In 2015, a blockbuster study came to a shocking conclusion: Middle-age white Americans are dying at younger ages for the first time in decades, despite our advances in medical technology and the positive trends in other wealthy countries.

The research, by Princeton’s Anne Case and Angus Deaton, highlighted the links between economic struggles, suicides, and alcohol and drug overdoses. Since then, researchers have been scrambling to fully explain the trend, which now seems to be affecting the entire population. The efforts have suggested it’s not just “deaths of despair”— from opioids, alcohol, and suicides — that account for the dip in life expectancy, but that violence and cardiovascular disease seem to be major contributors, too.

Now, a new study provides another clue about what’s behind the backward sliding of American mortality: the hidden toll of diabetes.

Diabetes’ prevalence has exploded in the US over the past 20 years. Nearly 30 million Americans live with the disease today — more than three times the number in the early 1990s.

And researchers have long known that diabetes is an underreported cause of death on death certificates, the primary data source for determining life expectancy trends. That’s because people with diabetes often have multiple health conditions, or “comorbidities,” such as cardiovascular disease, high blood pressure, high cholesterol, obesity, and even cancer.

When both diabetes and heart disease are listed on a death certificate, the decision to list diabetes as the primary cause of death is “highly variable,” said Andrew Stokes, assistant professor of global health at Boston University’s School of Public Health. “Often times, the [death certificate] certifier will code the death as being caused by heart disease rather than a death from diabetes,” he added. “So to some extent, deaths that should be attributed to diabetes go to other causes.”

For new research in the journal PLoS One, Stokes and his co-author Samuel Preston of the University of Pennsylvania decided to look at more granular administrative records and surveys — the National Health Interview Study and the National Health and Nutrition Examination Survey — to find out whether they could estimate how often diabetes might be the primary but overlooked cause of death.

To come up with an estimate, they calculated the prevalence of diabetes in the population and the increased risk of death among people with diabetes during five years of follow up.

Their results were astounding. While death certificates usually suggest that around 3.5 percent of deaths were caused by diabetes, the researchers found the number may be closer to 12 percent — three times higher than the typical estimates. Among obese individuals, the death rate from diabetes was even higher, at 19 percent.

That means that while diabetes is generally listed as the seventh most common cause of death in America, said Stokes, their results suggest it’s probably the third leading cause of death after cancer and heart disease.

“When we look at that surprising decline in life expectancy in 2015, and argue about the causes, we believe diabetes is somewhat obscured from this debate because it doesn’t jump out in the mortality statistics,” Stokes said. “For that same reason it may not be implicated in trends as much as it should be.”

As a result, relatively fewer health dollars and less policy attention are focused on diabetes compared to other more obvious contributors to this health crisis. If Preston and Stokes are correct, though, that needs to change.

“Mortality has been improving unusually slowly for about the last eight to 10 years,” said Preston. “We know the opioid epidemic is part of the problem. But I don’t think it accounts for all of the difficulties we’re facing. And an obvious place to look is obesity and diabetes.”