past 20 years.