(Washington Post illustration; iStock)

In 2014, the Centers for Disease Control and Prevention (CDC) released a promising report about American children. For decades, the collective waistline of this country's youths had only seemed to expand — between the late 1960s and early 2000s, the obesity rate among 6- to 11-year-olds quadrupled, while the rate for 12- to 19-year-olds more than tripled. Even among the youngest children, those ages 2 to 5, the rate had risen alarmingly fast — tripling. But finally things were changing.

Government researchers, using data from the National Health and Nutrition Examination Survey (NHANES), a nationally representative survey taken every year, found that obesity rates among 2- to 5-year-olds had actually fallen between 2003-2004 and 2011-2012. And not just by a little bit — over the period, the rate fell from 14 percent to 8 percent, marking a more than 40 percent drop.

The news fired off a chorus of media coverage. Tom Frieden, the director of the CDC, even suggested it showed we were capable of winning the battle against obesity.

"This confirms that at least for kids, we can turn the tide and begin to reverse the obesity epidemic," he said at the time.

It was a bold statement, rooted in emboldening data. But in retrospect, there was something off about the report, or the idea that we had really scored a major victory in the battle against obesity.

"The data they cited are perfectly accurate, but when you look at trends it matters where you start," said Asheley Skinner, who is a scientist at the Duke Clinical Research Institute, a research arm of the Duke University School of Medicine. "This is a case where you can tell two pretty different stories, but only one of them is right."

Skinner is the lead author of a new study published this month in the journal Obesity. The study used data from the same annual survey but reached a different conclusion. Despite a widespread belief, tied at least in part to the 2014 CDC report, that childhood obesity is trending downward, it argues that the opposite is true: Childhood obesity might not be growing quite as fast as it used to, but it's still growing. The progress among 2- to 5-year-olds, the study argues, is overstated.

What gives? Well, the CDC's story begins in 2003-2004, when the childhood obesity rate was significantly higher than in previous years. But the data actually go back further. And when you start earlier, the narrative changes considerably. "When you start in 1999, the first year in which the survey was taken, you no longer see the decrease, you no longer see this long-term trend," Skinner said. "Especially when you include each of the following years."

The chart below shows how the obesity rate for boys and girls ages 2 to 5 has changed over the years. It's fairly easy to see why comparing more current rates to 2003-2004, instead of looking at the broader trend, could give the impression that things are getting better faster than they actually are.

But that cursory look doesn't tell the whole story. For girls ages 2 to 5, the obesity rate was 10 percent in 2013-2014. While that's a percentage point lower than it was in 1999, the difference isn't statistically significant, according to Skinner. What's more, the rate is almost 40 percent higher than it was the year prior, casting doubt on the notion that there is a sustained downward trend.

For boys ages 2 to 5, the news is a bit better. The obesity rate was 8.5 percent in 2013-2014, almost a full percentage point lower than it was the year before, and almost two full percentage points lower than it was in 1999. But the broader dip is still far more modest than the CDC report might have led someone to believe. The study makes that very clear in its conclusion, where it says:

The significant decline previously reported in prevalence for 2- to 5-year olds for 2003–2012 is not evident in our results, for girls or boys, when using all data from 1999 to 2014.

Skinner's finding is even truer for children broadly than it is of the youngest age group. Have a look at the chart below, which shows the prevalence of class I, class II, and class III obesity among children ages 2-17. Class I is what most obesity rates, including those cited by the CDC, reflect, while classes II and III comprise what is known as severe obesity. Class I is defined as those whose body mass index (BMI) is greater than or equal to 95 percent of children of their same age group, while Class II and II are defined as those whose BMI is even greater (at least 120 percent and 140 percent, respectively, of the 95th percentile).

It's important to point out that new data have become available (specifically, data for 2013-2014), and that that is partly what makes it clearer now that the trend isn't as encouraging as previously thought. Still, the most recent numbers also underscore that it was premature to make some of the declarations of several years ago.

Why the CDC researchers decided to use 2003-2004 as their benchmark instead of 1999 is unclear. A spokesperson for the agency said it doesn't respond to or comment on research that is not its own. But Skinner is worried the choice has given credence to some unwarranted arguments, including those that suggest what we are currently doing to combat childhood obesity is enough.

"We haven't seen much movement in our national statistics, because we really haven't been doing much on the national level aside from the National School Lunch Program," she said. "The problem is so many things are propelling obesity. Kids might eat lunch at school, but then they eat at gas stations they pass on the way home, and then they eat the food that's being served at home."

Ruth Loos, who is the director of the Genetics of Obesity and Related Metabolic Traits Program at Mount Sinai Medical School, is sympathetic to Skinner's argument. In 2014, after the release of the CDC report, she told the New York Times that we should be cautious and wait until a clearer picture emerges. Today, she agrees that it's important for us to understand, both in public and in the research community, that childhood obesity rates don't appear to be falling.

"The important message is that we're not done yet in terms of battling childhood obesity, we haven't figured it out," she said. "I think Skinner is afraid previous data suggested we don't have to worry, I think she just wants to make sure policymakers don't think that that's the case when it's not."

"Do I wish the CDC report had gone back to 1999? Yes, ideally," she added. "But I don’t think it’s unreasonable to compare to obesity rates in the early 2000s. I don't think it was necessarily an inaccurate representation of the data. They may have been more worried about whether it was still going up compared to the worst times."

Until the early 2000s, childhood obesity has trended upward almost without interruption (the rise, as the CDC's own chart shows, is strikingly smooth). But more recently, there have been ups and downs, making year-to-year comparisons a bit more confusing.

Barry M. Popkin, a researcher at the University of North Carolina at Chapel Hill who has tracked American food purchases, told the New York Times he believes the fact that people are buying lower-calories foods has helped slow the trend. The change has been especially pronounced among those who participate in certain federally funded supplemental nutrition programs, such as the Special Supplemental Nutrition Program for Women, Infants, and Children, thanks to new regulations that limit the purchase of certain unhealthful items.

Still, when one digs into the numbers, there are certain trends that are ominous no matter how one chooses to present the data. Take the rate of severe childhood obesity, for instance, which is shown in the chart below. It has been on an upward trajectory since 1999, when it was 4.9 percent. Today it's almost twice that.

"Almost one out of every ten children are now severely obese in this country," Skinner said. "Think about that. These are kids who can't just grow out of it — they actually have to lose weight."

"Now consider that obesity rarely gets better, that it almost always gets worse," she added. "Obese children tend to become even more obese adults."

It's hard to sugarcoat the broader picture. Nearly a third of children in the United States are considered overweight, according to the Food Research and Action Center, and roughly 35 percent of them go on to become obese in adulthood. While the the problem isn't specific to the United States — childhood obesity has proved to be a growing problem for many countries around the world — it has crippled this country in ways it has yet to cripple others. The United States is still, by almost any measure, one of the fattest countries in the developed world.

There are few culprits as obvious as the modern American diet, which has only seemed to widen the gut — and worsen health. A study published this past December, which looked at what has happened to the children of Mexican immigrants, makes this painfully clear. The kids who adopted an Americanized lifestyle exhibited significantly less-healthy eating behaviors than those who did not.