Are yearly Covid-19 boosters the best we can do?

No. But alternative plans will take creativity and effort, and apparently, we're too tired. Nevertheless, lives are on the line. Timing Boosters: Part I.

People often ask me, “should I time my booster?” The answer is that for many people, it’s perfectly reasonable and even wise to do so. But to make national policy around this would take a lot of effort and would seem to carry some risk. Today, I make the case for why we should seriously consider this. I also share my theory for why officials are unlikely to do this, even though they should. Thanks for being in the Inside Medicine community!

---

The FDA vaccine advisory committee is scheduled to meet today to discuss future booster plans. It seems to me that we are headed towards a universe in which Covid-19 boosters are offered once a year for everyone, and perhaps more often for high-risk people.

Is that the best we can do? No.

Now, there’s a tension between the need for officials to provide clear and simple public health guidance and the reality that one-size-fits-all policies may come at the expense of the best-possible outcomes for individuals.

So, I’m going to make an argument that is biologically and epidemiologically unassailable but one that is also sacrilege among many mainstream public health experts: On an individual level, waiting to get a booster until a bad surge of Covid begins would almost always be better than getting vaccinated at the soonest moment that you become eligible for another dose.

We can understand this conceptually or mathematically. Let’s do both.

Conceptual framework for booster timing. Imagine I offer you an umbrella and you accept the gift. You carry it all the time, even when it’s not raining. When it happens to rain, you are protected every second. As long as you don’t mind carrying it all the time, there’s no downside to accepting my offer right away. This is analogous to getting boosted as soon as you become eligible for your next dose.

But what if I tell you that the umbrella only works for 3 months after I hand it over because it’s flimsy and will eventually fall apart (and there are no more umbrellas after until next year). Would you accept it right away, even if I gave it to you in the less rainy part of the year? Or would you wait until the winter and accept my gift then?

If the umbrella never fell apart and worked indefinitely, you’d want to take the gift right away. But if you know it’d fall apart after 3 months of carrying it around, you’d be wiser to wait a few months, when I’ll offer you that umbrella again.

Covid boosters clearly only work for a short period of time. As recently as yesterday, the New England Journal of Medicine published data showing that for the original formulation of the booster, added protection against hospitalization and death lasted less than 2-3 months and that the new bivalent booster may only work for 4 months (we’ll see).

Mathematical framework for booster timing. For simplicity, let’s say the added protection of a booster lasts 3 months and is 90% effective (the peak effectiveness of the bivalent vaccines looks to be more like 70%, but let’s just be generous) before dropping to zero. Remember, 90% effective means that instead of, say, 10,000 deaths per month (among the unboosted), there would only be 1,000 deaths per month, for three months.

When would you rather be boosted, during a relatively quiet 3-month period, or during a huge deadly Covid wave? Let’s imagine the quiet period is September-November and that a huge wave appears January-March. Here’s how that might play out:

Fall “quiet” wave among those boosted on September 1st: A booster which is 90% effective turns 5,000 deaths per month into 500—4,500 lives saved per month x 3 months = 13,500 lives saved during the booster’s useful period.

Winter surge among those boosted on January 1st: A booster which is 90% effective turns 50,000 deaths per month into 5,000—or 45,000 lives saved per month x 3 months = 135,000 lives saved during the booster’s useful period.

In this scenario, timing the booster would save 148,500 lives. So, obviously you’d want to wait until right before the winter surge to get boosted.

In a sense, this is what the FDA/CDC will be kinda sorta doing by rolling out an annual booster in the fall of every year, as opposed to, say July 1st. They’ll basically be saying, “Sure, let’s time boosters to the fall, but let’s not get too cute and try to really time it to when waves occur, because then people will wait around and many will never get boosted.”

Another complaint you’ll hear from public health experts is that if you wait until a surge, it’ll be too late for the boosters to work. That’s just not accurate. Data from Israel and now the US, clearly show boosters working very well within days, if not a week or two.

Thought experiment: A big 8-week surge occurs in January, but you are unboosted coming into it. Given what we know, you’re actually likely to be better off by not having boosted a few months prior, because you could then use your “one time” yearly boost right as the surge took off, which is when you would benefit from it the most.

Here’s how this would play out. You’d have 0% added protection for 1-2 weeks and then get 90% added protection for the next 6 weeks (rather than 0% added protection the whole time if your early fall booster had completely worn off). And since the boosters actually peak in effectiveness quite early, even for people boosted in, say, October or November, their extra protection in January would already have waned by perhaps 80%, and maybe 90%-100%. Even allowing 20% of added protection (i.e., the booster’s added protection didn’t fall to zero and lasted the entire wave), you’d still rather have not boosted earlier. Instead, you could get boosted early in the surge, and then have two weeks with 0% added protection (over and above the overall protection you already have from the primary series, mind you) followed by 90% protection for the remainder of the wave.

Now here’s the math, for our hypothetical 8-week surge:

People who got fall boosters as early as eligible would enjoy 20% added protection during a January wave. 10,000 deaths per week would become 8,000 deaths per week, meaning 16,000 lives saved (2,000 saved x 8 weeks) by boosting in the fall versus having received no booster at all. We could add some back for the lives saved during the quiet time, but it wouldn’t be much in comparison.

Meanwhile, people who waited to be boosted until the massive surge arrived would enjoy zero added protection for 2 weeks (20,000 deaths occur, 0 saved). Then for the next 6 weeks, 10,000 deaths per week would become 1,000 deaths per week. That means that even though 0 additional lives would be saved in the first 2 weeks, 56,000 (9,000 fewer deaths per week x 6 weeks) would be saved by having gamed it.

The logistical problem is that getting everyone vaccinated quickly may not be feasible in real life. But, I argue that we can do well enough on this that it would make it worthwhile. We could actually reach almost everyone we absolutely need to boost by rapidly vaccinating all nursing home and assisted living patients first. There are just 23 million Americans ages 75 and up. If we started there, also including other obvious high-risk groups, and focused all of our resources on that, we would save far more lives than than a one-sentence policy that reads: “Get your fall booster!”

We can do better, if we try. While I believe most scenarios favor timing boosters on the individual level (since we have tools like wastewater and better public reporting on real-time case counts than for any disease in human history), I am concerned that our leaders won’t act on it. Despite the fact that it would probably save untold numbers of lives, I’m pessimistic that plans like the ones outlined here will become official policy.

Why won’t timing boosters become official policy? Two reasons. First, too many experts are stuck in old thinking that we can’t game these things. In the past, they were right. With our tools and resources today, however, we can. Second, we’ve lost confidence in ourselves. At one point, we were vaccinating three million people per day in the United States. If we wanted to again—and it’s just a matter of effort—we could vaccinate 63 million people in just three weeks. We can reach those at-risk and who are willing, very quickly (regardless of whether that number is smaller).

A surge-triggered rollout might feel chaotic, but there are a large number of scenarios in which taking this approach would save an incredible number of lives. I’m worried that nobody is thinking about these things, or else are casting them aside quickly without the rigorous study that these questions demand.

Yes, what I’m proposing is far more difficult than what I suspect will probably happen, if our current inertia remains. But I worry that we are not achieving our best results for a simple and unsatisfying reason: Our brains are tired.