Three years ago, with childhood obesity on what seemed to be a scary rise toward infinity, Philadelphia was among the first cities to report a short-term drop.
With that tentative finding replicated, more or less, around the country, a new Philadelphia study has found longer-term evidence of a reversal in the 30-year trend.
Considering that obesity rates tripled over that period, the recent changes are small and the burden remains large: 20.3 percent of city public school students were obese in 2012-13, down from 21.7 percent in 2006-07.
In the frustrating world of obesity prevention, that's something.
The most interesting findings involved subgroups. Boys' obesity rates declined nearly twice as much as girls'. African American and Asian students also experienced sharper drops than other races and ethnicities.
Hispanics of both sexes had the highest obesity rates, and prevalence among Hispanic girls actually increased slightly over time.
The study did not examine reasons behind the trends. But the authors, all with the Philadelphia Department of Public Health, noted that no other community had found drops among African American students.
"There were interventions in the public schools in Philadelphia that particularly reached African American kids," coauthor Giridhar Mallya said in an interview Friday.
As far back as the mid-1990s, he said, the School District began working with the federal government to make entire schools with high poverty levels eligible for free or reduced-cost meals. It implemented comprehensive nutrition standards a decade ago, years before the Obama administration mandated a similar change nationwide.
"So they got as many vulnerable kids into the system [as possible], and then they improved the nutritional quality of the meals," said Mallya, the Health Department's director of policy and planning.
The study - published in the online journal Preventing Chronic Disease - found an average decline in obesity prevalence of 6.3 percent over seven years.
It found an even greater drop, 13.9 percent, in severe obesity.
Severe obesity can cause immediate health problems, such as high blood pressure, whereas obesity is considered more of a risk for future difficulties.
The authors speculated that the reasons that Hispanic students' obesity rates had changed the least (boys were down 1.7 percent while girls were up 1.5 percent) were because they were exposed, according to national studies, to more fast food and were targeted by Spanish-language advertising.
And Mallya guessed that differing levels of exercise might help explain the larger decline in obesity among boys overall (8.1 percent) compared with girls (4.3 percent). A small Philadelphia study of fourth, fifth, and sixth graders several years ago found that girls were five times less likely than boys to get the recommended 60 minutes of daily exercise, he said.
Stella Lucia Volpe, nutrition sciences department chair at Drexel University, said research had found that girls exercise less as their bodies begin to mature. She worries about the long-term consequences nationally as schools cut physical education to focus on academics.
Volpe, who had not read the new study, said that small changes made in Philadelphia, like removing sodas from schools years ago, can have significant benefits.
To determine the prevalence of obesity districtwide, the new study relied on height and weight measurements, which are used to calculate body-mass index, taken by school nurses every year. It did not follow individual students, although a shorter-term previous study did, with similar findings.
The student population has declined and changed over the seven years, and the percentage of children who were actually measured dropped significantly, perhaps due to layoffs of nurses.
Those limitations, along with relatively small percentage-point changes, prompted obesity researcher Robert C. Whitaker to interpret the study's findings more conservatively.
"I'm not certain that things have gotten better, but I think it says a lot that things have not gotten worse," said Whitaker, a Temple University professor of public health and pediatrics, who was not involved with the study.
Whitaker also wasn't sure that school districts can do much more to reduce obesity. To really drive down rates, he said, will likely require interventions in neighborhoods and homes, where children spend most of their time.
"Feeding and activity and sleep are probably the three core behaviors that have evidence of affecting obesity risk in children," he said, and they all involve parenting.
Changing policy in school districts has been hard enough, Whitaker said. "Try doing that in households, and to respect autonomy when you are trying to change the culture."