Relative risk, versus Actual risk

If you've ever read the Daily Mail, you'll likely have seen headlines such as "Eating food A can reduce your risk of cancer B by 50%".

Let's look at that. If my lifetime chance of cancer B was 5 cases per 10 people, and it was reduced by 50%, it would be reduced to 2.5 cases per 10 people (2.5 is half of 5). Before the risk was 50% ((5/10)*100), and eating food A reduced it to 25% ((2.5/10)*100) - Great, right?

However the above example illustrates a situation when the relative risk, aligns to the actual risk. This is rarely the case in reality.

What if my lifetime chance of cancer B was only 1 case per 1,000,000 people? Eating food A would reduce my risk of cancer B by 50% to 0.5 cases per 1,000,000 people (50% of 1). 

Before my actual risk of cancer B, expressed as a percentage would be ((1/1,000,000)*100), which is 0.0001%. After my risk of cancer B would be halved to 0.00005%. So my actual risk reduction is only 0.00005% (0.0001 - 0.00005) compared to what is was before, despite my risk having halved by 50%.

So, without knowing the actual original lifetime risk, you cannot know how much your actual risk of a disease decreases for any form of screening or procedure.