A fertility trial that missed its endpoint and found something it cannot explain

Design

single-centre, open-label randomised trial, 1:1 allocation

Population

254 women undergoing artificial insemination by husband

Primary outcome

DASS-21 distress score at day 35

Effect

no difference (all P ≥ .868); clinical pregnancy 26.8% vs 11.0% as an unpowered secondary, P = .001

Two hundred and fifty-four women undergoing intrauterine insemination with partner sperm were randomised, open-label, to routine care or routine care plus two 30-minute music-listening sessions either side of the procedure. The prespecified primary outcome was the DASS-21 distress score at day 35.

It did not move. Anxiety, depression and stress scores were indistinguishable after adjustment for baseline (all P ≥ .868), with absolute differences of 0.01 point or less for anxiety and depression and 0.08 for stress. The secondary outcome did move: clinical pregnancy occurred in 34 of 127 (26.8%) with music and 14 of 127 (11.0%) without, P = .001.

The authors are right to leave it there. A trial that fails on its mechanism and succeeds on an unpowered secondary has not shown that the intervention works — it has produced a question. The control arm's 11 per cent is also low for this population, which is at least as likely an explanation as any effect of music.

• Offer periprocedural music if a woman wants it — it is harmless and free — but not as a fertility intervention
• Do not quote the pregnancy figure to patients as a benefit of music
• When a trial's mechanism fails, treat downstream findings as hypotheses
• Watch for the replication the authors call for: attention-matched control and live-birth follow-up

Why it matters

It is a clean example of why the endpoint a trial was powered for is the one that counts.

Don't overread it

The pregnancy difference is an exploratory secondary outcome the trial was not powered for.

The statistics, in plain English

P = .001 on a secondary outcome in a trial powered for something else is not the same as a positive trial. With 127 per arm, chance differences in clinical pregnancy are large, the study was open-label, and multiple secondary outcomes were examined — all of which inflate the chance of a striking number appearing somewhere.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.