Non-inferiority trials: when good enough is good enough
What if a new drug is not better than the old one, but still good enough to use? Is that a failure or a success? A whole class of clinical trials exists to answer exactly this question. Non-inferiority trials help researchers decide when a treatment that is not the best is still worth using. Instead of asking whether a new drug beats the current standard, they ask a more modest question: is the new treatment at least as good as the old one within an acceptable margin? That distinction matters when the new option brings something else to the table, such as fewer side effects, lower cost, or easier dosing.
Three ways to compare treatments
Comparative trials come in three flavors. Superiority trials try to show that treatment A is better than treatment B. Non-inferiority trials try to show that A is not worse than B by more than a small, pre-specified amount. Equivalence trials try to show that A and B are essentially the same, within a margin on both sides. The margin is the key difference. Non-inferiority only looks at one side: is the new drug not too much worse? Equivalence looks at both sides: it must not be too much worse or too much better.
The delta margin
The non-inferiority margin, often called delta, is the maximum amount of effectiveness the new treatment is allowed to lose compared with the old one. Researchers choose delta before the trial starts, and it must be small enough that losing that much effectiveness is still clinically acceptable. For example, if the old drug cures 90 percent of patients and the new drug has fewer side effects, a team might accept a margin of 10 percent, meaning the new drug must cure at least 80 percent to be considered non-inferior.
Choosing the margin is one of the most debated parts of these trials. The margin must preserve a meaningful portion of the old drug's effect over placebo. Researchers look at historical trials of the active control to establish its effect size, then choose a margin that retains at least half of that effect. This is the M1 and M2 approach from the ICH E10 guideline, which covers the choice of control groups in clinical trials (ICH efficacy guidelines). M1 is the entire effect of the active control over placebo; M2 is the smaller margin used in the trial that preserves most of that effect.
Reading the confidence interval
Non-inferiority is established by looking at the confidence interval of the difference between treatments. If the entire confidence interval lies above the non-inferiority margin, non-inferiority is declared. Suppose the confidence interval for the difference runs from minus 2 percent to plus 5 percent and the margin is minus 10 percent: the new drug is non-inferior. But if the interval crosses below minus 10 percent, the trial fails to show non-inferiority. The position of the confidence interval is what matters, not just the point estimate.
When non-inferiority trials make sense
These trials are used when a placebo-controlled trial would be unethical. If an effective treatment already exists, you cannot give patients placebo. So the new drug is compared against the active standard of care (ClinicalTrials.gov on how studies work). They are also used when the new treatment offers meaningful advantages over the old one, like a better safety profile, a simpler dosing schedule, or lower cost. The goal is to show that the new treatment preserves enough of the old drug's benefit while offering those other advantages.
The challenges
Non-inferiority trials face real problems. First, assay sensitivity: can the trial actually detect a difference between treatments if one exists? Without a placebo group, you might not know if both drugs simply failed. Second, the choice of margin is sometimes criticized as too wide or too narrow. Third, these trials often need larger sample sizes than superiority trials because the margin is small. Fourth, there is the risk of "bio-creep": each generation of drugs becomes slightly less effective, and over time the standard of care quietly deteriorates.
What regulators require
Regulatory agencies are cautious with non-inferiority trials. The FDA and EMA require the margin to be justified with historical evidence, and they require proof of assay sensitivity, meaning the trial must be able to distinguish an effective treatment from an ineffective one. The FDA has published a dedicated guidance on non-inferiority clinical trials that walks through margin selection and interpretation (FDA guidance on non-inferiority clinical trials). ICH E10 is the key regulatory document; it defines when non-inferiority trials are appropriate and how the margin should be chosen (EMA ICH E10 guideline). Regulators are especially concerned about bio-creep, where successive non-inferiority trials gradually erode the effectiveness of the standard of care.
Sample size and common mistakes
Non-inferiority trials often require larger sample sizes than superiority trials because the margin is smaller than the effect size you would target in a superiority trial. The sample size calculation depends on the chosen margin, the expected event rate in the control group, and the desired statistical power. A common mistake is underestimating the required sample size and ending up with a trial too small to demonstrate non-inferiority. That is not the same as proving equivalence. An underpowered trial just fails to show anything.
Interpreting results without overclaiming
If the confidence interval lies entirely above the margin, non-inferiority is demonstrated. If it crosses the margin, non-inferiority is not established. That does not mean the new drug is inferior; it means the trial did not prove it is not worse. If the entire confidence interval sits above zero, the new drug might even be superior, but only if the trial was designed to test superiority as well. A non-inferiority trial cannot claim superiority unless that comparison was pre-specified.
Several misconceptions follow from this. Non-inferiority is not the same as equivalence, which requires the confidence interval to sit within a margin on both sides. Failing to show non-inferiority does not mean the drug is inferior; it means the evidence is not strong enough. And a wide confidence interval is not evidence of non-inferiority, it just means the result is imprecise.
Three practices that keep these trials honest
First, pre-specify everything: the margin, the analysis method, and the criteria for non-inferiority must be defined in the protocol before the trial starts. Second, ensure assay sensitivity by including a placebo group where possible or by using a sensitive patient population and endpoints validated in previous trials. Third, be transparent about the margin and its justification. Regulators and readers should understand why that particular delta was chosen and what it means for clinical decision making.
The bottom line
A non-inferiority trial is a study designed to show that a new treatment is not worse than an existing one by more than an acceptable margin. These trials are essential when placebo controls are unethical and when new treatments offer advantages in safety, convenience, or cost. The margin is the heart of it: choose it carefully, justify it with evidence, and interpret the confidence interval correctly. Non-inferiority is not about settling for less. It is about knowing when good enough is truly good enough.
This article is based on the Clarity Clinical Solutions video "Non-Inferiority Trials Explained - Superiority vs Equivalence, Delta Margin and Confidence Intervals." Watch it here: Non-Inferiority Trials Explained - Superiority vs Equivalence, Delta Margin and Confidence Intervals
References
- Clarity Clinical Solutions — "Non-Inferiority Trials Explained - Superiority vs Equivalence, Delta Margin and Confidence Intervals" (source video). https://www.youtube.com/watch?v=a7FP6eDlCi0
- ICH — Efficacy guidelines page (E10: Choice of Control Group, M1/M2 margin approach). https://www.ich.org/page/efficacy-guidelines
- FDA — Guidance: Non-Inferiority Clinical Trials to Establish Effectiveness. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/non-inferiority-clinical-trials
- EMA — ICH E10 guideline on choice of control group in clinical trials. https://www.ema.europa.eu/en/ich-e10-choice-control-group-clinical-trials-scientific-guideline
- ClinicalTrials.gov — About studies: how clinical trials work. https://clinicaltrials.gov/about-studies/learn-about-studies