Research Literacy
Relative risk vs. absolute risk explained
"Doubles your risk" sounds alarming regardless of what it's a risk of. The missing number is almost always the one that tells you whether to care.
A worked example
Suppose a rare condition affects 2 people out of 1,000 over ten years. A study finds that eating a certain food is associated with double that risk — 4 people out of 1,000.
| Affected | Not affected | Risk | |
|---|---|---|---|
| Without the food | 2 / 1,000 | 998 / 1,000 | 0.2% |
| With the food | 4 / 1,000 | 996 / 1,000 | 0.4% |
Reported as relative risk, that's a "100% increase" or "doubles your risk" — technically accurate, and exactly the kind of number that ends up in a headline. Reported as absolute risk, it's an increase from 0.2% to 0.4% — two additional cases per 1,000 people. Both numbers describe the identical result. Only one of them tells you how much it should actually change your behavior.[1]
Why relative risk dominates headlines
Relative risk numbers are almost always larger and more dramatic-sounding than absolute risk numbers, especially for rare outcomes — which makes them better headlines, not better information. Neither number is "the wrong one" in a technical sense; they answer different questions. Relative risk is useful for comparing the strength of different exposures. Absolute risk is what actually matters for deciding whether a personal behavior change is worth it.
The question to ask
When you see a relative risk figure ("40% higher risk," "doubles your chance of..."), ask: what was the baseline risk this is relative to? If the baseline is already high — say, a common condition affecting 20% of people — a 40% relative increase is a large, meaningful absolute change. If the baseline is 1 in 10,000, the same 40% relative increase might mean almost nothing in absolute terms.
Number needed to harm (or treat)
One practical way researchers translate this into something usable: "number needed to treat/harm" — how many people would need the exposure or intervention for one additional case to occur. In the worked example above, roughly 500 people would need to eat the food (at that exposure level) for one additional case to appear, compared to none eating it. Numbers like this make absolute impact much more concrete than a bare percentage.[1]
This same distinction is exactly why "50% increased risk" and "0.2 percentage points" can both be true about the same study — see how to read a nutrition study for the fuller checklist.
Frequently asked questions
What is the difference between relative risk and absolute risk?
Relative risk compares how much more or less likely an outcome is between two groups, as a ratio or percentage. Absolute risk is the actual probability of the outcome occurring. A food that 'doubles' a risk from 0.2% to 0.4% has a large relative risk increase (100%) but a small absolute risk increase (0.2 percentage points).
Why do headlines usually report relative risk?
Relative risk numbers are almost always larger and more attention-grabbing than absolute risk numbers, particularly for rare outcomes, which makes them more compelling headlines even when the real-world impact is small.
What is number needed to harm?
Number needed to harm estimates how many people would need to be exposed to something for one additional case of an outcome to occur, translating a risk ratio into a more concrete, real-world figure.