Friday, August 26, 2011

Thoughts on the Behavioral Risk Factor SURVEILLANCE Study

The Behavioral Risk Factor Surveillance System (BRFSS) is a telephone survey conducted every year. It's been growing and growing every year, and it gives a lot of people (including me) a data woodie every year in April when the annual data dump comes out.
Nominally, the survey is about "behavioral health", things like smoking & drinking, seatbelt use, exercise, diet, getting your cholesterol checked and a mammogram done. It has become a cornerstone of our data surveillance infrastructure - used to track progress against the Healthy People goals, and to reiterate endlessly repetitive health disparity analyses.

In 2010, 429,630 people responded to the survey. That's almost the population of Wyoming. Sure Wyoming is the least populated State, but do we really need to call that many people every year to look at trends in how often people smoke, use seat belts, and eat five fruits & veggies a day? Imagine talking to each and every resident of Kansas City, Missouri in one year, asking them some 75 questions about their personal behaviors. That's the scale of this thing.
Since 1988, there have been almost 5 million interviews - about the population of South Carolina, or the combined population of the 7 least populous States: Wyoming, Alaska, North & South Dakota, Delaware, Vermont & Montana.
And it has been growing at a rate of about 8-9% a year, which means that is has been doubling in size about every 7 years or so.

One of the ways that BRFSS is complicated is that it tries to have about the same number of responses from each State. Think Senate vs. House of Representatives. So whereas the response burden in New York, Illinois and California is a relatively manageable 1 in 2,000 or so residents getting called in any one year, in New Hampshire and Hawai'i, about 1 in 200 people have to answer this survey every year. And in Vermont, it's as low as 1 in 92 people! That means you Vermonters probably know several people who get surveyed this year, and given the survey growth rate, it will be almost impossible not to get interviewed at some point in your life.

The record goes to the Virgin Islands, though, where about 1 in 35 people get surveyed every year. We're going to know every detail about every resident of the Territory before long!

This raises a couple of issues for me. There's the inevitable risk of some hacker breaking into the State Health Department and snagging detailed information on tens of thousands of State residents linked to their phone numbers. Although that's a scary idea, it doesn't get me too exorcised, because there is probably very little value to that information - it is hard to imagine who would want to know about your dietary habits, or even drug use or sexual behavior.

Another issue is just the level of surveillance, or monitoring of the population. Gathering information from a small number of people to keep tabs on trends in the population as a whole makes sense, but it seems to me that BRFSS is getting out of control, moving towards a degree of surveillance that is quite intrusive on a high proportion of the population. I mean, at this rate, it might make more sense to just mass mail the survey to every State resident every five years or something like that.

But the biggest problem I have with BRFSS is only tangentially related to it's size. It's the fact that it asks really boring questions. How that's related to its size is that by becoming the largest health survey in the country by an order of magnitude, BRFSS is where lots of people will look for answers to what is causing our public health problems. It's the centrality of "behavioral health" that I've got a gripe with.
Let's take obesity as an example.
BRFSS can demonstrate in very great detail the growth in obesity rates over time, in very minute detail. But it can't tell us much of anything about why any individual or group is getting heavier. Partly that's because it's a prevalence survey, so there is no way to track individuals over time (talk about invasive surveillance). There is no way to know if an overweight person became overweight recently, or even if they have lost a lot of weight recently. But mostly it's because the questions are boring.
You would think that we'd have learned by now that asking people about their behaviors doesn't tell us much about behavior change. And even when we do learn something about behavior change, we have learned that such efforts are incredibly difficult, time-consuming and often barely effective, especially when they are administered at an individual level.
I won't pretend to know what's causing the growth in obesity rates, but I can assure you that the answer won't come from asking 2 million more people about their exercise habits and vegetable intake. If the answer lay there, we'd have licked the problem a long time ago.
One thing I guarantee that asking 2 million more people about their behaviors will accomplish is cementing in the minds of most public health researchers and practitioners that the answers to our public health issues lie in personal behaviors - and the corollary to that is that any health problems you have are because you have failed yourself. Is that the message we really want to send?

1 comment:

  1. Hi Bill,

    Thought you might enjoy this post:

    Much less polite than yours ;)