How to read a fertility study without being misled

1. What outcome was measured?

In fertility research the outcome that matters to patients is a live birth. Many studies report earlier steps instead: a positive pregnancy test, a "clinical pregnancy" seen on ultrasound, or even the number of eggs collected or embryos created. These are useful, but each step loses some cases along the way.

National data show the gap. In the UK in 2023, the average pregnancy rate per fresh embryo transferred was 31%, while the birth rate was 25% (HFEA, Fertility treatment 2023). A study that reports only pregnancies will always look better than one that reports births.

Researchers have examined how much this matters for trials. A study of randomised trials included in Cochrane reviews found that only a minority of trials in reproductive medicine reported live birth; among 143 trials that reported both outcomes, conclusions based on clinical pregnancy and on live birth agreed closely (Clarke J. F. et al., "Measuring outcomes in fertility trials: can we rely on clinical pregnancy rates?", Fertility and Sterility, 2010). Agreement on average does not guarantee agreement in every case, which is why trial guidelines now ask for live birth as the primary outcome.

2. Per what? Check the denominator

A success rate is a fraction, and the bottom of the fraction changes everything:

• per embryo transferred, as the UK regulator reports;
• per cycle started, which also counts cycles stopped before transfer;
• per patient over several cycles, sometimes called a cumulative rate, which is higher.

None of these is wrong. Comparing a clinic's cumulative rate with another clinic's rate per transfer is. Our article on IVF success rates shows the UK figures by denominator and age.

3. Relative or absolute?

"Doubles the chance" sounds dramatic. If the chance goes from 1% to 2%, it is still small. Relative effects (ratios) are useful for comparing treatments; absolute effects (percentage points) tell a reader what changes for them. Good systematic reviews give both: for example, they translate a pooled effect into "for a typical patient with a 31% chance, the chance with the new method would be between X% and Y%". If an article gives only the relative figure, ask what the baseline was.

4. Who was studied?

A treatment tested on women under 35 with one specific diagnosis may not work the same way for women over 40 with another. Check the age range, the diagnosis, the country and whether the study took place in one clinic or many. The 2017 study on AMH tests is a good example: its findings apply to women aged 30 to 44 without known infertility, which is exactly the group being sold "fertility tests" (see what an AMH test can and cannot tell you).

5. What kind of study is it?

• Laboratory and animal studies explain mechanisms and generate ideas. They rarely tell us what will happen in patients (see what animal research taught us about reproduction).
• Observational studies follow groups of people and look for associations. They can reveal patterns but struggle to separate cause from coincidence. Research on housing costs and birth rates is a clear case: studies find consistent associations, and the best ones use clever designs to get closer to cause, as our article on housing and fertility explains.
• Randomised controlled trials assign treatments by chance, which removes most other explanations. They are the standard for judging whether a treatment works.
• Systematic reviews and meta-analyses combine all the trials on a question. They are only as good as the studies they include.

6. One study is rarely the answer

Small early studies often find large effects that shrink or disappear when larger studies follow. That pattern is one reason the UK regulator rates most IVF add-ons as unproven or ineffective despite promising early results (see IVF add-ons: what the evidence shows).

A short checklist

1. Is the outcome a live birth?
2. What is the denominator?
3. Is the effect given in absolute terms?
4. Does the study population look like the person the claim is aimed at?
5. Is it a randomised trial or a review of trials?
6. Has it been repeated?

The original papers are usually reachable through their DOI. Our short guide to DOIs and the reference archive explains how to find them.

Red flags in health news

• "Miracle", "breakthrough", "cure": these words rarely appear in the studies themselves.
• No link to the original study, or a link to a press release only.
• A study in mice or cells reported as if it concerned people.
• A conference abstract presented as a finished result: many abstracts are never published in full.
• A product for sale attached to the article, whether a supplement, a test or a clinic package.

Where to find reliable summaries

National bodies are often the best starting point. In the UK, the HFEA publishes national fertility data and rates treatment add-ons; NICE publishes clinical guidelines that weigh the evidence; the NHS website summarises guidance for patients. Cochrane reviews include plain-language summaries of the evidence on many treatments. International organisations such as the WHO publish fact sheets on conditions such as infertility, endometriosis and polycystic ovary syndrome.

None of these sources is infallible, and guidance changes as evidence accumulates. The habit that matters most is asking where a claim comes from and how strong the evidence behind it is.

A worked example

In 2011, a study of 100 women suggested that low levels of anti-Müllerian hormone were associated with lower natural fertility. In 2017, a much larger study of 750 women aged 30 to 44, designed specifically to answer the question, found no such association. The second study was larger, prospective and focused on the population most likely to use the test, and its conclusion has shaped how the test is viewed. The lesson is general: a single small study is a reason to look further, not a basis for decisions.