This is going to be a dull and numbers-heavy statistics post, there’s no way around it. But the concepts of absolute and relative risk are fascinating and eye opening once you learn how they can mislead. Absolute risk simply refers to the risk of a disease or condition happening to someone with a certain characteristic over time, expressed as a percentage. For example, the absolute risk of a smoker getting lung cancer is 10-15%. Relative risk compares the risks between two or more groups of people relative to each other, such as vegans and non-vegans, or people given a drug and those given a placebo. For example, processed meat eaters are 18% more likely to develop colon cancer than people who shun processed meat. Why do these matter when we’re thinking about whether or not to take a drug or engage in some healthy or unhealthy behavior?
This article explains it well. Often research studies highlight relative risk reduction to make the benefit of a drug MUCH larger than it actually is. A study might report that a drug halves the risk of a certain event happening, but in reality the risk of that event happening is already so small that the risk might go from 2% to 1%, hardly a noticeable decrease but technically the risk has been decreased by half (50%). Often the public doesn’t have access to full research articles and can only see abstracts or sensational headlines reporting the larger numbers.
Take statins. They’re one of the many drugs whose benefit is often overstated (and here I’m going to round the numbers a bit to make the math easier). Frequently, studies will report somewhere around a 25% drop in coronary-related deaths for people taking a statin compared to those not on a statin, but what does the data say? In placebo groups, the risk of coronary-related death is around 8%, and in treatment groups the risk is around 6%. The researchers get the 25% by saying that 6 is 25% smaller than 8. It is. But here are the hard numbers: if we have 100 people in each group, 8 would suffer coronary deaths in the placebo group and 6 people would suffer coronary deaths in the statin group. By giving 100 people a statin, 2 extra people will not experience coronary-related death. It’s beneficial of course, but we can’t forget drugs have side effects.
Here’s another example with a diagram (I’m only using this diagram because I think it depicts the concepts clearly - let me be clear that processed meat is troublesome for many reasons and not just bowel cancer!!). If we have 100 people not eating processed meat, 5.6 of them will get bowel cancer. If we have 100 people eating processed meat, 6.6 of them will get bowel cancer. Thus 1 extra person out of every 100 people will get bowel cancer if they eat processed meat. What about the 18% figure that we discussed in the beginning of this blog? It comes from the fact that 6.6 is 18% higher than 5.6.
Source: euphic.org
Another concept along these same lines is NNT, or Number Needed to Treat. It refers to how many patients need to be treated to see one patient receive a benefit. For example, 1,667 people need to take the drug Aspirin on a daily basis for a year for one heart attack to be prevented. For sinusitis, 17 people need to be given antibiotics for one to see a benefit.
Bottom line: The above has been an abbreviated explanation of these concepts, so if you’d like to read further please see some resources below. If we’re talking about a drug, ask your doctor about how many people need to be treated with the drug they are prescribing for one person to see a benefit, and always ask them about the potential side effects.
Other Resources: