Study looks good for democrats, but misses an important confounder.

As a physician, I often think of health as the sum of a set of choices and behaviors an individual makes. Does this patient smoke, do they get exercise, do they eat well, do they take their medications as prescribed? Of course, the reality is that health is impacted by factors beyond an individual’s control all the time. But to be honest I rarely ask if my patients work in a factory where they may be injured or where the chemicals they work with may be carcinogenic, or if they make too little money to eat healthy, or can’t get exercise because they can’t get childcare. Why? Well, because I can’t really do anything about it.

These larger factors are choices we’ve made, as a society. They are the result of policies created by people we elect to, theoretically, represent us.  And these policy factors – community choices, rather than individual choices, may be responsible for one of the great epidemiologic mysteries of our age.  Why are so many people in the United States dying young?

Life-expectancy in the US is lower than most other high-income countries.

Digging into the causes, studies have shown that most of that difference is driven by early deaths – deaths that occur before age 50. This isn’t a failure to provide good healthcare to the very old – it’s a failure to protect people in the prime of their life.

But the United States is a big place, and there is substantial variation in life expectancy from one state to another. Someone born in Mississippi in 2019 can expect to live an average of 74.4 years. Someone born at the same time in Hawaii can expect 80.9 years. Why?

According to this paper in PLOS One, it may be those big factors – policy choices – that explain the difference between your Hawaiis and your Mississippis, between your Connecticut’s and your Oklahomas.

Researchers examined policy choices states made over time across eight domains: Criminal justice, Taxes, Environmental policy, Firearm safety, Healthcare, Marijuana, private labor policy like the minimum wage, and tobacco taxes.

Each state was given a score in each policy sector ranging from 0 to 1, on a conservative to liberal axis. For example, you can see that by 2014, my home state of Connecticut had the most liberal environmental policies in the country, but was middle of the road on gun safety policies.

With that set up, the researchers could do something pretty cool. They linked the relative liberalism of the state policies to state death rates – focusing specifically on deaths between ages 25 and 65 – that age bracket we seem to be failing. What’s critical here, is that they model each policy domain, accounting for all the others. This is something that is not often done in this type of study. A study linking minimum wage laws to state-level deaths may find an association, but the level of the minimum wage in a state correlates with a lot of other policies – blue states are blue after all – so it can be really hard to interpret the results. By adjusting for all of the policies, we can see if any individual policy has an independent effect on death.

So – the results. The researchers looked at overall mortality, as well as cardiovascular mortality and mortality due to what some have described as "deaths of despair” – suicide, alcohol abuse, and drug overdoses.

Let’s look at overall mortality.

A couple of interesting take-home points here.  Broadly, we see that more liberal policies tend to be associated with lower mortality rates, with the notable exception of marijuana, where conservative policies seem protective. Surprisingly, we see that health and welfare policies –like Medicaid expansion - which you would expect to have the greatest effect on health and welfare, have really no effect on overall mortality. And finally, we see that, if you were to believe this data, the single best thing you could do to lower mortality rates in your state would be to enact stricter gun control measures.

This figure summarizes the results of the study across the various policy domains and among other causes of death.  I’m showing you the results for women here, though the men are broadly similar. Again, broad benefit for liberal policies except for marijuana legalization.

The authors then engage in a hypothetical. How many working age lives would be saved if the entire US went “maximum liberal” or “maximum conservative”.

According to this analysis, the Blue America scenario would lead to about 86,000 fewer deaths a year among working-age adults. The Red America scenario would lead to around 108,000 more deaths. And a mixed scenario (mostly blue but with stricter marijuana policy) would save 92,000 deaths.

OK do you buy it? I mean, if you’re running for office as a democrat, by all means, use this study. It’s not bad science. But there are a lot of limitations here. First off, this study is subject to the ecological fallacy. This is to say that we have no way to know which policy, if any, led to which death, if any.

Although this analysis adjusted for multiple policies, there’s two big things they didn’t adjust for – deliberately – according to the authors.  Median income and education level.

The authors state they did not adjust for state resident’s education levels or income because these might lie along the causal pathway from a policy choice to the benefit.

And yes, you should not adjust for things that lie on the causal pathway. But are we sure this is the causal pathway? Do policy choices lead to benefits via the education level of the state residents?

Or, could it be that more educated voters tend to enact more liberal policies. And, as has been shown before, education level does associate with longer lifespan. Classic confounding.

I tend to favor the latter model and would love to see an analysis that fully accounts for these key characteristics.

Still, despite the limitations of this particular study, I feel that studies like this are important in medicine. All too often, doctors have our blinders on to the pressure society puts on our patients. Sure, I can treat your blood pressure, but I can’t give you a living wage, childcare so you can find a better job, or a feeling of safety at home. Those things do matter, even if they are tricky to quantify, and it’s time for physicians to take a broader look at the world outside our clinic windows.

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