Table 1.
Strengths and limitations of trial endpoints.
| Endpoint | Strengths | Limitations |
|---|---|---|
| Singular | ||
| Clinically observed | Routinely collected information Typically well-accepted approach by scientific community |
Doesn't consider that an intervention may impact on the patient in different ways May need supportive secondary analyses to be persuasive |
| Surrogates | Reduction in sample size Shorter trial duration Decreased cost of trial Accelerated approval/dissemination of effective therapies |
May fail to predict clinically meaningful endpoints May not be sensitive to change at all stages of disease Validation process often challenging Therapeutic advances may alter the validity of the surrogate measure Cost Reproducibility may be problematic Utility limited to early phase trials (1/2) |
| Multiple or combination | ||
| Multiple primary | Useful if more than one important outcome exists & demonstration of 1 is enough to support clinical efficacy | Adjustment for Type 1 error is required. Hard to interpret if results occur in different directions |
| Co-primary | Useful if demonstration of two or more outcomes is necessary to establish clinical benefit | Adjustment for Type 2 error is required |
| Composites | Improves statistical efficiency and precision. Increases power (reduces sample size requirement). Ability to measure small effects. Lower cost. Earlier trial completion |
Implementation may be complex and resource-intensive. Components may be inappropriately combined or reported. May be difficult to interpret study findings and determine which of the component endpoints are impacted by the intervention; the effect is often smallest for the most important component and biggest for the less important components. Prone to post-hoc analyses/bias. Key data often missing or unclear. Can lose meaning if components of composite move in opposite directions |
| Multi-component endpoints | Allows single evaluation of numerous components without creating multiplicity issues | Individual components may not have clear meaning. If components aren't concordant, study power may be compromised |
| Weighted endpoints | More complex/robust evaluation of the effectiveness of treatment intervention(s) that considers the relative importance of components | Process of assigning weights not standardised, and can involve lengthy processes. May be costly |
| Endpoints that are participant specific | Best reflects clinical decision-making. Theoretically would represent the gold-standard for informing personalised, evidence-based medicine. May result in increased power to detect real treatment effects for patients |
Complex; logistically difficult. Generalisability of trial results may be limited. Requires large data capture |