Skip to main content

Welcome to a Grasp lesson

Create a free account to keep studying and create your own study path.

Lesson 8

Interpreting Clinical-Trial Results: Absolute, Relative, and Confidence-Interval Effects

Recommended resources

Welcome to the clinical-evidence module. The course now shifts from how GLP-1 medicines act in the body to what trials show about outcomes that matter to patients. This lesson gives you a way to read a result without letting a striking percentage obscure its size or its uncertainty.

You will work with one cardiovascular trial of semaglutide. By the end, you should be able to describe how often an event occurred in each group, how large the difference was in absolute and relative terms, and what the confidence interval does—and does not—tell you. This is a method for interpreting evidence, not for choosing a prescription.


Start with the event, the groups, and the time period

A trial percentage has meaning only when attached to an outcome and a population. An event rate of 8%, for example, might refer to strokes, hospital visits, or a combination of several events; it might have accumulated over months or years. Before interpreting a treatment effect, establish what was counted and in whom.

The short video below introduces the two ways of expressing a difference that you will use throughout the lesson.

NCCMT - URE - What’s the Risk? Understanding Absolute and Relative Risk Reduction

Watch “What’s the Risk? Understanding Absolute and Relative Risk Reduction” from NCCMT | CCNMO for its worked example of event rates in treatment and comparison groups.

Watch absolute risk for the group event rates and their difference. Then watch relative risk for the ratio of those rates. Focus on why a percentage-point difference and a percentage reduction answer different questions.

In the video’s example, a fall from 30% to 18% is a 12-percentage-point absolute reduction. Relative to the original 30% risk, it is a 40% relative reduction. Neither description is inherently wrong; each leaves something out if reported alone. The relative figure describes the proportion of baseline risk removed. The absolute figure describes the difference in events over the period studied.



Read the semaglutide result

The trial below compared semaglutide with placebo in people with overweight or obesity and established cardiovascular disease, but without diabetes. Its primary outcome was the first occurrence of any one of three events: death from a cardiovascular cause, a nonfatal heart attack, or a nonfatal stroke. That is a composite outcome: one participant contributes a primary-outcome event when the first of those events occurs.

Semaglutide and Cardiovascular Outcomes in Obesity without ...

Read the trial report’s efficacy result and its accompanying table. The report supplies the actual event counts, the trial’s relative-effect estimate, and the uncertainty around that estimate.

In Results, under “Efficacy End Points,” begin with the paragraph reporting the first primary cardiovascular event. Read the results paragraph through its explanation of why later confirmatory outcomes were not tested for superiority. Then find Table 2, “Primary and Secondary Time-to-First-Event Efficacy End Points” on PDF page 7. In its first two outcome rows, compare the two results for the primary composite and cardiovascular death. Below the table, read the footnote beginning “The primary efficacy end point” to confirm exactly what the primary outcome includes. Stop after the footnote explaining the confirmatory secondary endpoints; the other rows are not needed yet.

The same table is reproduced here to make its rows and footnotes easier to inspect:

Table 2 from the semaglutide cardiovascular-outcomes trial: event counts and percentages for each group, alongside hazard ratios, 95% confidence intervals, and notes defining the outcomes.

Read across the primary cardiovascular composite row before looking at the headline hazard ratio. Among 8,803 participants assigned to semaglutide, 569 had a first primary-outcome event (6.5%). Among 8,801 assigned to placebo, 701 did (8.0%). Those are the observed event proportions over the trial’s follow-up, not predictions for every person who might take the medicine.



Put the effect on an absolute scale

Subtract the treatment-group event percentage from the placebo-group percentage:

In plain language, about 15 fewer people per 1,000 in the semaglutide group experienced a first primary-outcome event during the trial than in the placebo group. This is an average difference between randomized groups, not a way to identify which particular people avoided an event.

You may also encounter the number needed to treat (NNT), calculated approximately as the inverse of the absolute difference expressed as a proportion:

Here, that shorthand means roughly 67 people treated for a period like the trial’s follow-up per one fewer first composite event, using the rounded percentages. It is not “67 people to prevent one death,” and it is not a timeless property of semaglutide. A precise time-to-event NNT would require an analysis at a specified follow-up time; the calculation above is a useful approximation from the reported proportions.

Absolute effects depend strongly on starting risk and the observation period. Even if two populations had the same relative reduction, a population with fewer events to begin with could have a smaller difference in events per 1,000 people. This is why the trial’s 1.5-percentage-point difference should not be transferred automatically to a population with a different cardiovascular history.



Put the effect on a relative scale—without mixing up two ratios

Using the rounded event percentages, the observed risk ratio is approximately

The corresponding relative reduction in the observed event proportions is approximately

or 19%. That answers the question “How much smaller was the observed event proportion relative to the placebo group’s proportion?” It does not mean that 19 out of every 100 treated people benefited. The absolute difference was about 1.5 out of every 100 people.

The paper’s reported relative measure is instead a hazard ratio of 0.80. A hazard ratio incorporates when first events occurred during follow-up, as well as whether they occurred. A careful plain-language reading is: the estimated rate of first primary-outcome events among participants still at risk was about 20% lower in the semaglutide group over the trial follow-up. A hazard ratio is not literally a ratio of the final event percentages, although 0.80 and the approximately 0.81 risk ratio happen to be close here. Do not substitute “20% fewer people had an event” for the hazard-ratio finding without checking the event proportions.

Both scales belong in an account of the result. “About 20% lower hazard” describes the relative time-to-event finding; “6.5% versus 8.0%, or about 15 fewer first events per 1,000 participants” gives its absolute context.



What the confidence interval adds

The reported hazard ratio is 0.80, with a 95% confidence interval from 0.72 to 0.90. The point estimate, 0.80, is the trial’s single best estimate under its analysis. The interval conveys uncertainty: values from 0.72 to 0.90 are compatible with the observed data under the statistical model. Expressed as hazard reductions, its ends correspond to approximately 28% lower and 10% lower hazard, respectively.

A useful way to understand “95% confidence” is to imagine repeating the study procedure many times under the same conditions. Intervals constructed this way would contain the underlying effect about 95% of the time, assuming the method’s conditions hold. It does not mean there is a 95% probability that this particular, already-calculated interval contains the true value. Nor is 0.72 to 0.90 an interval for the absolute reduction: the table does not supply an absolute-difference confidence interval.

For a ratio such as a hazard ratio, 1 represents no difference between groups. The primary outcome’s interval, 0.72 to 0.90, lies entirely below 1, supporting a lower hazard for that prespecified outcome. Now compare the next row of Table 2: cardiovascular death had a hazard ratio of 0.85, but its interval was 0.71 to 1.01. Because that interval extends just above 1, the trial does not establish a reduction in cardiovascular death as a separate outcome. This does not prove the medicine has no effect on cardiovascular death; it means this result leaves more uncertainty about that specific claim.

The distinction matters because the primary result is for a composite of three events, not for each component individually. A clear result for the composite must not be silently rewritten as a clear result for death.



A reusable plain-language reading

A compact interpretation of the primary finding would be:

In this trial population, a first cardiovascular death, nonfatal heart attack, or nonfatal stroke occurred in 6.5% of participants assigned to semaglutide and 8.0% assigned to placebo during follow-up—about 15 fewer first events per 1,000 participants. The estimated time-to-event hazard was 20% lower with semaglutide; its 95% confidence interval was compatible with roughly 10% to 28% lower hazard. This finding concerns the composite outcome in the people studied, not a guaranteed benefit for every individual or a separately established reduction in cardiovascular death.

Notice how each clause has a job. The outcome and population set the scope. The two event rates make the absolute effect visible. The hazard ratio states the trial’s relative time-to-event estimate. The confidence interval communicates uncertainty. The final sentence prevents a claim about the composite from expanding into claims the trial has not separately established.



Key takeaways

A clinical-trial headline becomes much more informative when you attach its relative effect to the underlying event rates. Here, 6.5% versus 8.0% is an absolute difference of 1.5 percentage points, while the reported hazard ratio of 0.80 describes an estimated 20% lower time-to-event hazard. Its 95% confidence interval, 0.72 to 0.90, gives a range of effects compatible with the data—not a promise about an individual result or an interval for the absolute difference.

The next lesson uses this reading method to examine cardiovascular and kidney outcome trials, paying particular attention to which benefits have been established for which drug and which studied population.