What’s Wrong With That NYT Story About Fat Kids?

July 26, 2008

Not much – as long as you’re willing to do your own research and your own math.

A front page story in today’s New York Times doesn’t seem to add up – until you figure out that the missing statistic in the story is the fact that there are about 80 million children in the U.S.

Stephanie Saul’s story says

Hundreds of thousands of children are taking medication to treat Type 2 diabetes, high blood pressure, high cholesterol and acid reflux – all problems linked to obesity that were practically unheard-of in children two decades ago.

But a chart accompanying the story shows children taking Medco prescriptions for diabetes at a rate of 1.2 per 1000. The chart says the database has 600,000 children in it.

Do the math: If there are 1.2 kids on meds for every 1,000 kids, then in a database of 600,000 there are only 720 kids on meds. Right?

If you do the same math for the rest of the chart, you come up with 12,840 kids on fat pills in total. That’s a long way from “hundreds of thousands.”

The story goes on to say that the Times also used data from Express Scripts that was presented at the American Public Health Association last November. After a search, that study seems to be the one cited here (scroll down) and here and here and here.

(Even if it’s not the same study, it’s a good proxy because it covers kids on diabetes meds over the same time period.)

That study presents similar math: It has 4,333,344 kids in its database. It then says that the prevalence of kids on diabetes meds is just over 2 per 1000 in 2005 for type 1 diabetes, and 0.6 per thousand for type 2.

The back of the envelope says: (2 x 4,333) + (.6 x 4,333) = 11,265.

Again, a long way from “hundreds of thousands.”

So where does this story get its beef? The story mentions in passing that its numbers come from “extrapolating.” As we saw recently with the ludicrous JAMA “women and Viagra” study, extrapolating has its problems. But the answer is that it is only when you take these per-thousand rates and multiply them across the roughly 80 million kids in America, that you start to get hundreds of thousands, and then only in two categories: blood pressure and heartburn (665,600 and 944,000, respectively).

It would have been nice if the 80 million number was in the story, because the piece makes no sense unless you know this already.

Advertisements

4 Responses to “What’s Wrong With That NYT Story About Fat Kids?”

  1. Merrill Says:

    Way to dissect the numbers! I thought you did such a good job, I linked to it on my website. Welcome back!
    Merrill

  2. Jennifer Says:

    Of course they extrapolated. Are you suggesting that they should have a database with the info of all 80 million American kids in it?

  3. jimedwardsnrx Says:

    I’m only suggesting that the story makes no sense unless you already know that the extrapolation is done through all 80 million US children. I’m also slightly curious as to why, if there are “hundreds of thousands” on diabetes meds, only 720 show up in the Medco database and only 8666 show up in the Express Scripts database. These are not small companies with small databases … Just to be extra cynical, the 944K kids on heartburn pills could be as much a function of the trend toward childhood coffee drinking (think Starbucks) as it is about obesity. I think Saul buried the bit that she actually nailed: 665K kids on blood pressure meds, a number that appeared nowhere in the story. That’s the lead.

  4. veinglory Says:

    Jennifer, you can’t extrapolate on the assumptionthat kids not on the database, and those that are, are equally likely to be medicated, or indeed obese.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: