New research isolates an old foe — smoking — as the principal culprit behind U.S. midlife mortality gaps defined by place and education.

Working with colleagues from Dartmouth, the University of Pennsylvania, and the Federal Reserve Bank of Boston, Ellen Meara of the Harvard Chan School sought to shed light on the gap in mortality among Americans 25 to 64, which widened from 2.6 years in 1992 to 6.3 years in 2019. The work included close scrutiny of several potential drivers, including “deaths of despair,” the changing composition of college graduates, and globalization. The variable that best fits the evidence, the researchers say, is tobacco use: “Smoking emerges as an exceptionally powerful predictor of mortality trends.”

In this edited conversation, Meara, Richard L. Menschel Professor of Health Economics and Policy, discusses why and where smoking habits linger, to devastating effect.

Previous theories for the midlife mortality gap — deaths of despair, income inequality, education — were accepted by many. Why not you?

These explanations had not really been tested. So, we started with three puzzles. One, the education gap in mortality is widening, which other researchers have shown. Two, place matters a lot more than it used to. Our earlier work showed that in 1990, for example, residents of Arkansas, Ohio, and New York could expect pretty much the same lifespan. That has really changed since 1990. Third, also about place, people in rural America now die at significantly higher rates than those in cities and suburbs — a new and growing divide.

We wanted to understand these puzzles. We found that people with college degrees are becoming more alike. It doesn’t matter where you live; college graduates are living longer. Regarding health behaviors, college graduates quit smoking quickly after the 1964 Surgeon General’s report. In nearly every U.S. county, about 5 percent smoke.

What about those who didn’t graduate college?

It’s more complex. In cities, mortality rates have fallen for folks without college degrees; you also see big declines in smoking. But in small towns and rural areas, smoking rates remain high.

What role do other health factors — obesity, diabetes, lack of exercise — play in this story?

We were a little surprised. We thought more things would matter in this calculus. Obesity is one health risk that people raise, but while smoking is uniformly low among college graduates — in contrast to the noncollege population — obesity rates are high and vary widely among college graduates. So, obesity does not explain mortality differences by place and education. Since rising obesity is a more recent trend, I do worry about its contribution to mortality inequality in the future.

One thing we do is rule out some common hypotheses that simply do not fit the data. Some people attribute the widening education gap to changes in who completes college. Or they say migration explains the trends: The healthy people leave rural areas. We looked at both explanations and the data don’t fit. The percent of people finishing college has changed little since 1990, yet gaps in mortality by education have widened substantially. Similarly, accounting for migration patterns does not explain mortality trends. These findings are important. Now we can focus on the question: “What is it about rural residents that leads to high rates of smoking and rising death rates?”

And, as we look at younger generations, will mortality disparities fall, as we predict, or is smoking just a stand-in for something else?

“Deaths of despair” — largely understood as mortality linked to suicide and alcohol and drug abuse — is another hypothesis that you say doesn’t answer this question. Why not?

Deaths of despair are clearly an important cause of death at ages 25 to 64, but not as important in this “place” story. Even removing “deaths of despair,” the growing mortality divide by education and place remains large. In many high-income places, like here in New England, rising drug-related death offsets dramatic declines in deaths from other causes. Although deaths of despair contribute to premature deaths, these trends are swamped by trends in mortality due to causes like cancer or cardiovascular disease, especially among people older than 50. And since the vast majority of midlife deaths occur after 50, deaths of despair do not explain the growing mortality inequality across places.

How can researchers and/or policy makers build on these findings?

We are very interested in rural-urban differences. Historically, rural areas were always healthier. This is part of the tragedy. We took the healthiest parts of the country — in a country as rich as ours — and not only are they not enjoying the same gains in longevity, but they’re seeing shorter lives. Smoking is a very effective marker for where places are struggling, that’s why we’re trying to understand the underlying factors that may explain strong links between smoking and deaths in those areas. There are likely underlying factors, alone or in combination, that trigger both persistent smoking and, in other ways we do not yet understand, lead to premature deaths among these populations.

This gets a lot of press attention but I’ve been stunned by how hard it is to get funders or the government to go after this problem. If a disease doubled your chance of death, we’d treat it like a four-alarm fire. Yet adults living in states like West Virginia, Oklahoma, and Alabama are twice as likely to die in midlife as residents of Minnesota, California, and New York, and few resources are available to find causes or remedies for these disparities. I look at these differences and think, “No wonder our country’s divided.”