Two ways to describe the same change
Imagine that, over a stated period, 4 out of 100 people in one group experience a health outcome. In a comparable group receiving an intervention, 2 out of 100 do. The absolute risk is the chance of the outcome in each group: 4% and 2%. The absolute risk reduction is the difference between them, or 2 percentage points. Put another way, there are two fewer outcomes per 100 people in the intervention group.
Relative risk compares one risk with another. In this example, 2% is half of 4%, so the intervention group has a 50% lower relative risk. Both “2 percentage points lower” and “50% lower” describe the same result. Neither is inherently wrong, but each answers a different question. Absolute risk shows the size of the difference in the groups studied; relative risk shows its size compared with the starting risk.
Why the starting risk matters
A large relative change can describe a small difference in the number of people affected. Suppose an outcome occurs in 2 out of 1,000 people in one group and 1 out of 1,000 in another. That is a 50% relative reduction, but an absolute reduction of 1 person per 1,000. By contrast, a change from 200 to 100 out of 1,000 is also a 50% relative reduction, yet the absolute difference is 100 people per 1,000.
When a headline gives only a relative figure, look for the baseline risk: how often did the outcome occur in the comparison group? If the article does not provide it, the headline alone cannot tell you how many people might be affected. The comparison is also meaningful only when both numbers refer to the same outcome, a similar population and the same length of time.
Check what is being counted
Risk figures need a denominator and a time frame. “Ten cases” means something different among 100 people than among 100,000. Likewise, a risk measured over a month should not be casually compared with one measured over a decade. A useful way to restate a claim is: “Among how many people, how many experienced what outcome, over what period?”
The outcome itself matters too. A study might count a laboratory measurement, a symptom or an event such as hospital admission. These are not interchangeable. Check whether the claim describes developing a condition, experiencing a complication or dying from it. Also notice whose risk was measured: a result from one group may not describe people with different ages, histories or exposures equally well.
Ask how the comparison was made
A difference between groups does not automatically mean that an intervention or exposure caused it. People who choose different habits, for example, may differ in other ways that affect their outcomes. Researchers can account for some of these differences, but unmeasured factors may remain. In a randomized trial, assigning participants to groups by chance can make the groups more comparable, although trial design and follow-up still matter.
Risk estimates are also uncertain. If only a small number of outcomes occur, the estimated difference may be imprecise. Reports sometimes show a range of plausible values around an estimate; a wide range calls for more caution than a narrow one. Rather than treating a single percentage as a guarantee, read it as a summary of what happened in a particular comparison.
Use both numbers for a clearer picture
When evaluating a health claim, try to find the risk in each group and express both using the same denominator—for example, outcomes per 1,000 people. Subtract the figures to see the absolute difference, then compare them to understand the relative difference. This simple translation can reveal whether an attention-grabbing percentage corresponds to a large or small change in the setting studied.
Finally, consider what the numbers leave out. A potential benefit may come with unwanted effects, burdens or costs, and an average result may not reflect an individual’s circumstances. Absolute and relative risk are tools for understanding evidence, not instructions for making a personal health decision. Read together with the study’s population, time frame and limitations, they make health claims easier to interpret without oversimplifying them.
