My childhood obesity story seemed doomed until I dug deeper into the data

Everyone in Merced County seemed to know that childhood obesity is a major local issue.

Food bank, health department and public health officials told me the children in the county have significantly higher rates of obesity than other areas of California.

One of every four Merced County children between the ages of 5 and 12 are obese, which is triple the national rate, according to a 2016 county health assessment. But beyond analyzing census data, no one had conducted a deep data analysis on the issue.

The county’s high poverty rate means most schools in the county give students free lunch through state and federal programs, and many students get most of their nutrition through meals such as pizza, burgers, burritos, fruit juice and chocolate milk.

That also means schools need to keep and report meticulous data. So when I joined the Center for Health Journalism’s 2018 Data Fellowship, I was hoping the data analysis techniques I would learn could help me expose whether schools were knowingly or unknowingly contributing to the epidemic.

But data is sometimes elusive. And I learned you need to take different approaches to data to tell a meaningful and accurate story.

Collecting data from schools

While requesting granular school lunch numbers and nutritional data from the 21 different school districts in Merced County, I quickly learned many of them didn’t have the data readily available.

For many school districts, my initial request yielded reports, not datasets. And I learned some didn’t analyze or even look at their own data. (A story for the future.) They had to request it from their food vendors.

One school administrator was upset he had to work to get the data in the desired format for my records request. I had to get our lawyer involved, and he ended up complying.

A few districts were cooperative, including the largest elementary school district in Merced County. So I focused my reporting on their students.

Have patience and plan for delays with data collection. It took several months to work with nutrition officers at the school district to gather the data.

One of the more valuable lessons the fellowship taught me was to ask and learn about how the data is gathered. It took me several lunchroom visits. That helped me direct the school district to what I needed.

It’s often easier and more efficient to go through accommodating sources. These school district officials were very open with their program, which helped me provide a clearer picture on the school meal program besides just the “negative stuff.”

With a deeper data project, allow for several weeks for data analysis as well. It took time to organize the datasets and figure out what aspect of the nutritional content I wanted to pursue.

Don’t just analyze, visualize

When I got the nutritional content for the average school breakfast meal and lunch meal, I immediately noticed the high sugar content.

“This was it!” I thought. The school district complied with federal standards. But students participating in school lunch and breakfast consumed an average of 68 grams, or about 18 teaspoons, of total sugar per day.

Recent research on obesity points to sugar as the major culprit, not dietary fat, which was low for the average school district meal. Dr. Robert Lustig from UCSF, a leading national voice on sugar and children’s diets, wasn’t surprised. He told me the sugar levels were very high.

I had Lustig and other researchers’ similar comments. But when I looked for any health organization’s recommendations on the total amount of sugar children should eat, there wasn’t any. The school district defended its meals, noting they complied with USDA standards.

I only found recommended amounts of “added sugars” children should eat: no more than six teaspoons, according to the American Heart Association and World Health Organization.

Added sugars are the amount of total sugars that don’t come from whole and unprocessed foods like fruits, vegetables and milk. But I learned that measure is hard to nearly impossible to get without knowing exactly what foods are being used in each meal recipe. And the school district doesn’t track it.

I tried creating my own measurement by excluding the sugars of natural whole foods like fruits, vegetables and the natural lactose in white and chocolate milks served. My added sugar measurement ended up higher than recommendations, but it still wasn’t accurate because it doesn’t account for natural sugars in cooked meals.

Maybe, I thought, the data can’t accurately conclude the school district serves too much sugar.

But my fellowship mentor, Cheryl Phillips, told me to dig deeper and visualize the data. I made some pie charts that revealed chocolate milk and fruit juice accounts for more than half the sugar consumed by kids.

A student consuming just one fruit juice at breakfast and a chocolate milk at lunch is drinking more than eight teaspoons of sugar, about five teaspoons of added sugar if you don’t count the natural lactose in the milk.

It was an important find because these sugars, nutrition experts told me, aren’t accompanied by fiber. That means they are absorbed and used by the body no different than the sugar in soda, a drink that the school district stopped serving years ago due to its sugar content.

Whether the school district should stop serving juice and chocolate milk all together is another issue. At that point, school officials said, kids could lose the nutrition in fruit juice and milk because they may not substitute it with natural fruits or other foods high in calcium. But, they said, “it is something we can look into further.”

While this wasn’t the way I envisioned the story moving, the data pointed me to a finding that was still important, impactful and insightful.

Keep analyzing the data, even if it isn’t at first cooperative.

You can check out Vikaas Shanker’s stories here.