Relative risk versus absolute risk: how to read a scary health headline

A single study result can be described in two very different ways, and the more alarming version is usually the one that reaches the public. A treatment "triples the risk." A daily habit "raises the odds by 70 percent." A food "doubles your chance" of a disease. Each of these can be technically accurate and still leave a reader with the wrong impression, because each reports a relative figure while quietly omitting the one number that tells you whether the change matters: the baseline risk it started from.

This is the distinction between relative risk and absolute risk. Understanding it is one of the highest-leverage skills in reading health research, because so much of what feels frightening in a headline dissolves the moment the absolute numbers are put back in. The goal here is narrow and practical: to define both terms with a real worked example, to explain why the alarming version tends to win, and to leave a careful reader with a short set of questions to bring to any risk story.

This is educational and not a substitute for personal medical advice.

The same result, told two ways

Start with definitions, because the whole problem lives in the gap between them.

Absolute risk is the actual chance that an outcome happens in a group over some period. If 4 people out of 1,000 develop a condition in a year, the absolute risk is 4 in 1,000, or 0.4 percent. The absolute risk difference between two groups is simply one group's absolute risk minus the other's.

Relative risk is a ratio: the risk in an exposed group divided by the risk in an unexposed group. If the exposed group's risk is twice the unexposed group's, the relative risk is 2, often reported as a "100 percent increase" or "double the risk."

The key insight is that these two numbers answer different questions. A relative figure describes the size of a change. An absolute figure tells you whether that change matters for an actual person. A risk can double and remain trivially small, or rise by a modest-sounding percentage and move a great many people, depending entirely on where it started. The ratio alone cannot tell you which situation you are in.

A worked example, with real numbers

Consider a widely used illustration from Cancer Research UK, drawn from research on CT scans in childhood. The relative figure sounds serious: children who had received the scans were described as roughly "three times as likely" to develop leukaemia or a brain tumour.

Now put the baseline back. The starting absolute risks are very small. As the organization lays out the numbers, roughly 0.4 per 10,000 children develop brain tumours and 0.6 per 10,000 develop leukaemia in that age group. Tripling a very small number leaves a still-small number. Worked through, the scans translate to about one additional case of brain cancer and one additional case of leukaemia for every 10,000 children scanned. The relative change is large. The absolute change is roughly one extra case per 10,000.

The contrast becomes obvious when the baseline is high instead of low. In the same discussion, a reported "5 percent increase" in breast cancer risk from regular drinking, applied to a common cancer with a substantial baseline, works out to roughly 60 additional cases per 10,000 women. Here a small-sounding relative percentage moves many more real people than the dramatic "three times" did, precisely because it acts on a larger starting risk.

That is the entire lesson in one comparison. The size of the starting absolute risk decides whether a big relative change is meaningful or negligible. A relative figure with no baseline attached is not wrong so much as unreadable.

Why the alarming version usually wins

If the absolute framing is so much more informative, why is the relative one so much more common? Several forces push in the same direction.

The simplest is arithmetic drama. Relative figures are larger, rounder, and more quotable. "Triples the risk" makes a stronger headline than "adds one case per 10,000," even when both describe the same study. Press releases and news write-ups reward the version that sounds like news.

A subtler problem is what science-communication researchers call mismatched or inconsistent framing. A detailed toolkit from HealthNewsReview describes a recurring pattern in which benefits are presented in relative terms, which makes them look bigger, while harms are presented in absolute terms, which makes them look smaller. When the two sides of a trade-off are reported on different scales, a reader cannot compare them honestly, and the comparison tilts without anyone stating a falsehood.

The distortion often begins upstream of the newsroom. The same reviewers note that the absolute numbers are frequently missing from study abstracts and press materials themselves, so a journalist working from those sources may never see the baseline at all. The reader inherits an omission that started several steps back.

Relative reduction, absolute reduction, and number needed to treat

The same asymmetry appears in reverse when a study reports a benefit rather than a harm, and the vocabulary is worth knowing.

Relative risk reduction expresses how much a treatment lowers risk as a proportion of the starting risk. A drug that cuts an event rate from 2 percent to 1 percent has reduced relative risk by 50 percent, which sounds substantial.

Absolute risk reduction is the plain difference in event rates between the groups. In that same example it is 2 percent minus 1 percent, or one percentage point. The relative reduction of 50 percent and the absolute reduction of 1 percentage point describe the identical result. They simply frame it differently, and the relative version will always be the larger-sounding one when the baseline is low.

A useful companion measure is the number needed to treat, the number of people who would have to receive a treatment for one additional person to benefit. It is the reciprocal of the absolute risk reduction, and it translates statistics back into people. An absolute reduction of one percentage point corresponds to a number needed to treat of 100, meaning about 100 people take the treatment for one to avoid the outcome. Whether that is impressive depends on the condition, the cost, and the harms, none of which a relative percentage reveals on its own. These are descriptive tools for reading evidence, not a basis for any individual decision, which belongs with a clinician who knows the specific case.

The methodological point beneath all of this is not new. In a review in Nephrology Dialysis Transplantation, methodologists argued the case directly in the title itself.

Noordzij M, van Diepen M, Caskey FC, Jager KJ (2017). Relative risk versus absolute risk: one cannot be interpreted without the other. Nephrology Dialysis Transplantation.DOI: 10.1093/ndt/gfw465

Their argument is that the relative figure, by compressing two numbers into one ratio, conceals the underlying absolute risks and invites readers to overestimate the effect. The ratio is not the villain. Reported alone, without the baseline it is built on, it is simply not interpretable.

Questions to ask of any risk headline

None of this requires statistical training. A few questions, asked in order, do most of the work.

• What was the baseline risk to begin with? A percentage change is meaningless until you know what it changed from.
• Is this a relative change or an absolute one? "Doubles" and "triples" are relative. "One additional case per 10,000" is absolute. If the story does not say, assume relative, since that is the version usually chosen.
• Are the benefits and harms reported on the same scale? If a benefit is quoted in relative terms and a harm in absolute terms, the comparison is rigged, even if unintentionally.
• How many real people does this change per 1,000 or per 10,000? Converting back to a count of people is the single most clarifying move available, and it is exactly the number the alarming framing tends to leave out.

Reading risk well is a skill rather than a talent, and these few questions carry most of the load. The next time a headline announces that something has tripled a risk, the useful response is not alarm and not dismissal. It is a question: tripled from what?

Readers who want to see how these ideas play out in real study design can explore the peer-reviewed publications behind this work.