A superiority trial is a clinical trial designed to demonstrate that a new treatment is statistically and clinically superior to a control treatment. Evidence is provided via a pre-defined primary endpoint and a hypothesis aimed at a positive difference in favour of the investigational treatment.
Core concept: hypotheses, endpoint and effect size
In superiority trials, the null hypothesis is typically formulated as there being no difference between the groups. The alternative is that the investigational treatment is better, for example through a higher response rate, longer progression-free survival, or a lower event rate. Which effect size is considered relevant must be clinically justified, for instance via a minimal clinically important difference and historical data.
The choice of endpoint and analysis population is critical. Intention-to-treat analyses are often preferred because they respect randomisation. In addition, per-protocol analyses can help assess robustness without replacing the primary analysis. It should also be pre-defined how missing values and intercurrent events (e.g., treatment discontinuation, rescue medication) will be handled.
Study design and comparator arms
Superiority trials can be conducted against placebo or an active control. A placebo control arm is only ethically justifiable if no effective standard therapy exists or if an add-on design is used in which all participants receive standard therapy and additionally receive either the investigational product or placebo.
Randomisation is intended to balance known and unknown confounders. Blinding reduces bias in endpoint assessment and safety reporting. For objective endpoints, an open-label study may be feasible; for subjective endpoints, however, the risk of expectancy effects increases. In some indications, blinded endpoint adjudication is used to increase objectivity.
With an active control, the comparator therapy should be selected to reflect current clinical practice. Otherwise, a positive study outcome may be statistically correct but considered of limited value from a regulatory perspective. The rationale for the control arm is therefore a recurring topic in scientific advice and later marketing authorisation dossiers.
Statistical planning: power, alpha and multiplicity
Sample size planning is based on the expected effect size, variance and the desired statistical power. Typical values are 80% or 90% power at a 5% significance level. With multiple endpoints or interim analyses, control of the alpha error must be considered, for example via hierarchies or alpha spending.
A common practical issue is handling missing data and protocol deviations. Sensitivity analyses, pre-defined imputation rules and consistent data management processes are important to avoid “creating” or losing superiority through methodological artefacts. For time-to-event endpoints, attention is also paid to censoring and follow-up, as differences in follow-up duration can bias effect estimates.
Especially in multicentre studies, centre effects may also occur, for example due to differences in recruitment or standard of care. Therefore, stratification factors and central monitoring approaches are often used to strengthen data quality and consistency.
Interpretation and clinical relevance
A statistically significant difference is not automatically clinically relevant. For benefit–risk assessment, effect sizes, confidence intervals and patient-relevant thresholds are considered together. Particularly in large studies, small effects can become significant without changing clinical practice.
Conversely, a study may show clinically relevant trends that are not statistically significant, for example with rare events or too short a study duration. Therefore, transparent presentation of uncertainty is essential, including sensitivity analyses, subgroup analyses (with caution) and consistent presentation of safety data.
From a regulatory perspective, it is also important that conclusions are consistent with the pre-specified analysis plan. Post-hoc changes to endpoint definitions or analysis rules can undermine credibility and lead to major objections.
Relevance for clinical trials
From an operational perspective, superiority trials require a precise definition of the primary endpoint, a robust monitoring concept and a stable data basis for endpoint evaluation. For sponsors, consistent argumentation across the protocol, SAP and clinical study report is crucial, as deviations can weaken interpretability.
When working with a CRO, it is important that recruitment, data quality, query management and timelines are managed so that the primary endpoint can be analysed without delays and with minimal missing-data rates. Mediconomics supports such studies, among other things, with project management, data management and regulatory documentation.
For stakeholder communications (investors, partners, authorities), it is helpful to define clear narratives already during the study: What does a positive finding mean clinically, and how will ambiguous results be handled? This preparation later reduces the risk of exaggerated statements in press releases or submission documents.
Frequently Asked Questions (FAQ)
When is a superiority trial preferable to a non-inferiority trial?
If a true added benefit is expected and can be measured clinically, a superiority trial is the obvious choice. Non-inferiority is more likely to be chosen when the new therapy primarily offers advantages in safety or usability.
Can a superiority trial also be conducted against an active control?
Yes, particularly when an established standard therapy exists. In that case, the choice of comparator arm and the justification of the clinically relevant effect size are especially important.
What are common reasons why superiority is not demonstrated?
Overly optimistic effect assumptions, unexpectedly high control rates, insufficient adherence, protocol deviations or missing data can reduce power. A suboptimal endpoint definition or follow-up that is too short can also mean that a real effect is not captured.
Regulatory References
- ICH E9 (R1): Statistical principles, hypothesis testing, sample size planning and sensitivity analyses.
- ICH E6 (R3): Requirements for study conduct, data integrity and traceability.
- EU Regulation 536/2014 (Clinical Trials Regulation): Framework for the authorisation and conduct of clinical trials in the EU.