Absolute and Relative Risk in Plain Terms

Two ways to describe one change

Health news often reports a risk as a percentage change. A habit or a treatment "cuts risk by half". The figure may be accurate. It is also incomplete.

Two terms help. Absolute risk is the number of people who experience an event in relation to the population at large. Relative risk is a comparison between two groups, or between the same group over time. Both are legitimate descriptions. They answer different questions, and a claim that gives only one leaves the reader with half the picture.

An illustration with invented numbers

Suppose, purely as an illustration, that 2 in every 1,000 people in one group experience an event, against 1 in every 1,000 in another. The relative reduction is 50 per cent. The absolute reduction is 1 person in 1,000, or 0.1 percentage points.

Now suppose the starting figure were 200 in 1,000, falling to 100 in 1,000. The relative reduction is again 50 per cent. The absolute reduction is 100 times larger, at 10 percentage points. The headline percentage cannot tell these two cases apart. These numbers are invented to show the arithmetic. They are not drawn from any study.

What the research says about presentation

A Cochrane review by Akl and colleagues examined how people respond when the same risks are presented in different statistical formats. It pooled 35 studies and 83 comparisons. Relative risk reductions may be perceived as larger and are more persuasive than absolute risk reductions. Yet compared with absolute figures, relative figures made little or no difference to understanding. The difference lies in persuasion, not comprehension.

The authors caution that relative risk reduction is likely to misinform decisions, particularly when the baseline risk is low. They also say it is uncertain whether presenting relative figures helps people choose in line with their own values. The evidence on the best format is not settled.

Two questions we ask of any risk claim

We think every risk claim should answer two questions. What was the risk to begin with? What is it afterwards, in the same units? If a source gives only a percentage change, we treat the claim as unfinished and look for the baseline.

We cite our sources so that readers can check this reasoning for themselves. If we have misread the Cochrane review, or any other source, we would like to be told.

Informational only, not medical advice.