The Trouble With Ranking Life-Expectancy Numbers

A widely reprinted recent Associated Press story painted a dire picture of the U.S. health-care system. Headlined “U.S. Slipping Down Life Expectancy Rankings,” the article reported the U.S. had fallen to 42nd in the world in life expectancy in 2004, down from 11th two decades earlier.The story isn’t quite so simple. Multiple data sources confirm that the U.S. has slipped in these rankings since the 1980s. But the United Nations and the U.N.’s World Health Organization, which each maintain their own life-expectancy numbers, have the U.S. ranking higher overall.Among the complicating factors: The AP’s international data source, the U.S. Census Bureau, includes nearly 30 more places in its rankings than are covered by the U.N. and the WHO. Many of these places are territories or small parts of other countries. Nonetheless, the AP’s article began, “Americans are living longer than ever, but not as long as people in 41 other countries.”

Seven of the places ranking ahead of the U.S. in 2004 had a population under 50,000, including Montserrat and San Marino, while another 10 had populations under 500,000, including the U.S. Virgin Islands and Andorra. (Andorra topped the ranking with a life expectancy of 83.5 years.) Such places have so few deaths each year that their mortality numbers are subject to big swings. Meanwhile, 18 of the 41 places ranked ahead of the U.S. by just a year or less, a small difference.

Comparing the latest data with two decades ago also obscures that the U.S.’s ranking hasn’t budged much in recent years — it was 41st in 1997, by the AP’s ranking method. (And keep in mind, as I wrote in a recent post, that life expectancy is more complicated than the popular perception that it predicts how long someone born today is likely to survive.)

Other life-expectancy trackers show the U.S. ranking higher, thanks in part to the absence of dozens of territories that aren’t independent countries. The U.N., which hews closely to its list of 192 member states for its data collection and reports life expectancy over five-year periods, shows the U.S. at 35th in 2000-2005, down from 31st in 1995-2000 and 17th in 1980-1985. The U.S. ranked 31st in the WHO’s 2005 numbers.

The Census’s own data included far fewer countries in the 1980s (just 133 in 1984, compared with 222 in 2005), which helps account for the U.S.’s big decline from 11th in that decade.
There are other differences between the data sets. Each agency has its own statistical models and supplements data supplied by countries, where available, with its own considered judgment. This can result in some discrepancies — not so much for the U.S. and other large countries with reliable health statistics, but for some nations alongside the big ones near the top of the rankings. The WHO, for instance, shows a shorter life expectancy for Albania (by about five years) and Andorra (three years), and a longer one for Monaco (two years) than does the Census Bureau.

“We wouldn’t pay too much attention to [life-expectancy data from] Andorra and Monaco,” Mie Inoue, a Geneva-based WHO statistician involved in producing the agency’s life expectancy numbers, told me. “There are very few deaths, so there are lots of fluctuations, so the result wouldn’t be very reliable.” Added Thomas McDevitt, chief of the population studies branch within the Census Bureau’s population division, “For small areas, constructing life tables is a challenge.”

The Associated Press combined the Centers for Disease Control and Prevention’s domestic figures with the Census Bureau’s world numbers (explained here). “I combined the most accurate U.S. numbers available with the most accurate numbers available for the rest of the world,” Stephen Ohlemacher, author of the AP article, told me. “Several researchers recommended this strategy, and several others agreed it was the best way to get the most complete and up-to-date numbers for the U.S. and the world.”

Asked why he included places that aren’t countries, some of which are very small, he replied, “I relied on the Census Bureau’s expertise on these issues.” Of the non-countries, Mr. Ohlemacher added, “I believe that many of those areas are listed separately because their demographic characteristics are different than the rest of the country. Think Hong Kong and China.”

What do you think? What should be included and excluded from such rankings? Are differences of a year or less significant? What’s the best way to compare the health of different countries? Please let me know in the comments.

Further reading: Speaking of life-expectancy comparisons, New York Magazine recently reported that New Yorkers live longer than their counterparts around the country.