Over 355,000 excess deaths could have been prevented from July 2021 to September 2022. Is this a Red State/Blue State problem?

A friend sent me a fascinating piece in yesterday’s Washington Post entitled “How Red-State Politics Are Shaving Years Off American Lives.” It details the credible and devastating reality that where you live has a large effect on your health and longevity—not because of geography, but because of policy.*

Many things in the Post piece ring true and I encourage you to read it and appreciate the graphs. The story looks at outcomes in three adjacent counties along Lake Erie, which happen to form the borders of three states (Ohio, Pennsylvania, and New York). The mortality differences are staggering. The political differences can’t be ignored.

We had the same thought.

Last year, my colleagues (including Inside Medicine’s Benjy Renton and our wonderful collaborators led by Dr. Harlan Krumholz at Yale’s Center for Outcomes Research and Evaluation) and I had questions reflecting similar concerns.

Did where you lived matter in terms of the rates of mortality in the US during the Covid-19 pandemic? Were Red States really that much worse than Blue States? Just how many lives could we have saved if, say, Red States like West Virginia had fared as well as Blue ones like Connecticut? To find out, we decided to do an analysis of excess mortality by state, and run the math out from there.

What we found both didn’t surprise us (yes, it matters) and did (the magnitude blew us away). It turns out that where you live for a long time does indeed correlate to if you live for a long time. Even though we explicitly avoided doing a strict Red-State Blue-State analysis, the political patterns were right there, staring us in the face…

The remainder of this article is for paid subscribers through November 1st. It’ll be free after that.

At first, politics and policies really had nothing to do with outcomes. In the early days of the pandemic, mortality rates tracked the virus, as it worked its way through the country, with states dropping to the virus like flies. But eventually the virus was everywhere, and vaccines were available. We realized that it would be interesting to see how states fared from that point forward as a way of measuring whether state policies and behaviors really matter. Unsurprisingly, the differences were substantial—and they often fell along Red-State Blue-State divisions (a point we downplayed in the article, because it was a study of medical epidemiology, not a political one).

Nevertheless, the period we studied certainly reflected the various political moods in the country. By the time the Delta and Omicron variants surfaced, some people (and places) were already done worrying about Covid-19. Others remained quite concerned. How willing people were to do things that decrease the spread and severity of Covid-19 varied a lot by geography, be it masking, vaccinating/boosting, or access to and uptake of effective anti-viral treatments.

So, did the US states that “did more” have better outcomes? The answer is yes. Earlier work by Dr. Alyssa Bilinksi and colleagues indicated that states with better vaccination rates had lower all-cause mortality during the Delta and early Omicron periods. If all US states had performed as well as the 10 states with the highest vaccination rates, >266,000 fewer all-cause excess deaths* would have occurred during that period.

*Excess deaths are the deaths that go above and beyond the “expected” normal mortality rates. If 1,000 deaths from all causes are expected in a week, but 1,250 occur, there were 250 excess deaths.

My colleagues and I thought this was very important. But we wanted to learn a bit more and make the analysis even purer. Besides, there’s more to pandemic response than vaccines…

So, Inside Medicine’s Benjy Renton** and I joined up with our usual collaborators Led by Dr. Harlan Krumholz at Yale’s Center for Outcomes Research and Evaluation, and looked at this question from a more holistic viewpoint. (We also extended the analysis into the summer and early fall of 2022.) We wanted to know how the US would have done if the bottom 40 states had the same excess mortality rate as that of the 10 which performed the best. We published our findings in the Journal of General Internal Medicine this summer, having found that over 355,000 fewer excess deaths would have occurred in the US during Delta and Omicron (through September, 2022) had places like West Virginia, New Mexico, Mississippi, Oklahoma, and Kentucky experienced as small of an increase above their own expected mortality rates as the 10 best—including the very home states of my research team (Massachusetts, Connecticut, and Washington D.C.; I swear, we did not intend to find that our own states were in the top 10. It’s just what happened).

While we were “purist” (by not just looking at states with the highest vaccination rates—instead pegging the counterfactual to best overall performance on excess mortality regardless of vaccination rates—), we were asked to look at that issue by the journal’s reviewers. So, we published a series of graphs showing a compelling inverse relationship between vaccination rates and all-cause excess mortality during Delta and Omicron (see below).

Here’s how to read them: The vertical axis is excess deaths per 100,000 people. The horizontal axis is the percent of people fully vaccinated in each state. Each of the five panels is a different three-month period, from Early Delta (Summer of 2021) to Omicron BA.5 (Summer of 2022). Each dot is a state.

Take a look:

Renton et al, Journal of General Internal Medicine. Annotations: Inside Medicine.

See the inverse relationship? The higher the vaccination rate, the less all-cause excess mortality there was. Also notice the two red lines I added to the first and last panels. This shows how different states’ outcomes were earlier, back when many people were having their first exposure to the virus, and many without the benefit of vaccination. But by mid-2022, most US residents had been vaccinated, infected, or both. This meant that the inverse correlation, while still there, was less drastic. These graphs underscore the importance of our successful campaign to rapidly vaccinate US adults. Honestly, we should be proud of that, and not take it for granted. Still, if we’d done even better, more lives would have been saved.

Now, I should note that this type of 1:1 correlation between two things (vaccination rates and all-cause excess mortality by jurisdiction) is often frowned upon in medical research as an “ecological analysis”—another way of saying that states are just too different to simply compare them without adjusting for other factors. In fact, JAMA published a similar kind of data in the paper I mentioned above, but snuck into a table so that nobody would notice. The reason we were permitted to publish graphs like these in a very good journal (we think) is the 5-panel trend we showed. Whatever other differences there might be that could provide alternative explanations for the inverse relationship between vaccine uptake and all-cause excess mortality, none of those “confounding variables" could possibly explain a trend wherein the magnitude of the relationship neatly decreased in mirror image to the increasing immunity in the population.

So what? The implications, we think, are large. Look, there’s no good reason that the bottom 40 states could not have done as well as the best 10. The top 10 states weren’t living in some alternative universe. They’re just other states. While you might argue that West Virginia entered the pandemic with a population with higher risks, that falls flat on two counts. First, we looked at excess mortality. West Virginia indeed had the highest pre-pandemic mortality rate of any state. So for that state to also have had the most excess mortality means that it had to exceed its already very high baseline. Second, shrugging this off is defeatist. If West Virginians had (and have) worse health than people in, say Massachusetts, that in large part reflects policies, such as healthcare access and other systemic choices. We get the results that we accept, both at baseline and during pandemics.

That said, the design of our work intentionally avoided any deep inquiry into the specific explanations for state differences, including their Redness or Blueness. We simply “reported the facts” and concluded that it’s worth understanding the driving forces behind these mortality differences going forward. If politics has anything to do with it—as I am forced to conclude it does—you’d think that might matter to people who are still alive to vote in elections.

---

*Aside: I can’t help but mention that my friend—solar physicist Dr. Dan Seaton—does research on the sun, which has led him to focus on the corona; meanwhile, I study the coronavirus. We often ponder the odds of this coincidence, given that we were once freshman hallmates back at Williams College.

**Congratulations to Benjy on his first ever first authorship on a peer reviewed medical/public health paper!