Skip to content
ContentLora

    Tip: press / anywhere to search.

    Explainer

    Relative vs absolute risk: how to read a health headline

    Absolute risk is the chance that something happens to a person over a set time, while relative risk compares that chance between two groups.[1][2] The same benefit, a fall in risk from 4% to 3%, can be reported as a 25% relative reduction or a 1 percentage point absolute reduction, and clinicians may be more inclined to prescribe a treatment described the first way.[3] To judge a headline, look for the starting risk, the absolute difference and, where given, the number needed to treat.[4][5]

    Editor reviewedStrict sourcingUpdated Explainer14 claims · 100% primaryHealth and medicine, Science
    Reading level: 1 passage differs; the rest is the same for everyone.

    This page explains how risk numbers are reported. It is general information, not medical advice. Talk to a clinician about what a study result means for you, since your own starting risk may differ from the people studied.

    Health headlines often say a food, drug or habit “raises risk by 50%” or “cuts risk by a quarter”. Those are relative figures, and on their own they cannot tell you how many people are affected. The tools below come from the methods handbook used by Cochrane, which produces systematic reviews of health research, and from the US National Cancer Institute (NCI).[3][1]

    Absolute risk: your chance over a period of time

    Absolute risk is the likelihood that a person who does not have a condition at a given age will develop it over a stated period.[1] NCI gives an example: a 35-year-old woman with no known risk factors has an absolute lifetime risk of breast cancer, to age 90, of about 13%. That is a 1 in 8 chance, and it means an 87% chance of never developing it.[6] An absolute risk always needs a time frame. “13%” means little until you know whether it covers a year or a lifetime.

    Relative risk: comparing two groups

    Relative risk, also called the risk ratio, compares the risk of an event in one group with the risk in another. A value of one means no difference between the groups.[2] The relative risk reduction re-expresses a ratio as a percentage, so a risk ratio of 0.75 is a 25% relative reduction.[7] The risk difference, the gap between the two absolute risks, is often called the absolute risk reduction (or absolute risk increase when risk goes up). It can be shown as a percentage, a decimal or a count such as 10 out of 1,000.[8]

    The Cochrane Handbook’s own example shows why both numbers matter. If a treatment lowers the risk of death from 4% to 3%, it is fair to say it “cuts the risk by 25%”. It is equally fair to say it lowers the risk by 1 percentage point. The handbook notes that clinicians may be more inclined to prescribe the treatment when it is described the first way, even though both describe the same benefit.[3]

    Counted in people, that example looks like this: of 1,000 people, about 40 would die without the treatment and about 30 with it, so about 10 fewer deaths per 1,000.[3] Showing results as events per 1,000 people is the format Cochrane uses, because evidence suggests these “natural frequencies” are easiest for the public to understand.[9]

    A relative effect is not a fixed amount of benefit. The same risk ratio gives different risk differences, and different numbers needed to treat, in groups with different comparator risks, unless the intervention does nothing.[4] So a 25% relative reduction in a high-risk group can prevent many more events than the same 25% in a low-risk group. Evidence-grading systems therefore look at both measures. NCI’s PDQ boards, for example, judge the magnitude of an effect “both absolute and relative” and note it may vary between populations.[10]

    Number needed to treat

    The number needed to treat (NNT) is the expected number of people who need to receive a treatment, rather than the comparison, for one additional person to have or avoid an event over a given time.[5] It comes from the risk difference, which makes it an average rather than a promise. An NNT of 10 does not mean one extra event in every group of 10, and by convention NNTs are rounded up to a whole number.[11] In the 4%-to-3% example, 10 fewer deaths per 1,000 people treated works out to an NNT of 100.[3] Because the NNT depends on the starting risk, it is only meaningful for a stated group and time period.[4]

    Checklist for reading a health headline

    • Find the starting risk. A relative change means little without the absolute risk it applies to, and for whom.[1][4]
    • Ask for the absolute difference. Look for “per 1,000 people” or percentage points rather than a percentage change alone.[8][9]
    • Check the time frame. Absolute risks and NNTs always apply to a stated period.[1][5]
    • Look at the uncertainty. Cochrane asks review authors to report the confidence interval and exact P value, not just whether a result was “statistically significant”.[12]
    • Ask what kind of study it was. A link found in an observational study is weaker evidence than a randomised trial. See how epidemiology studies work and how systematic reviews work.[13]
    • Weigh harms the same way. Absolute numbers apply to side effects too, which is why screening guidance such as the cancer screening explainer sets benefits against harms.[8][14]

    Questions readers ask

    What is the difference between relative and absolute risk?

    Absolute risk is the likelihood that a person will develop a condition over a certain period. Relative risk compares the risk in one group with the risk in another, and a relative risk of one means no difference between them.[1][2]

    Why can a "25% lower risk" be misleading?

    Because the same change can be tiny in absolute terms. The Cochrane Handbook's example is a fall from 4% to 3%, which is a 25% relative reduction but only a 1 percentage point absolute reduction, and it notes clinicians may be more inclined to prescribe a treatment described in relative terms.[3]

    What is the number needed to treat?

    It is the expected number of people who need to get a treatment rather than the comparator for one more person to have or avoid an event over a given time. It is an average, so an NNT of 10 does not mean one extra event in every group of 10.[5][11]

    Does a treatment with the same relative effect help everyone equally?

    No. The same relative effect produces different absolute differences and NNTs in groups with different starting risks, unless the treatment has no effect.[4]

    What is the clearest way to show a risk?

    The Cochrane Handbook says evidence suggests natural frequencies, such as events per 1000 people, are best understood by patients and the public, which is why Cochrane summary tables use them.[9]

    Sources

    · 14 claims from 4 publishers

    Each numbered claim is a statement we checked against the sources listed with it. Tap a number in the text to see its sources in place.

    1. [1]

      The National Cancer Institute defines absolute risk as the likelihood that a person free of a given disease at a given age will develop it over a certain period of time. confirmedas of 2026-11-07

    2. [2]

      Relative risk (also called the risk ratio) compares the risk of an event in one group with the risk in another group; a relative risk of one means no difference between the groups. confirmedas of 2026-11-07

    3. [3]

      The Cochrane Handbook notes that a fall in risk from 4% to 3% can be described either as a 25% relative reduction or as a 1 percentage point absolute reduction, and that clinicians may be more inclined to prescribe a treatment described the first way. confirmedas of 2026-11-07

    4. [4]

      The same relative effect produces different risk differences and NNTs in groups with different starting (comparator) risks, unless the treatment has no effect. confirmedas of 2026-11-07

    5. [5]

      The number needed to treat (NNT) is the expected number of people who need to receive a treatment rather than the comparator for one additional person to have or avoid an event over a given time. confirmedas of 2026-11-07

    6. [6]

      NCI's example is that a 35-year-old woman with no known risk factors has an absolute lifetime risk of breast cancer (to age 90) of about 13%, or a 1 in 8 chance, which also means an 87% chance of never having it. confirmedas of 2026-11-07

    7. [7]

      The relative risk reduction re-expresses a risk ratio as a percentage reduction; a risk ratio of 0.75 equals a 25% relative risk reduction. confirmedas of 2026-11-07

    8. [8]

      The risk difference between two groups is often called the absolute risk reduction (or absolute risk increase) and can be shown as a percentage, a decimal or a count such as 10 out of 1000. confirmedas of 2026-11-07

    9. [9]

      The Cochrane Handbook says current evidence suggests absolute risk differences are best understood by patients and the public when shown as natural frequencies, which is why Cochrane summary tables give events per 1000 people. confirmedas of 2026-11-07

    10. [10]

      NCI's PDQ boards judge the magnitude of an intervention's effect using both absolute and relative risks. confirmedas of 2026-11-07

    11. [11]

      An NNT is an expected value derived from the risk difference, so an NNT of 10 does not mean one extra event in every group of 10 people, and by convention NNTs are rounded up to the next whole number. confirmedas of 2026-11-07

    12. [12]

      Cochrane guidance tells review authors not to lean on P-value thresholds or the label "statistically significant", and to report the confidence interval with the exact P value instead. confirmedas of 2026-11-07

    13. [13]

      NCI's PDQ evidence ranking puts randomized controlled trials first, then nonrandomized controlled trials, then cohort or case-control studies, then ecologic and descriptive studies, and expert opinion last. confirmedas of 2026-11-07

    14. [14]

      The USPSTF lists the potential harms of screening mammography as false positives leading to extra tests and procedures, overdiagnosis and overtreatment, and radiation exposure. confirmedas of 2026-10-11

    Revision history (1)
    1. Page created.

    Created Nov 7, 2026. Last reviewed by an editor on Nov 8, 2026. Next scheduled review: Nov 7, 2027.

    Cite this page

    "Relative vs absolute risk: how to read a health headline." ContentLora, updated Nov 7, 2026. https://contentlora.com/explain/relative-vs-absolute-risk-explained

    Spotted an error? Suggest a correction or emailcorrections@contentlora.com.