NEW YORK (AP) -- A new study offers more compelling evidence that life expectancy for some U.S. women is actually falling, a disturbing trend that experts can't explain.
The latest research found that women age 75 and younger are dying at higher rates than previous years in nearly half of the nation's counties - many of them rural and in the South and West. Curiously, for men, life expectancy has held steady or improved in nearly all counties.
The study is the latest to spot this pattern, especially among disadvantaged white women. Some leading theories blame higher smoking rates, obesity and less education, but several experts said they simply don't know why.
Women have long outlived men, and the latest numbers show the average life span for a baby girl born today is 81, and for a baby boy, it's 76. But the gap has been narrowing and data from the Centers for Disease Control and Prevention has shown women's longevity is not growing at the same pace as men's.
The phenomenon of some women losing ground appears to have begun in the late 1980s, though studies have begun to spotlight it only in the last few years.
Trying to figure out why is "the hot topic right now, trying to understand what's going on," said Jennifer Karas Montez, a Harvard School of Public Health sociologist who has been focused on the life expectancy decline but had no role in the new study.
Researchers also don't know exactly how many women are affected. Montez says a good estimate is roughly 12 percent.
The study, released Monday by the journal Health Affairs, found declining life expectancy for women in about 43 percent of the nation's counties.
The researchers, David Kindig and Erika Cheng of the University of Wisconsin, looked at federal death data and other information for nearly all 3,141 U.S. counties over 10 years. They calculated mortality rates for women age 75 and younger, sometimes called "premature death rates," because many of those deaths are considered preventable.
Many counties have such small populations that even slight changes in the number of deaths produce dramatic swings in the death rate from year to year. To try to stabilize the numbers, the researchers computed some five-year averages. They also used statistical tricks to account for factors like income and education.
They found that nationwide, the rate of women dying younger than would be expected fell from 324 to 318 per 100,000. But in 1,344 counties, the average premature death rate rose, from 317 to about 333 per 100,000. Deaths rates rose for men in only about 100 counties.
"We were surprised" by how much worse women did in those counties, and by the geographic variations, Kindig said.
The study wasn't the first to reach those conclusions. Two years ago, a study led by the University of Washington's Dr. Christopher Murray also looked at county-level death rates. It too found that women were dying sooner, especially in the South.
Some other studies that focused on national data have highlighted steep declines in life expectancy for white women who never earned a high school diploma. Meanwhile, life expectancy seems to be growing for more educated and affluent women. Some experts also have suggested smokers or obese women are dragging down life expectancy.
The Murray and Kindig studies both spotlight regional differences. Some of the highest smoking rates are in Southern states, and the proportion of women who failed to finish high school is also highest in the South.
"I think the most likely explanation for why mortality is getting worse is those factors are just stronger in those counties," Murray said, adding that abuse of Oxycontin and other drugs also may add to the problem.
Some also think the statistics could reflect a migration of healthier women out of rural areas, leaving behind others who are too poor and unhealthy to relocate. That would change the rate, and make life expectancy in a county look worse, explained Bob Anderson of the CDC's National Center for Health Statistics
"We shouldn't jump to the conclusion that more people are getting sicker in these geographic areas than previously," he said.
But that is open to debate. Migration didn't seem to affect male death rates. Murray disagrees with the theory, saying he has tracked a great deal of movement from urban areas to less-populated counties.