A Guide on How NOT to Message the COVID Vaccine's Effects

Co-published on Publish0x.

As a disclaimer, this article is not a commentary on the efficacy or safety of the COVID vaccines and only serves to critique the messaging behind the vaccines' risk reduction.

There's an ongoing debate on whether children need to receive the COVID vaccines or not. People on the left side of the political spectrum generally argue for it whereas those on the right argue against it. Earlier today, I stumbled across a 33 second video on YouTube. In it was NBC News contributor Dr. Natalie Azar attempting to explain why children should receive the COVID vaccines. Long story short, she did not do a good job.

Other than the reduction of the number of ICU visits and deaths per million children, Dr. Azar was vague. "Hundreds" of prevented hospital visits can range from 200 to 900 for all we know. "Thousands" of prevented infections has an even larger range. Assuming she's distinguishing between "thousands", "tens of thousands", and "hundreds of thousands", the former can range from 2000 to 9000. Quite simply, it was terrible messaging and here's why.

A Small Epidemiology Lesson

In epidemiology, there are several important variables to consider including risk ratio (RR), relative risk reduction (RRR), absolute risk reduction (ARR), and number needed to treat (NNT). Below is an example of how to calculate for each.

Here's how to interpret the numbers:

• For the 100 subjects taking Drug B, 3% of them contracted Disease X.

• 5% of the control group contracted the same illness.

• A risk ratio of 0.6 means that if you were to take Drug B, you have a 60% chance of catching Disease X compared to not taking the drug.

• A relative risk reduction of 0.4 means that taking Drug B will lower the chance of contracting the illness by 40%.

• An absolute risk reduction of 0.02 means taking Drug B will lower the chance of contracting the illness by 2 percentage points.

• A NNT of 50 means you need 50 people to take Drug B to prevent a single case of Disease X.

What's Wrong with Dr. Azar's Messaging?

Let's say for every 1 million children vaccinated against COVID, 9000 of them are prevented from contracting the virus. Dividing 1 million by 9000, the NNT will be 111.11. That would mean you need to vaccinate 111 children in order to prevent one COVID infection. For reference, a well-matched influenza vaccine has a NNT of 37. A poorly-matched flu vaccine has a NNT of 77. According to Brisson et al. (2007), for 12 year old females receiving the human papillomavirus (HPV) vaccine, the NNT for preventing a single case of genital warts is 8 and the NNT for preventing a single case of cervical cancer is 324.

We can also take the NNT and calculate for the ARR. An NNT of 111.11 would mean the ARR is 0.009. In other words, a child receiving the COVID vaccine will have a lower risk of getting infected by 0.9 percentage points. You may be tempted to conclude that the vaccine is therefore useless because the absolute risk reduction is so minuscule. However, that is actually not the problem.

The problem is that Dr. Azar did not provide the relative risk reduction or any numbers that would help us calculate for it. In order to calculate for the RRR, you need to subtract the incidence of the control group from the incidence of the exposed (i.e. getting the vaccine) group and then, divide that difference by the incidence of the control group. The reason why the ARR hardly tells you the full picture is that it only tells you how many percentage points drop. Did the risk of contracting COVID drop from 10% to 9.1% or 2% to 1.1%?

While the ARR in both scenarios are identical, the RRR are drastically different. In the first scenario, the RRR will be |0.091 - 0.10|/0.10 = 0.09, meaning the vaccine will reduce the risk of contracting COVID by 9%. In contrast, in the second scenario, the RRR will be |0.011 - 0.02|/0.02 = 0.45, meaning the vaccine will reduce the risk of contracting COVID by 45%.

Unfortunately, because Dr. Azar does not provide any incidence numbers, we are left completely in the dark.

Lastly, vaccinating 1 million children to prevent one death is not particularly impressive. Here's an excerpt from Leong et al. (2021) on the effectiveness of the COVID vaccine on Canadian adults:

On the basis of these data, the estimated NNV to prevent 1 COVID-19 death in Canadian adults overall is 33,595 and in men aged 70-80 years is 8722, and in women aged 70-80 years is 9060.

While NNT does not provide the whole story as I established earlier in the section, the NNT for children is several magnitudes greater than the NNT for Canadian adults (29.7 times to be more precise). It is even more so if we are to compare the NNT of children to the elderly 70-80 year old cohorts (over 100 times).

One can conclude from those numbers that the elderly population and adults, in general, are at higher risk of dying from COVID. As a result, it would be more beneficial to focus efforts towards vaccinating the adult population instead of the children population.