RRR (Relative Rate Reudction) is just for marketing because it sounds good. We should be using ARR (Absolute Rate Reduction) to describe vaccine effectiveness.
Consider this hypothetical 2-month vaccine trial:
100,000 participants: 50,000 vaccinated / 50,000 unvaccinated
Cases detected: 50 vaccinated, 500 unvaccinated
The RRR is 90%, ARR is 0.9% (1% to 0.1%), and NNTV is 111
All are valid, but which best communicates the VE (effectiveness)?
Explanation for RRR:
90% of cases that would have occurred otherwise are prevented with vaccination.
Explanation for ARR:
The risk of infection is reduced from 1% to 0.1% for 2 months.
Notice how we had to include the time period. Unless the VE drops to 0% suddenly at the end of the trial, the ARR is only valid for the time period indicated by the study.
Also notice that we included the risk before and after, rather than just the reduction of 0.9%. An absolute reduction of risk from 1% to 0.1% is a very significant drop! Whereas if the risk dropped from 90% to 89.1%, this would imply that the vaccine is completely INEFFECTIVE.
Explanation for NNT (same as NNTV):
For every 111 people vaccinated, we reduce the number of cases by 1 within 2 months.
Notice that like ARR, we needed to specify the time period for the figure to be of any value whatsoever, since the vaccine continues to work (even if there is declining effectiveness) after the trial ends.
The missing piece of information in RRR is the level of background risk, but most people have a general awareness of how much risk COVID-19 poses, and if not, this can easily be explained. RRR is clearly the easiest to understand and most useful figure for measuring vaccine effectiveness.
The FDA publication Communicating Risks and Benefits: An Evidence-Based User's Guide (https://www.fda.gov/about-fda/reports/communicating-risks-and-benefits-evidence-based-users-guide) recommends using ARR.
Absolute risk, relative risk, and number needed to treat is discussed on page 56 in Chapter 7 ("Quantitative Information"). The example provided:
When explaining the benefits of taking chemoprevention to prevent breast cancer, risk reduction could be described as (1) a 50% risk reduction (relative risk reduction), (2) a reduction from 6% risk of breast cancer to 3% (absolute risk reduction) or (3) the number of women needed to take chemoprevention to prevent cancer in one of them (NNT).
NNT was found to be the most difficult to understand, and that risk reduction seemed more favourable when described using the relative risk reduction than the absolute risk reduction. Absolute risk reduction is recommended.
IMPORTANT! The example doesn't describe ARR as 3%, but as a drop from 6% to 3%. The anti-vaccination arguments never call for the range to be specified.
IMPORTANT! The example is not a vaccine or preventative treatment. The word "vaccine" does not appear in chapter 7.
When you read the example provided, it's easy to see how the ARR (with the before/after risk stated) is more useful if you know nothing about the background risk for breast cancer. It simply conveys more information.
But ARR falls down in usefulness in a pandemic, where your risk of exosure to the virus increases over time and varies widely depending on your living situation and location.
Let's say the trial length is doubled. Surely you would expect the cases to rise, and surely you would expect them to rise proportionately. This would result in a completely different ARR, while the RRR would be relatively consistent.
The RRR will also be relatively consistent for those at higher and lower risks, whereas the ARR would vary widely depending on age, health, and location.
RRR is the clear winner for accurately communicating COVID-19 vaccine effectiveness.
How to calculate RRR, ARR, and NNT/NNTV: https://bestpractice.bmj.com/info/toolkit/learn-ebm/how-to-calculate-risk/
Relative risk rather than absolute risk reduction should be preferred to sensitise the public to preventive actions
https://gut.bmj.com/content/71/6/1045