No, Seatbelts are not like a Covid19 booster shot in kids 5 to 11

Your bad analogy is harming evidence based medicine

We're living in the pandemic of bad analogies.

A new one is: Just like we didn't require randomized control trials to establish the value of seatbelts in cars, we don't need a randomized control trial of a booster shot in 5 to 11-year-olds measuring severe disease.

There are 2 parts to the analogy:

1. Some things are obviously beneficial and need not be tested.

2. Some things have a very small individual risk reduction, but can matter in aggregate

Unfortunately, the seatbelt analogy fails for boosting 5 to 11-year-olds.

Seat belts are a mechanical device placed outside the body for a set period of time, and is meant to lower the risk of an entirely human created problem: Injury during automobile accidents. The problem exists solely because we created the car. A seatbelt is a way to lower the unique harm we created.

Because seat belts are a piece of fabric outside your body, worn transiently, it is impossible for seat belts to have side effects no one anticipates. There may be injury from seat belt in a car accident. There may even be discomfort. But seat belts cannot result in fever, chills, night sweats, or other secondary consequences. Seat belts cannot cause myocarditis or any other adverse events. There is no chance that by wearing seat belts you will be less likely to respond to a new formulation of an mRNA virus that targets different epitopes (due to imprinting/ original antigenic sin), or actually have some negative future adverse outcome that you cannot anticipate. Frankly, it's always misguided to analogize a mechanical, external device to an internal, immunogenic product.

Second, seat belts are ancient history. They were developed in an era where we had not fully embraced the virtues of randomization (50s-60s). And it is something that evokes a strong intuition— people believe seatbelts are beneficial after decades of supporting evidence (and in concert with other things like soft dashboards, crumple resistant cars, airbags, etc).

But boosting children is not in this category. As long as we're going to play imperfect analogies let me make a slightly better one.

One can imagine in the modern world there is an alternative to a seatbelt. There is a five-point harness. Five-Point harness is what race car drivers use.

A 5 Point harness is to booster what seat belt is to the first two doses. More is better!

We could immediately demand that every human being who sits in a car for any purpose wear a five point race car driver harness. Before you say, kids use 5 point harnesses, to continue the analogy, we can take that those 5 up to 7 points, and add a racing helmet and roll cage to the backseat to boot! In fact people have tried!

A [5 point harness: seatbelt:: booster: first 2 doses] is a far better analogy for getting a third dose of vaccine after you've gotten the first two doses. You already have the lion share of protection that comes from vaccination regarding severe outcomes. Can you improve upon it with more points?

The analogy also makes sense for another reason. Older people are like the race car driver. They're subject to the greatest risks. They are the ones most likely to benefit from the five-point harness. But a regular person driving to the grocery store is much less likely to be injuried or benefit. Do they also need to wear a 5-point harness?

We could totally perform a randomized control trial of the current seatbelt versus a 5-point harness for regular day-to-day use. Apple watch accrued rapidly for its’ trial. The TASTE RCT in the Netherlands accrued 60% of all people with the condition. Even the original Salk polio virus trial recruited hundreds of thousands and measured the outcome that matters. Technology makes large randomized trials easier!

Yet, as you do the power calculation, you will find the sample size needs to be incredibly large, because the postulated benefit has got to be very very marginal at best. But, it is possible.

If I had to bet, I believe a 5-point harness will not win. Whatever extra safety (very marginal) will be offset by misuse/ disuse/ pain. But let's say it wins. It has some survival benefit. You'll have to offset that by all the life years lost by buckling it up. A very trivial absolute risk reduction will not persuade people that it's worth their time. In fact, it may result in more life years lost.

Now let's tackle the second argument. Something can have a small risk reduction individually but good effects on the population level.

This is absolutely true. And applies nicely to things like brighter lines on the road. Safety precautions in the subway, such as sliding doors, preventing falling onto the track before the train arrives. Smoke detectors. And numerous other safety interventions. The main downsides here are annoyance and cost.

But when it comes to medical products that are injected, swallowed, or implanted this argument makes little sense. The downsides of an injectable immunogenic drug product are not fully understood, may occur at later date, nearly impossible to know at the outset, and must be weighed against the trivial benefits. No one knew metal on metal hips would grind and lead to problems.

The purpose of a randomized trial is to demonstrate if there is in fact a net clinical benefit. A huge benefit on an important outcome (death) early will almost never be offset by unanticipated harms. But a trivial benefit early might easily be offset by even measured harms. You simply don't know that until you do the study. You cannot just guess what a medication does.

Moreover, you have to have the consent of the person receiving the product. We still live in a country where most people believe that you should have the right to control your bodily choices as long as those choices do not infringe upon other choices.

Boosting children will do absolutely nothing to slow pandemic spread. It does not have a durable protection against transmission. It cannot help others. It is only being done out of a personal health benefit calculation. And frankly no one can make this calculation because the trials are woefully underpowered to show net clinical benefit.

The Pandemic of bad analogies

We are living in a pandemic of bad analogies. They took place for masking, which is now in year 3, and still lacks any randomized control trial evidence supporting its use as it's being done today. We have no data in vaccinated people. No data in children.

Boosting low risk individuals is another intervention that has no credible data. The randomized controls are deeply inconclusive. It's one thing to offer a choice, but that choice is prayed upon by anxious people. It is not an informed choice because there is not enough evidence.

Giving Paxlovid to low risk vaccinated people is another choice made with very little data. We all know young, healthy people who took paxlovid. Given all the medications that must be stopped (due to interactions), there may be real downside. Another decision where we had ample opportunity to compel proper randomized control trials.

Saying you don't need a adequately powered randomized trial for boosting children, because you didn't have one for seatbelts is as stupid as saying you married your spouse without a randomized trial, therefore, you don't need one for your heart failure medication.

Very different choices require very different types of evidence.

Analogies with a lot of discordance are not useful. When we make an analogy more appropriate. The first two doses are a seat belt. The third dose is a 5-point harness. What evidence do you want to persuade you to use the 5-point harness. Should race car drivers be encouraged to do it or should everyone? As the analogy becomes closer to what you're talking about, the intuitions align.

Finally, I want to add one caveat. When it comes to healthy kids 5 to 11, I have not seen credible data that even the first 2 doses lowers risk of severe disease. We should have done a large RCT for even that question. My analogy is also imperfect. As I said, comparing injectable products to mechanical ones is always imperfect.

Why do bad analogies dominate COVID thinking?

In this section, I will discuss the most important issue, why do these bad analogies dominate COVID thinking…