Research
Human trials
How clinical trials are designed and phased, and how to read a trial result from a compound page without over- or under-reading it.
- Last updated:
- 2026-07-22
Overview
Human clinical trials are structured investigations of a candidate intervention in defined human populations. They are usually staged: phase 1 (safety, small healthy volunteers), phase 2 (efficacy signal, small patient populations), phase 3 (efficacy and safety at scale, large populations), and post-marketing surveillance after approval.
The most informative single trial design is a randomised, double-blind, placebo-controlled trial with pre-specified endpoints and adequate power. Deviations from that design — open-label, non-randomised, retrospective — reduce the strength of the conclusion, not always to zero.
Why it matters
The evidence sections on compound pages cite specific human trials. Understanding what those trials do and do not establish keeps readers from over- or under-interpreting the summary.
Key concepts
Primary vs. secondary endpoints
A trial is designed to answer a primary question with pre-specified statistical power. Secondary endpoints are hypothesis-generating; a positive secondary endpoint in a trial that missed its primary is not the same as a positive result.
Effect size and confidence intervals
A trial can be statistically significant with a small effect that is not clinically meaningful, and clinically meaningful with a wide confidence interval that leaves the true effect uncertain.
Generalisability
Trial populations differ from real-world populations. Age, comorbidity, and background medication may not match the reader.
Common misconceptions
- "A positive trial proves the drug works." A well-designed positive trial supports the conclusion; it is not a proof.
- "A phase 2 result is a phase 3 result." Many phase 2 signals fail to replicate at scale.
- "Approval means universally effective." Approval means the trial data supported a specific claim in a specific population under specific conditions.
Practical interpretation
When a compound page cites a trial, read the citation for design, size, and endpoints. Weight the conclusion in proportion to those details.
Limitations
Compound pages summarise trials in a few lines. The trials themselves are longer. Where a decision depends on a specific detail, read the underlying paper.