Punisher wrote: ↑Thu Nov 30, 2023 8:25 pm
Can someone explain what a relative risk reduction is and why it's misleading?
Relative risk is an epidemiological measurement where we compare what happens when we look at a single factor affecting one group vs that factor missing in another (the control). As an example, let's give baby aspirin to a population of people above the age of 65 that experienced a stroke to see if reduces the chances of having a second stroke. Or let's compare rates of cancer between groups of smokers and non-smokers.
Relative risk is reported as a percentage.
"People that received 5mg of baby aspirin daily experienced a 37% relative risk reduction for experiencing a second stroke in comparison to a similar cohort that did not."
"Smokers have a 37% higher relative risk of experiencing a cancer diagnosis than non-smokers."
Relative risk reduction (RRR) is a specific type of relative risk calculation (ratio) focused on the application of some kind of medical treatment or intervention. The reason it's potentially misleading is because you're only looking act the impact of the intervention as a function of the group that didn't.
Instead, studies will usually report an Absolute Risk Reduction (ARR) which mathematically accounts for this by not turning the calculation into a ratio (like hte RRR) and instead just compares the actual risk reduction between the experimental and control groups.
In practice, this will usually mean a high RRR (37%) will have an ARR of something like 3.1 or 2.7 instead - which is why RRRs are potentially misleading.
Textbook definitions:
Relative risk is the ratio of the probability of an event occurring with an exposure versus the probability of the event occurring without the exposure.
Absolute risk is the actual risk of some event happening given the current exposure.
I wasn't expecting this question here or on a Thursday night.