A placebo arm looks ethically unavailable, so the team defaults to an active-control, non-inferiority design and treats the choice as settled. ICH E10 — Choice of Control Group and Related Issues in Clinical Trials, reaching Step 4 on July 20, 2000 and adopted as FDA guidance the following year — does not let that reasoning stand on its own. An active-control design only answers the question it is asked to answer if assay sensitivity can be established: historical proof that the comparator reliably beats placebo under conditions like the ones in your planned trial. Skip that case, and a non-inferiority finding cannot tell a reviewer whether the new treatment worked or whether the trial simply couldn't detect that neither treatment did.

What ICH E10 actually requires

E10 catalogs the principal control-group types — placebo, dose-response, active-control superiority, active-control non-inferiority, and external or historical control — and is explicit that no type is uniformly preferred. The choice is a scientific decision tied to what the trial needs to prove, made as part of broader global regulatory strategy rather than defaulted to. Placebo-controlled superiority trials are, in E10's own framing, the most efficient at establishing that a treatment has any effect at all; active-control non-inferiority trials are harder to interpret because a 'no difference' result is ambiguous by construction. That ambiguity is exactly what assay sensitivity is meant to resolve.

Jul 20, 2000
ICH E10 reached Step 4 — the final, harmonized text sponsors and reviewers both work from.
Assay sensitivity
The historical evidence requirement that has to hold before a non-inferiority result means anything.
May 2001
FDA adopted E10 as final guidance for industry, where it remains in effect.

Why ‘no placebo available’ isn’t the whole argument

Ethical unavailability of a placebo is a real and legitimate design constraint — E10 says so directly. What it does not say is that the constraint, by itself, licenses an active-control design. The sponsor still has to show the historical trials establishing the comparator's effect are close enough to the planned trial that assay sensitivity can be assumed to carry over. Several things commonly break that assumption:

  • Different populations or dose. The historical trials that proved the comparator beats placebo enrolled a different population, used a different dose, or ran in a different disease stage than the planned trial.
  • Endpoint drift. The historical effect was shown on an endpoint that differs from the one the new trial will use, even if both are plausible measures of the same underlying benefit.
  • Standard-of-care creep. Background treatment has improved since the historical trials ran, shrinking the comparator's true advantage over placebo in a way the old margin no longer reflects — the ‘biocreep’ problem E10 flags directly.
  • Thin historical evidence. A single, older, or inconsistently replicated placebo-controlled trial is a weaker foundation for assay sensitivity than a consistent body of evidence across settings.
A non-inferiority margin borrowed from a trial run under different conditions isn’t a conservative choice. It’s an untested one. Why the historical match matters as much as the number

Setting and defending the margin

The margin itself has to come from the same historical evidence that establishes assay sensitivity — conventionally a fraction of the comparator's own effect size over placebo, conservative enough to preserve a meaningful amount of that effect even in a worst-case reading of the new trial's result. E10 expects this margin pre-specified in the protocol and justified against the cited trials, not fitted to the data afterward. How that margin interacts with the treatment-effect question the trial is actually asking is also where the estimand framework and the control-group choice meet — a margin defined against the wrong estimand answers a question nobody asked. Sponsors who bring this case to an FDA meeting before the trial starts get a design objection on the record while it is still a protocol amendment; sponsors who don’t get it as a complete response letter.

A non-inferiority case worth building before you meet
  1. Establish assay sensitivity first. Assemble the historical, placebo-controlled evidence under comparable trial conditions before assuming a non-inferiority design is viable.
  2. Derive and justify the margin. Set it from that historical effect size and pre-specify it in the protocol.
  3. Match the comparator and conditions. Confirm dose, population, and endpoints mirror the trials that established the comparator's effect.
  4. Raise the design at the meeting, not after. Bring the case to End-of-Phase 2 or scientific advice, where it is still a protocol amendment.

None of this is a reason to avoid non-inferiority designs where a placebo genuinely cannot be used — E10 was written to make that route workable, not to close it off. It is a reason to treat the assay-sensitivity case as the actual design work, built on the same rigor as any other part of a pharmaceutical development program, rather than as a formality that follows automatically from an ethics argument.

Frequently asked questions

When does ICH E10 allow a non-inferiority design instead of a placebo-controlled trial?

ICH E10 allows it when assay sensitivity can be established — historical evidence that the active comparator, in trials conducted under similar conditions, has reliably and consistently outperformed placebo. Without that evidence, a non-inferiority finding cannot distinguish 'the new treatment works' from 'neither treatment worked and the trial could not tell the difference.'

Does an unavailable ethical placebo automatically justify an active-control design?

No. ICH E10 treats the ethical unavailability of a placebo as a reason to consider an alternative design, not as evidence that an active-control, non-inferiority design will answer the scientific question. Sponsors still have to make the assay-sensitivity case and justify the non-inferiority margin against historical data.

How is the non-inferiority margin set under ICH E10?

The margin is derived from the historical, placebo-controlled evidence for the active comparator — typically a fraction of its established effect size over placebo — and must be pre-specified and justified in the protocol, not chosen after the fact to fit the trial's results.

Sources & further reading

  1. ICH. E10: Choice of Control Group and Related Issues in Clinical Trials (Step 4, 20 July 2000). ich.org
  2. FDA. E10 Choice of Control Group and Related Issues in Clinical Trials — guidance for industry (May 2001). fda.gov
  3. EMA. ICH E10 — Choice of control group in clinical trials — scientific guideline. ema.europa.eu

This article is provided for general informational purposes and reflects the regulatory landscape as of November 2026. It is not legal or regulatory advice. Confirm current ICH E10 expectations with FDA, EMA, or qualified counsel before finalizing a trial design.