On this page
Reading a clinical trial's primary endpoint means finding the exact outcome the trial was designed to measure in the protocol or statistical analysis plan, checking whether it achieved statistical significance at the pre-specified threshold, and understanding what that result actually demonstrates about the drug or device, separate from secondary endpoints or subgroup findings.
- The primary endpoint is pre-specified before data analysis begins and is listed in the trial registration and statistical analysis plan, not chosen after results are known.
- An endpoint definition includes the patient population it applies to, the specific outcome measured, and the exact timepoint, results at other timepoints or in other populations are secondary.
- Statistical significance at p < 0.05 means the result is unlikely to be due to chance, but a statistically significant result can still be clinically small or irrelevant.
- A trial that misses its primary endpoint but emphasizes a secondary endpoint result is a sign of selective reporting, not a successful trial.
- Surrogate endpoints like tumor shrinkage measure a proxy for clinical benefit, not clinical benefit itself, and do not prove a drug improves how patients feel, function, or survive.
Clinical trial results appear in press releases, SEC filings, regulatory documents, and published papers. A company or researcher will often highlight a finding and call it successful. But the finding that gets highlighted may not be the endpoint the trial was designed to measure. Learning to locate the primary endpoint, read its exact definition, verify it was met, and distinguish it from secondary results is the core skill for assessing what a trial actually showed.
This article teaches you where to find the primary endpoint in trial documents, how to read its definition precisely, how to check whether it achieved statistical significance, and how to spot when results are being cherry-picked from secondary endpoints or subgroups. You will be able to move past press releases and read the trial on its own terms.
What you'll be able to do
By the end of this article, you will be able to locate the primary endpoint in a trial protocol or results document, understand exactly what outcome was measured and in which patients, verify whether the trial met its statistical threshold for success, and recognize when a company or media report is emphasizing secondary endpoints instead of the primary one. You will know how to spot a trial that succeeded on its primary measure versus one where results were selected from secondary findings.
Locate the primary endpoint in the trial protocol or results document
The primary endpoint is stated in the trial protocol before patients are enrolled and before any data analysis begins. This prevents researchers from changing the definition of success after seeing the results. You need to find this pre-specified definition, not rely on a company's summary of what the trial showed.
Check the trial registration first
Start at ClinicalTrials.gov (for trials in the United States) or EudraCT (for trials in the European Union). Search for the trial by name or identifier. On the trial's registration page, look for the section labeled "Primary Outcome Measures" or "Primary Endpoint." This entry was filed before results were known and cannot be changed after the fact. It is the official, pre-specified definition of what the trial was designed to measure.
If you are reading a published paper or a company press release, cross-check the primary endpoint stated there against the trial registration. If they differ, the trial definition changed, which is a red flag.
Find the statistical analysis plan
The statistical analysis plan (SAP) is the document that specifies exactly how the primary endpoint will be analyzed: which patients are included, which statistical test will be used, and what p-value threshold counts as success. The SAP is written before the data are analyzed and locked before anyone looks at results.
For published trials, the SAP is often referenced in the methods section of the paper or available as a supplement. For regulatory submissions (FDA approvals, for example), the SAP is part of the clinical trial dossier. If you are reading a company announcement or press release, ask whether the SAP is available. If it is not, you cannot verify that the statistical test was pre-specified rather than chosen after seeing the data.
Read the endpoint definition exactly as written
The primary endpoint definition is precise. It specifies which patients it applies to, what outcome is being measured, and when it is measured. Read it word for word. Vague language like "improvement" or "benefit" is not an endpoint; endpoints measure specific, quantifiable outcomes.
Identify the patient population
Endpoints often apply only to a subset of the patients enrolled in the trial. The definition will state which patients are included in the primary analysis. For example, an endpoint might apply only to patients with a certain baseline disease severity, or only to patients who completed the full treatment course, or only to patients who did not require a rescue medication.
This matters because it affects how broadly the result applies. If the primary endpoint is measured only in patients with moderate disease, you cannot assume the drug works the same way in patients with severe disease. Note the inclusion and exclusion criteria for the primary endpoint population. If a press release emphasizes results in a different population (e.g., "patients over 65"), check whether that population was part of the primary endpoint definition or whether it is a secondary subgroup analysis.
Understand the outcome measure
The outcome measure is the specific thing being measured. Examples include time to disease progression (how long until the disease worsens), change in blood pressure from baseline, proportion of patients who achieve a certain response, or survival time.
The definition will specify the exact metric. For example, "time to disease progression" is different from "proportion of patients with no disease progression at 12 weeks." The first is a time-to-event measure; the second is a binary measure at a fixed point. The statistical test used and the way results are interpreted depend on which type of measure it is.
If the definition uses a scale or score (e.g., a pain scale, a cognitive test), note the full range and what direction counts as improvement. A 10-point improvement on a 100-point scale is different from a 10-point improvement on a 20-point scale.
Note the timepoint
The primary endpoint is measured at a specific time. This might be 12 weeks, 6 months, 2 years, or at the end of treatment. Results at other timepoints are secondary. For example, if the primary endpoint is measured at 12 weeks, a result at 8 weeks or 16 weeks is not the primary endpoint, even if it looks favorable.
Press releases sometimes lead with results at a timepoint that was not the primary endpoint. Check the definition to confirm which timepoint is primary. If the company is emphasizing a different timepoint, that is a signal to look more carefully at whether the primary endpoint was met.
Check the statistical test and success threshold
The primary endpoint is met only if the trial achieves statistical significance at a pre-specified threshold. This threshold is almost always a p-value of 0.05 or lower. You need to find the p-value (or confidence interval) reported for the primary endpoint and verify it meets this threshold.
Locate the p-value or confidence interval for the primary endpoint
In published papers, the p-value or confidence interval for the primary endpoint is usually in the results section, often in a table. Look for the row labeled with the primary endpoint name and find the column labeled "p-value" or "p." If the p-value indicates statistical significance, the result met its primary endpoint. If the p-value does not indicate statistical significance, the trial did not meet its primary endpoint.
Confidence intervals work the same way. A confidence interval that does not cross zero (for a difference measure) or does not cross one (for a ratio measure) indicates statistical significance.
In press releases, companies often state the p-value explicitly. If the p-value is not stated, ask for it. A company that will not provide the p-value for the primary endpoint is hiding something.
Verify the test was pre-specified
The statistical test (t-test, chi-square test, log-rank test, etc.) should have been chosen before the data were analyzed, not selected after seeing the results. This is stated in the statistical analysis plan. The SAP will specify which test applies to which endpoint.
If you see a phrase like "exploratory analysis" or "post-hoc analysis" applied to the primary endpoint, that is a warning sign. The primary endpoint should never be exploratory. If the statistical test for the primary endpoint is not mentioned in the SAP, or if the SAP was written after the trial ended, the test was likely chosen after seeing the data, which inflates the false positive rate.
Distinguish the primary endpoint from secondary endpoints
A trial has one primary endpoint and usually several secondary endpoints. The primary endpoint is what the trial was powered to detect. Secondary endpoints are additional measures that may provide supporting evidence but were not the main focus.
If a trial misses its primary endpoint but succeeds on a secondary endpoint, that is not a successful trial. It is selective reporting. A press release that leads with a secondary endpoint result while burying or omitting the primary endpoint result is a red flag.
Interpret what the result means for the drug or device
A successful primary endpoint means the trial showed a statistically significant effect on the outcome it was designed to measure. But statistical significance does not automatically mean the effect is large, important, or clinically meaningful. You need to assess the magnitude of the effect and understand what the endpoint does and does not prove.
Assess the magnitude of the effect
A result can be statistically significant and still be clinically small. Suppose a trial measures blood pressure reduction and finds that the treatment group's blood pressure dropped 2 millimeters of mercury more than the control group, with statistical significance. The result is statistically significant, but 2 mmHg is a tiny difference and unlikely to matter for patients.
Look at the actual size of the effect, not just the p-value. The effect size is usually reported as a difference, a percentage change, or a ratio. Ask: Is this difference large enough to matter? How does it compare to existing treatments? A methodological study of 500 trials published between 2018 and 2022 found that 10.3% of published trials were statistically significant but not clinically important. Statistical significance alone is not enough.
Recognize what the endpoint does not measure
A successful primary endpoint on one outcome does not prove benefit on other outcomes. If the primary endpoint is tumor shrinkage, a positive result shows the drug shrinks tumors. It does not prove the drug extends survival or improves quality of life. If the primary endpoint is a biomarker (like a blood test), a positive result shows the drug changes that biomarker. It does not prove the change translates to clinical benefit.
Surrogate endpoints are particularly important to understand. A surrogate endpoint is a proxy for a clinical outcome, measured because the clinical outcome would take too long to study. For example, a reduction in blood pressure is a surrogate for stroke prevention; a drop in cholesterol is a surrogate for heart attack prevention. The FDA approved 45% of new drugs between 2010 and 2012 based on a surrogate endpoint. Surrogate endpoints are useful, but they are not the same as clinical outcomes. A drug that improves a surrogate endpoint may or may not improve the clinical outcome. If the primary endpoint is a surrogate, look for evidence that improving it actually leads to clinical benefit.
Spot common misrepresentations of endpoint results
Companies and media outlets sometimes present trial results in ways that obscure whether the primary endpoint was met. Learn to recognize these patterns.
Watch for emphasis on secondary endpoints
If a press release leads with a secondary endpoint result, the primary endpoint likely missed or was weak. Compare the headline and opening paragraph against the actual primary endpoint. If they do not match, read further to find the primary result.
For example, a trial might have a primary endpoint of survival and secondary endpoints of tumor response and symptom improvement. If the press release leads with a favorable tumor response result but buries the survival result deep in the text, the survival result was probably not favorable. Ask: What was the primary endpoint? Did it succeed? If the answer is not in the first paragraph, the company is not leading with good news.
Check for subgroup analysis claims
A subgroup is a smaller group within the trial population, defined by a characteristic like age, disease severity, or genetic marker. Results in a subgroup are exploratory unless the subgroup was pre-specified in the statistical analysis plan as part of the primary analysis.
Suppose a trial's primary endpoint was measured in all enrolled patients and the result just missed statistical significance. The company then reports that in patients over 65 (a subgroup), the result was statistically significant. That subgroup result is exploratory. It was not pre-specified, and it was likely identified after seeing the overall result. Testing many subgroups increases the chance of finding one that looks good by random variation alone.
If a press release emphasizes a subgroup result, check the statistical analysis plan to see whether that subgroup was pre-specified as a primary analysis. If it was not, the result is exploratory and should not be treated as evidence the drug works in that population. The PROSERA trial of seralutinib in pulmonary arterial hypertension illustrates this: the overall trial missed the primary endpoint at the pre-specified alpha threshold, but the company reported that a prespecified intermediate and high-risk subgroup showed improvement. The subgroup was pre-specified, so it is more credible than a post-hoc subgroup, but the overall primary result still missed.
Do this automatically
Manually locating primary endpoints across trial registries, protocols, and statistical analysis plans requires hours of document review per trial, multiplied across the dozens or hundreds of trials a researcher or analyst might need to evaluate in a year. Software can extract and standardize these endpoint definitions automatically, reducing the time spent on document retrieval and interpretation. Healthcare research reads ClinicalTrials.gov records, regulatory filings, and published literature directly, and cites each claim to the record it came from, across as many subjects as you need, with Mimir attaching a citation to every figure. It will not predict a trial's outcome or advise on anyone's treatment. Those decisions stay with clinicians and regulators, and deliberately so. Every plan includes it, including the free one.
Start your first report and run it on something you already know well.
This article is general information about healthcare and life-sciences topics. It is not medical advice, diagnosis, or treatment guidance, and it is not a substitute for a qualified professional. Any treatment decision should be discussed with a clinician.
