Relative Risk Sells. Absolute Risk Informs.

"This drug cuts your risk by half." "This drug moves your risk from two in a thousand to one in a thousand." Read those again. They can be the identical result — same trial, same patients, same numbers. One of them ends up in the headline and the press release. The other ends up buried in a table on page nine, if it ends up anywhere at all. The preference for the first phrasing over the second is the most efficient distortion in all of health communication: technically true, reliably misleading, and almost never corrected once it leaves the building.

It is efficient because it costs the writer nothing. No fabrication, no spin you could underline and object to. The relative figure is real. It just answers a different question than the one the reader thinks they asked — and the gap between those two questions is where a great deal of unnecessary fear, and a fair amount of misplaced confidence, gets manufactured.

The arithmetic, once, and then never again

Three numbers, and they are not hard. Suppose a condition strikes 2 people in every 1,000 over some period. A drug brings that down to 1 in 1,000. The absolute risk reduction is the plain subtraction: 2 minus 1, so 1 fewer case per 1,000 people. The relative risk reduction is that same drop expressed as a proportion of where you started: 1 out of 2, which is 50 per cent. (These figures are illustrative — invented to show the machinery, not lifted from any real product or study.)

Notice what just happened. Nothing about the underlying effect changed between those two sentences. The drug did exactly one thing. But "50 per cent" and "1 in 1,000" land in the body like different-sized objects entirely, and the writer gets to choose which one you catch.

There is a third number worth keeping, because it translates the effect into something human-scale: the number needed to treat. If you prevent 1 case for every 1,000 people who take the drug, then roughly 1,000 people take it for 1 to benefit. That figure is unglamorous on purpose. It is the honest unit of "how much, for how many" — and it is precisely the unit that relative framing is built to keep out of view.

The whole trick lives in the baseline. When the starting risk is low, the relative number is spectacular and the absolute number is humble, every time. Halving a tiny risk produces a still-tinier risk and a gorgeous percentage. The lower the baseline, the wider that gap stretches — which is why the most modest real-world effects so often arrive wearing the largest, most quotable headlines.

Who the big number is for

Be clear about something: nobody has to lie for this to work. The relative frame is simply the more useful tool for everyone in the chain except the reader.

A press office exists to produce a sentence people will repeat. "Cuts risk by half" is that sentence. "One fewer case per thousand" is not, and no communications team that wants its work picked up is going to choose the second when the first is sitting right there, equally accurate. The incentive isn't villainy. It's the job.

The same gravity acts on the literature itself. Abstracts — the most-read and least-varnished-resistant part of any paper — tend to lead with the relative effect, because the relative effect is the larger and more publishable number, and the people writing the abstract believe in their result and want it read. The absolute difference is often present somewhere in the full text, fair and square. It simply isn't the figure doing the talking.

And then there is the move that should bother you most, because it happens inside a single document. Watch how benefits and harms get dressed in opposite clothes. The benefit is reported relatively — halves the risk, the big proportional number. The harm, two paragraphs later, is reported absolutely — affects 1 in 5,000, the small reassuring count. Both framings are available for both figures. Choosing the flattering one for each, in the same breath, is how a thoroughly true page can leave a thoroughly false impression. It reads as rigour. It functions as salesmanship.

Where it actually draws blood

This isn't a parlour problem. The relative frame, loosed into the world, has predictable places where it does damage.

Screening is the classic battleground. A programme is announced to "reduce deaths from some condition by 20 per cent", and the figure is true, and it is also almost useless to a person trying to decide whether to take part — because without the baseline, 20 per cent of a large risk and 20 per cent of a vanishingly small one are indistinguishable in the sentence, and they imply utterly different decisions. The benefit gets the proud relative figure; the costs of screening — the false alarms, the chase-down of things that were never going to matter — tend to get described in flatter, absolute terms, if they make the page at all. Same asymmetry, higher stakes.

Then there is the medication headline, the one that runs "everyday drug DOUBLES your risk" of something frightening. Doubling sounds like a verdict. But doubling a risk that began at 2 in 10,000 gets you to 4 in 10,000 — an extra 2 people per 10,000, which is a real effect and also nothing remotely like what the verb "doubles" detonates in a worried mind at eleven at night. The word is accurate. The impression is not. (Again: illustrative numbers, attached to no real drug.)

And the impression is the product that gets delivered. The reader does not retain "4 in 10,000". They retain doubles, and risk, and the name of the drug they take every morning. The downstream cost of a health-information diet built on relative framing is a population walking around with their personal risk estimates inflated by orders of magnitude — and a chunk of that miscalibration eventually arrives, in person, in front of a clinician who now has to spend the visit dismantling a number a headline installed.

One question dismantles most of it

The defence is almost insultingly simple, which is the best thing about it. Five words: from what, to what?

Any risk claim that won't answer those five words is not yet information. "Cuts risk by half" — from what, to what? From 2 in a thousand to 1 in a thousand, or from 4 in 10 to 2 in 10? Until that's answered, you genuinely do not know whether you've been handed a reason to act or a reason to shrug, and the percentage alone cannot tell you which. Treat a bare percentage with no baseline the way you'd treat a price with no currency: an unfinished number, withholding the part that would let you judge it.

So demand the baseline. Distrust, on reflex, any proportional claim that arrives without one — not because someone is necessarily deceiving you, but because the missing denominator is exactly the piece that decides whether the headline means anything, and its absence is rarely an accident.

For anyone doing the writing — the journalist, the comms team, and increasingly the person supervising a machine that drafts the first pass — the obligation runs the other way and it is not onerous. State both framings. Lead with the absolute one, because it is the one that tells the reader what is actually at stake for a person like them. Keep the denominator visible. And apply the symmetry test before you publish: are the benefits and the harms on this page wearing the same clothes, or have you quietly handed the good news the big proportional number and the bad news the small reassuring count? If the framings don't match, you have written persuasion and called it information.

What this means

A risk number stripped of its baseline isn't information. It's persuasion with units. It tells you a direction and a magnitude that feels enormous, while withholding the one fact — the starting point — that would let you decide whether the magnitude is worth a moment of your attention. The relative figure is the version of the truth that travels: light, dramatic, eminently quotable, and built to move. The absolute figure is the version that stays still long enough to actually be understood. Health writing earns trust in the gap between the two — and the discipline is nothing grander than refusing to let the spectacular number stand in for the honest one, every single time, especially when the spectacular number is your own.