A 50% reduction. But 50% of what?

Start with the denominator

Imagine a hypothetical study: 2 of every 100 people in one group experience an event, compared with 1 of every 100 in another group, over the same period. These are invented numbers for explaining the arithmetic.

For this exercise, assume both groups are observed for one year, with complete follow-up and each person counted once. The illustration zooms to a 0–2% scale: its full-length bar represents 2%, not 100%.

Two correct numbers, two different questions

The absolute difference is 2% minus 1%: one percentage point. The relative reduction is that difference divided by the starting risk: 1% ÷ 2% = 50%. Both describe the same example. The relative number alone does not tell you how common the event was.

Carry the time frame with the number

Ask which event was measured, over how long, and in whom. A one-year risk and a ten-year risk cannot be substituted for each other. A study average is also not a prediction for a particular person.

What to look for in the source

Find the event counts, group sizes, follow-up period and uncertainty interval. A hazard ratio describes events over time; it should not automatically be read as a simple difference between percentages. For a real study, use its reported statistical analysis rather than applying this toy example mechanically.