A new study finds that the ability of vaccination to protect the community, as opposed to just the individual, may have waned during the pandemic.

Someday, perhaps in the not too distant future, we will be able to have a calm, and informed debate about what interventions saved the most lives during the coronavirus pandemic. My personal vote, supplemental oxygen, doesn’t get enough appreciation.

Other interventions are no doubt up there as well – dexamethasone, mechanical ventilation, ECMO, proning, and of course and without a doubt – vaccines.

Vaccines have performed better in this pandemic than we had any right to hope for.

Before the vaccines had come out, I had written an article arguing that a 50% effective vaccine would still be a huge boon to humankind during the pandemic. When we saw 95% efficacy in the original clinical trials, it felt like a miracle. Of course, as the virus mutated efficacy has waned, and yet we still see a clear benefit of vaccination in terms of preventing the worst outcomes.

At least, on an individual level.

The vast majority of studies of vaccination focus on the individual – comparing the death rate, for example, between the vaccinated and unvaccinated.

But cast your mind back to the heady days of 2021. Back then, we spoke of vaccination as not merely something good for you – but something good for us – vaccination of individuals within a community to protect the whole community.

But was that right? The data is unequivocal at this point that vaccination protects the vaccinated, but does it help others as well? Did it ever?

We didn’t get a great answer to this question until this study, appearing in the BMJ last week, came out. And the major take home is yes, vaccination helped unvaccinated people, at least for a while.

What makes this study unique is that it uses county-level, rather than individual-level data. Instead of looking at how vaccinated individuals fared during the pandemic, it looked at how counties – 2558 of them representing around 80% of all Americans – fared, based on how vaccinated that county was.

The authors, led by Amitabh Suthar at the CDC, were able to do this because the variation in vaccine uptake by county in the US was immense, as you can see here. Over the course of 2021, some counties rapidly hit essentially 100% coverage of adults within 4 months. Some never got into double-digits.

The authors classified counties into four categories of vaccination coverage, though it is worth noting that the analysis allowed counties to change categories as time went on. At a county level, then, this is the effect of vaccination coverage on covid mortality, with very low coverage as the reference group, during the alpha wave.

At first glance, this graph seems unsurprising. Deaths from COVID were lower in counties that had higher vaccination coverage. But there is something surprising here. The deaths are lower than you would expect, even if vaccination was 100% protective against covid mortality.

In other words, deaths weren’t 40% lower in the counties with up to 40% vaccine coverage, they were 60% lower. What’s going on?

One obvious explanation is that vaccines were protecting more than just the vaccinated. By reducing transmission rates, even among unvaccinated individuals, they decreased the death rate among unvaccinated individuals.

But of course there are other factors at play here. Counties that are highly vaccinated are no doubt more likely to engage in other mitigation behaviors, masking and social distancing for example, that weren’t adequately captured in this dataset. It could be a “healthy county” effect.

And as I mentioned, this was during the alpha wave – vaccine efficacy was very high against that particular COVID strain.

Adding in the Delta data, the authors were able to generate a model to predict death rates across the entire range of vaccine coverage integrating both the alpha and delta waves, which looks like this.

Adding our “perfect vaccine” line is not quite as impressive once delta is thrown in.

Here we can see that if a community is 50% vaccinated, for example, the COVID death rates in that community are only around 40% lower. This could be because the vaccines weren’t quite as effective against delta as they were against alpha – some vaccinated people died. But more likely this reflects that vaccines lost some of their ability to reduce community transmission – since vaccinated people were still getting (mild) infections. And of course, since delta was more infectious than alpha, the herd immunity threshold was higher.

We don’t have data yet to see how this looks in the omicron wave, which would be quite interesting, and of course the whole process could reset if a new vaccine comes out that can restore significant protection against infection and transmission.

But taken together, this study confirms what a lot of us had suspected. Early in the pandemic, vaccination helped all of us, not just the vaccinated. Later, the benefit was more concentrated. It may be at this point that the benefit of vaccination accrues ONLY to the vaccinated. Does that mean we can lay off trying to convince vaccine holdouts that they should get the shot? Not really, it’s still a good choice on the individual level. But it does mean maybe we can worry less if they politely decline.

A version of this commentary first appeared on Medscape.com.