40,177 Transfers: What Surrogacy Success Rates Count

A number of transfers is not a number of babies. That sounds obvious, and it is the distinction that most of the figures quoted at you during a surrogacy search quietly breaks.

The largest published study of transfers to surrogates in the United States analysed 40,177 transfers performed between 2014 and 2020, and its figures come from SART CORS, the reporting system the professional society for United States fertility clinics maintains. It is a useful document precisely because of what it does not do: it does not promise you anything, and it does not measure the thing a journey is actually made of.

What a rate is a rate of

Four different denominators get used in this subject, and they produce four different numbers from the same clinic in the same year.

• Per transfer. Of the transfers performed, how many produced a pregnancy or a live birth. This is the denominator the study above uses.
• Per cycle. Of the treatment cycles started, how many produced anything. A cycle can produce no transferable embryo at all, so a per cycle figure is always lower than the per transfer figure drawn from the same period.
• Per retrieval, or per intended parent. How many people who started ended with a child. This is the number closest to what you are actually buying, and it is the number registry data is least able to give you directly.
• Per program. A provider’s own count, across cases they selected, with a definition they chose and usually no published methodology.

Ask which of those four a quoted figure is, and watch how often the answer is that nobody has counted. If a page tells you a rate and does not name the denominator, it is not a statistic. It is a mood.

What that cohort reported

The study’s own figures are per transfer, and they are worth having in front of you because they are the opposite of the vague reassurance a search returns.

1. 62.4 percent clinical pregnancy per transfer. The paper defines a clinical pregnancy as an intrauterine pregnancy that can be identified, which is a different thing from a baby going home.
2. 47.9 percent live birth per transfer. Fewer than half of the transfers in that cohort ended in a live birth, which is the arithmetic reason a plan that funds one transfer is a plan that assumes a coin fall.
3. 28.9 percent multiple births among those live births. Roughly one live birth in three in the cohort was a multiple birth, and a twin or higher order pregnancy is a cost and a medical risk that no quoted package prices in advance.
4. 32 percent of those transfers were for intended parents living outside the United States. If you are arranging care in the US from another country, that is not a trivia item. It means a large share of the cohort is travelling into the treatment rather than living beside it, which is exactly the situation in which scheduling and cost assumptions fail.

What the study does not report is how many transfers it took per person. There is no mean or median number of transfers performed in the published figures, so nobody can convert that 47.9 percent into a number of attempts, and a provider who quotes you one is working from their own files rather than from this data.

What the registry behind it cannot tell you

The description of SART CORS is a useful corrective here, because the same dataset is quoted in so many marketing pages.

• It holds cycle level clinical data from more than ninety percent of United States clinics providing assisted reproduction, which is why it is worth reading at all.
• It records cycle starts, patient demographics and linked cycles. It does not record surrogacy agreements.
• It records nothing about what you paid, what a contract said, or whether a program delivered on it. A program outcome is not the same object as a clinical outcome, and the registry only holds the second.

So the honest sentence is that published registry data can tell you roughly how a transfer performs, and cannot tell you how a program performs. Those are different purchases.

The five questions a rate should be tested against

Put these to a clinic or an agency in writing, and record the answer against the name that gave it. The question checklist puts them into one file, because an answer you cannot attribute is an answer you cannot rely on later.

1. Is the figure you are quoting per transfer, per cycle, per retrieval, or per intended parent who started with you?
2. Over what period, and does it include the cycles that produced no transferable embryo?
3. How do you count a journey that needed three transfers, and does your published rate include the first two?
4. What is your multiple birth rate, and what do you tell patients about the cost and risk of a twin pregnancy?
5. For intended parents living outside the United States, how many trips does your own data show, and what happens to the count when a transfer is postponed?

Question five is the one that turns a clinical statistic into a budget. The quote comparison tool is where the answer belongs, against the clinic and medication cost, which is the part of this subject a rate never mentions and a total always contains.

What this does not mean

None of the above says a United States surrogacy program is unlikely to work. It says the published numbers are per transfer, that per transfer is not per journey, and that the difference between those two is where a plan either holds or does not.

The reason there is no reliable average for a whole journey is the same reason there is no reliable average cost, which is set out in why no average surrogacy cost exists. Both are questions about a sequence of events that only your own file will resolve, and both are answered by asking for the denominator rather than the number.

What is reported above is published figures and the definitions attached to them. It is not a prediction about your treatment, and it is not medical advice.