Relative risk describes how much something changes your chance compared with a baseline. Absolute risk describes the actual chance. Confusing the two is the single most common way health information misleads people, and it happens in both directions.
A worked example
If an act carries a baseline risk of one in ten thousand, and a factor doubles it, the relative increase is one hundred per cent and the absolute risk moves from one in ten thousand to two in ten thousand. Both descriptions are accurate. One sounds alarming and the other sounds negligible, and the second is more useful for deciding anything.
Why is the relative figure usually reported?
Because it is larger and therefore more striking. A headline saying a behaviour doubles your risk gets read. One saying it moves your risk from 0.01 to 0.02 per cent does not. This is not usually deliberate distortion; it is what makes a finding sound worth reporting.
Where does it matter in sexual health?
Frequently. Having an untreated rectal infection increases the chance of acquiring HIV during receptive anal sex by a meaningful multiple, which is a real and important finding. If you are on PrEP, the baseline it multiplies is already close to zero, so the absolute effect on you is negligible. The same statistic means different things depending on where you start.
How does this apply to prevention?
Interventions reported as reducing risk by a percentage are describing relative reduction. Doxy-PEP reducing chlamydia by around ninety per cent means ninety per cent of the infections you would otherwise have had, which is a large absolute benefit if you get several a year and a small one if you have never had any. This is precisely why it is provided in a targeted way rather than to everyone.
What about the number needed to treat?
A useful figure that is rarely quoted. It asks how many people need to take an intervention for one to benefit. Where the baseline risk is high, the number is small and the intervention is clearly worth it. Where the baseline is low, the number is large and the calculation becomes a question of cost and side effects rather than obvious benefit.
How should you read a study?
Look for the baseline rate before looking at the effect size. A large relative reduction in something that was rare to begin with may matter less to you than a modest reduction in something common. Ask what population was studied, since a figure from a group with very different infection rates may not transfer.
Where does this go wrong in the other direction?
People dismissing genuinely large absolute risks because they are expressed as small percentages. A one per cent per-act risk sounds small and, repeated across a year, is not. Absolute figures need reading alongside how often the exposure occurs.
What is the practical use of all this?
When you encounter a claim that something increases or reduces risk, ask two questions: from what to what, and how often does this apply to me. Most alarming and most reassuring statistics survive one of those questions and not both.
The rule of thumb
Percentages describing change are relative. Numbers describing chances out of a total are absolute. If a source only gives you the first, it has not given you enough to decide anything.
What about risk expressed as odds?
Odds ratios appear frequently in research and are not the same as risk ratios, though they approximate each other when the outcome is rare. For everyday purposes they can be read similarly, and where an outcome is common they diverge enough to matter. This is one reason study abstracts can be misleading without the full paper.
How does this affect prevention decisions?
It explains why interventions are targeted rather than universal. Doxy-PEP given to someone with four infections a year prevents several; given to someone with none it prevents almost nothing while adding antibiotic exposure. The same relative effectiveness produces very different absolute value.
The habit worth building
When you see a percentage, ask what it is a percentage of. That one question resolves most of the confusion.