Relative risk, absolute risk, and why the headlines got HRT badly wrong
Let’s start with some fruit.
Imagine you have a bowl with three lemons and two limes. Lemons are good. Limes, for the sake of this exercise, are not. Someone adds another lime. You now have three limes – a 50% increase. That sounds significant.
Now imagine a room packed floor to ceiling with lemons. Thousands of them. And somewhere in the corner, two limes. Someone adds another lime. Still a 50% increase in limes. Technically true. But the story it tells is completely different.
This is the difference between relative risk and absolute risk. And it matters enormously – not just in statistics lectures, but in real decisions that real people make about their lives.
What happened with HRT
In 2002, the Women’s Health Initiative study published findings linking combined hormone replacement therapy to an increased risk of breast cancer. The findings were reported widely, and the headline figure – a significant percentage increase in risk – caused alarm. HRT prescriptions fell sharply. Women who had been using it stopped. Women who might have benefited from it didn’t start.
The number being reported was a relative risk increase. What was less clearly communicated was the absolute risk – which translated to approximately 4 additional cases of breast cancer per 1,000 women taking HRT over five years. That is not nothing, and it is a number worth knowing. But it is a very different number to the one most people took away from the coverage. The room was full of lemons. The headline counted limes.
A generation of women made health decisions based on a misreading of what the data actually said. That is not a minor data communication issue. That is a real-world consequence of presenting relative risk without context.
Why this keeps happening
Relative risk is not wrong or dishonest. It is a legitimate and useful way of measuring and communicating change. The problem is that it is almost always more dramatic than absolute risk, which makes it more likely to be the number that gets reported, shared, and remembered.
A 50% increase sounds alarming. Four in a thousand over five years prompts a more nuanced conversation. Both can be true at the same time. Only one of them gives you enough to make an informed decision.
This is why, when you see a percentage change in risk – in a news headline, a report, a briefing document, or a dashboard – the first question worth asking is: 50% more than what? What is the baseline? What does this look like in absolute terms?
What to ask instead
When you encounter a relative risk figure, a few questions help put it in context:
What is the baseline rate? A 50% increase from a rate of 2 in 1,000 is very different from a 50% increase from a rate of 200 in 1,000.
Over what time period? Risk figures can be per year, per decade, or over a lifetime. The same absolute number looks very different depending on the window.
Compared to what else? In the HRT case, the risk of breast cancer from HRT is now understood to be lower than the risk associated with several common lifestyle factors – but that comparison rarely made the headlines.
The point of this series
Numbers are not neutral. They are always a choice – a choice about what to measure, how to frame it, and what to leave out. That does not make data untrustworthy. It makes data literacy essential.
Each post in this series takes one common data concept that tends to get misread, misreported, or misunderstood – and tries to explain it in plain terms. Not to be cynical about data, but to help you use it with more confidence.
Because yes, the number might be technically correct. But what does it actually mean?