IVF Add-Ons: What the Randomised Trials Actually Found

That the add-ons tested in adequately powered trials have not improved live birth. This is not a selective reading — it is what the largest trial of each found, and the consistency across quite different interventions is the striking part.

• Endometrial scratching — 26.1% vs 26.1% live birth.
• Time-lapse imaging — 33.7% vs 33.0% live birth.
• PGT-A as routine screening — value not demonstrated.
• Immune testing — specifically not recommended.

Each was introduced with a plausible mechanism and encouraging early data. Each was in routine paid use before an adequately powered trial existed.

Live birth was 26.1% in both the scratch and control groups of a 1,364-woman randomised trial, with an adjusted odds ratio of 1.00 (95% CI 0.78–1.27). No significant differences in ongoing pregnancy, clinical pregnancy, miscarriage or ectopic pregnancy.

Covered in detail in our review of that trial.

The TILT trial randomised 1,575 patients into three arms and found live birth of 33.7% with time-lapse imaging against 33.0% with standard care. The undisturbed culture arm reached 36.6% — suggesting any benefit lies in leaving embryos alone rather than in the imaging.

Covered in our review of TILT.

ASRM's 2024 committee opinion states that the value of PGT-A as a routine screening test for all patients undergoing IVF has not been demonstrated, and that recent multicentre trials found similar pregnancy outcomes via frozen transfer with and without it.

On immune testing, ESHRE's 2023 recommendations are unusually direct: peripheral NK cell testing is not recommended, uterine NK cell testing is not recommended, and blood cytokine assessment is not recommended. On endometrial receptivity testing, there are insufficient data to recommend routine use.

• PGT-A as routine screening — value not demonstrated (ASRM 2024).
• Peripheral and uterine NK cell testing — not recommended (ESHRE 2023).
• Blood cytokine assessment — not recommended.
• Endometrial receptivity testing — insufficient data for routine use.

More detail in our article on PGT-A and recurrent implantation failure.

Because the incentives favour early adoption and disfavour testing. An add-on can be sold as soon as it is plausible; the trial that would settle it is expensive, slow, and commercially unattractive to whoever is already selling the thing.

• A plausible mechanism is enough to start offering it.
• Small early trials produce unstable, often positive, results.
• Positive small studies are more likely to be published and cited.
• By the time a large trial reports, the practice is established.

None of this requires anyone to act in bad faith. It is what happens when a treatment can enter routine practice without first having to prove it works.

Fertility treatment is unusually exposed to this. Patients are motivated, often paying privately, and frequently on a limited number of attempts — a combination that makes 'anything that might help' a persuasive offer even when the evidence behind it is thin.

A randomised trial, large enough to detect a difference that matters, measuring live birth, with the result published whichever way it fell. That is the standard the add-ons above were eventually held to, and none met it.

• Randomised, so the groups are comparable.
• Adequately powered, so a null result means something.
• Live birth as the outcome, not clinical pregnancy.
• Pre-registered and published regardless of result.

It is worth knowing that this standard is achievable — TILT and the scratching trial both met it. The problem is not that fertility questions cannot be studied properly; it is that the studies tend to arrive years after the treatment is being sold.

Where an add-on has no trial of this kind, the honest position is that nobody knows whether it works. That is different from knowing it does not, and both are different from the confidence with which such things are usually offered.

Three questions handle most of the menu, and a clinic that answers them clearly is telling you something useful either way.

• What is the largest randomised trial of this, and what did it find?
• Is the outcome live birth, or something easier like clinical pregnancy?
• What would change in my treatment if I declined this?
• Is it charged separately, and how much?

Our article on choosing a clinic covers what else is worth verifying, and the cost article covers where add-ons sit in what you are quoted.