What is a primary endpoint?
The primary endpoint is the main outcome a clinical trial is designed to measure. It answers the trial's central question, and the sample size is calculated so the study can detect an effect on it.
- Primary endpoint in plain language
- How it works in practice
- Related terms
Free sandbox · No credit card · 21 CFR Part 11 aligned
Primary endpoint
noun · clinical research
The pre-specified outcome that addresses the trial's primary objective and on which success is judged. Also called the primary outcome measure.
Related terms
In short
- The primary endpoint is the one outcome the trial is built to answer. It is chosen before the study starts and written into the protocol.
- It must be specific: what is measured, when, how it is summarised, and in whom. "Improvement in mood" is a topic; "change from baseline in PHQ-9 total score at week 12" is an endpoint.
- The sample size is calculated for the primary endpoint, so changing it later invalidates the plan.
- Most trials have one primary endpoint. Several are possible, but then the analysis plan must say how the overall false-positive risk is controlled.
- Registries call it the primary outcome measure; ICH E9(R1) adds the estimand, which states exactly what effect the endpoint estimates.
Explained
What a primary endpoint has to say
A trial can measure dozens of things. The primary endpoint is the one that decides whether the study met its main objective. Because the trial is judged on it, it is fixed in the protocol before the first participant is enrolled and cannot quietly change once results start arriving. Switching the primary endpoint after seeing data is one of the best-known ways a trial can mislead, which is why registries and journals check the registered primary outcome against the published one.
A usable definition has five parts. The variable (for example, systolic blood pressure). The time point (week 12). The measurement method (seated, mean of two readings, same device). The summary (change from baseline, or proportion with a reduction of at least a stated amount). And the analysis population (everyone randomized, or only those who completed treatment). Leave any of them open and two statisticians will analyse the same data differently. For the broader family of endpoint types, see clinical trial endpoints.
Why it drives sample size
The number of participants needed depends on how large an effect on the primary endpoint matters, how variable the measurement is, and the chosen error rates. A noisier endpoint needs more people. A binary endpoint needs a different calculation from a continuous one. That calculation belongs to the sample size calculation, and you can try one in the sample size calculator.
One primary endpoint or several
A single primary endpoint keeps interpretation clean. Some designs have two co-primary endpoints that must both succeed, or two alternative primaries where success on either counts. Each approach changes the statistical rules: with alternatives, the plan must adjust for multiple testing; with co-primaries, power for the pair is lower than for either alone. The choice and the rule belong in the statistical analysis plan.
The estimand
ICH E9(R1) introduced the estimand framework to state precisely what treatment effect is being estimated, including how events after randomization, such as stopping the study drug or taking rescue medication, are handled. The primary endpoint is one part of that statement. Check the current ICH text when writing it.
Examples
From vague aim to buildable primary endpoint (demo wording)
| Study aim | Vague wording | Specific primary endpoint |
|---|---|---|
| Hypertension | Lower blood pressure | Change from baseline in mean seated systolic blood pressure at week 12 |
| Depression | Improve mood | Change from baseline in PHQ-9 total score at week 12 |
| Vaccine | Protect against infection | Proportion with a protective antibody titre 28 days after the final dose |
| Oncology | Longer survival | Time from randomization to death from any cause |
| Pain | Reduce pain | Change from baseline in weekly mean of daily pain scores (0 to 10 scale) at week 8 |
Illustrative demo wording only; real endpoints come from your protocol and statistician.
Make sure your form collects what the endpoint needs
Draft visits and forms from a protocol in the free sandbox and check that the primary endpoint data is captured at the right time points. No credit card.
Practice
Building the primary endpoint into data capture
Once the endpoint is written, the data collection follows from it. Every input to the endpoint must be collected at the right visit, within a defined visit window, in a form that cannot be left incomplete. If the endpoint is a score, the item answers and the total need to be traceable. If it depends on a calculated value, build the calculation into the form instead of leaving it to a spreadsheet. If it comes from a patient diary or scale, the questionnaire timing matters as much as the content.
It is worth checking the endpoint against the schedule before go-live: is there a baseline measurement, a time point for the final value, and a plan for missing assessments? The protocol template pairs objectives with endpoints, and protocol to study setup shows how a protocol turns into visits and forms. A primary endpoint measured on a questionnaire belongs in ePRO, where collection can be time-stamped.
Endpoint
PHQ-9 change, wk 12
Checks
4 of 5 ready
Status
Review
- Baseline and week 12 visits in the scheduleVisit window set
- All 9 items required on both formsRequired fields
- Total score calculated, not typedCalculated field
- Out-of-range answers raise a queryEdit check
- Plan for missed week 12 visit written downNeeds the analysis plan
In Capture
Endpoint data collected the way the protocol says
In Capture, upload a protocol and let the AI study builder draft the visit schedule and forms, then review and edit them; nothing is saved without human review. Add required fields, edit checks and calculated fields so endpoint inputs are complete and consistent, set visit windows, and export data with an automatic data dictionary for the analysis.
- AI drafts visits and forms from a protocol, with human review.
- Calculated fields shown read-only during entry.
- CSV and Excel exports with data dictionary.
What is a primary endpoint?
The main outcome a clinical trial is designed to measure. It addresses the primary objective, is fixed in advance in the protocol, and determines the sample size.
What is the difference between a primary and a secondary endpoint?
The primary endpoint answers the main question and powers the trial. Secondary endpoints give supporting evidence or answer additional questions, often with less statistical certainty.
Can a trial have more than one primary endpoint?
Yes, but the analysis plan must state how success is defined and how the overall chance of a false positive is controlled across them.
Can the primary endpoint be changed during a trial?
Only through a documented protocol amendment, and changing it after seeing outcome data casts serious doubt on the results. Trial registries expect the registered primary outcome to match what is reported.
Is primary endpoint the same as primary outcome?
In practice yes. Trial registries use the term primary outcome measure, while protocols and regulators often say primary endpoint.
What is a surrogate primary endpoint?
A measure, such as a lab value, used in place of a clinical outcome because it is expected to predict clinical benefit. Whether a surrogate is acceptable depends on the disease and the regulator.
Build the primary endpoint into the study
Draft visits and forms from your protocol in the free sandbox. No credit card.