Are We Really Losing 1% of GDP Due to Poor Health? Also, a Poll on Polling

(Digital Vision)

We’ve been writing a lot about obesity recently. First, it was this study about projected future obesity rates, then we covered Denmark’s saturated fat tax, which Steve Sexton then criticized for being inefficient. So, if you’re tired of reading fat-related posts on our blog, I get it. But as long as reports like this one from Gallup keep coming out, we’re going to keep writing about them, especially when they include so many interesting conversation points.

Here are the top-line numbers:

About 86% of full-time American workers are above normal weight or have at least one chronic condition. These workers miss a combined estimate of 450 million more days of work each year than their healthy counterparts, resulting in an estimated cost of more than $153 billion in lost productivity per year. That’s roughly 1% of GDP.

Respondents were asked the question: “During the past 30 days, for about how many days did poor health keep you from doing your usual activities?” Here’s a sample of the responses:

  • 13.9% reported being normal weight and not having chronic conditions. They reported an average of .34 unhealthy days per month.
  • 30.2% reported being overweight or obese, as and having one to two chronic conditions. They reported an average of 1.08 unhealthy days per month.
  • 17.8% reported being overweight or obese and having three or more chronic conditions. They reported an average of 3.51 unhealthy days per month.

The data comes from Gallup’s Well-Being Index, which interviews at least 1,000 U.S. adults every day, and asks them questions pertaining to their health to keep a real-time measurement of our overall “well-being.” The result is this chart showing self-reported well-being scores over the last four years.

Though we started out 2011 all right, our well-being has been on the decline since about June, roughly the peak of the stock market before its latest swoon. In terms of the survey data, check out the fine print on Gallup’s interview methodology:

Results are based on telephone interviews conducted as part of the Gallup-Healthways Well-Being Index survey Jan. 2-Oct. 2, 2011, with a random sample of 270,695 adults, aged 18 and older, living in all 50 U.S. states and the District of Columbia, selected using random-digit-dial sampling. Of these, 109,875 were employed full-time at the time of the interview and they provided height and weight data to calculate body mass index scores.

Yikes, that’s only a 40% full-time employment rate for the initial sample of people who actually answered the phone. Even the most ghastly under-employment stats showing the rates of underemployed (those working part-time who want to work full-time) aren’t anywhere near that high. There’s more:

Each sample includes a minimum quota of 400 cell phone respondents and 600 land-line respondents per 1,000 national adults, with additional minimum quotas among land line respondents by region.

As The Economist pointed out last year, there’s reason to believe we’ve entered a “dark age” of polling recently, since it’s gotten so much harder to track people down. Calls to people’s land-line no longer capture a broad enough sample, since roughly a quarter of Americans are now cell-phone only. Not only is it more expensive to poll people by cell phone, it’s also not as effective, since “cell-phone onlys” (polling industry jargon) are less likely to take part in a poll.

Just how accurate a measure is this? Gallup appears to have more than compensated for the “cell-phone onlys” by using a 40% cell-phone polling baseline. But are the people taking these calls somehow more likely to be unhealthy and to have missed work? Were they perhaps on a sick day when they answered the phone? Think about it: How often do you answer a call on your cell phone from an unknown number while at work? Come to think of it, just out of curiosity, I’d be interested to know how many readers have received a call from a polling firm. So how about a poll:

[poll id=”21″]

Joshua Northey

Back when I had a land-line I would get polling calls (I think I dropped the land-line in 2001). I have never gotten one on my cell phone.

Also as you suspect I do not answer calls on my cell phone at work unless it is my wife, a call I am expecting, or someone I haven't spoken to in a long time. I certainly do not answer unknown numbers. Then again I probably only answer 40% of the calls from unknown numbers even when I am home (it depends on my mood).

As for:
"These workers miss a combined estimate of 450 million more days of work each year than their healthy counterparts, resulting in an estimated cost of more than $153 billion in lost productivity per year. That’s roughly 1% of GDP."

I am extremely skeptical of that kind of analysis. By that type of analysis you can find things that add up to a cost of 10,000% of GDP. Or factors which contribute 10,000% of GDP. Basically you can find anything you want. I have seen very convincing numbers that say if we spent twice as much on health-care GDP would go up 16% (which is frankly ridiculous). Of course they were produced by the health-care industry and if you pick around the edges they fall apart completely.

In this particular case some portion of the extra absenteeism is surely due to other factors which correlate with obesity, but are not directly related.

I don't take phone based polling very seriously at all. When my grandparents have participated in 20x the phone based polls I have (they mention them fairly regularly) over a similar period you know there are huge sampling issues.



Another thought on Gallup's polling methodology: It seems that it's likely to be significantly biased towards the unhealthy/sedentary population, 'cause those of us who are healthy/active are probably out doing something instead of sitting at home.

I'm also curious as to how GDP is figured in cases like this. Say a person is off a week with a kidney stone, an example I pick because I remember that the cost of a typical lithotripsy is about $15K. The economy has lost a week's productivity from that person, say $1K, but hasn't it statistically gained the $15K for the lithotripsy?


Broken window fallacy. GDP would not be increased 15-fold by giving everyone a lithotripsy. The medical procedure merely diverts from production of other goods and services that otherwise would have been consumed. The economy would not be on its production possibilities frontier.


I think you're getting into what we might call a zero-sum fallacy, which is the incorrect belief that if $15K wasn't spent on a lithotripsy, it would all be spent on other goods & services. But in fact some of us (me, for sure) might not actually want to spend that $15K on "stuff", and would save/invest it instead, thus reducing GDP.


I don’t take phone based polling very seriously at all. When my grandparents have participated in 20x the phone based polls I have (they mention them fairly regularly) over a similar period you know there are huge sampling issues. Swiss watch trends is great brands


My biggest concern is with the particular polling questions used - I'm overweight, and I have multiple chronic conditions that are in no way related to my body weight. My blood pressure, average heart rate, cholesterol, and blood glucose levels are all perfectly normal, but I have chronic joint problems due to a rare genetic condition called Ehlers-Danlos Syndrome. I also have asthma and have had for 20 years - this is another thing influenced by the EDS.

My being overweight and my having unhealthy days due to my chronic conditions are two entirely separate things, and one does not significantly influence the other - and it certainly doesn't cause it!

Mike B

I'll answer your poll if you call me on me cell phone.

Mike B

What's interesting is that I got on some sort of list of people who respond to polls and now I get called for polls all the time. I also get paid to take mail and online surveys. If polling firms are resorting to lists of people who are known to respond to polls then there has got to be some degree of bias creeping in.

Mike Lemmer

I suspect the sample's unemployment rate is so high because those of us with jobs (and thus only 5-6 hrs a day to do things) would hang up the moment we heard the word "poll". I'd also like to know what time they conducted this poll. I wonder how much the survey unemployment rate would vary if the same poll was done at 2 PM Wednesday, 7 PM Wednesday, and 2 PM Saturday.

Enter your name...

I'm overweight, and I have four chronic medical conditions (three quite mild).

But my doctors have confirmed (I asked) that my weight has created exactly 0% of my health problems, as all four are known to be 100% genetic. In fact, one of them actually makes being overweight even less likely to cause healthy problems than it would for a normal person, because my lifetime risk of developing ischemic heart disease is dramatically lowered.

I had a friend who was obese and had two chronic medical conditions—but the medical problems came first: she was badly injured in a bike wreck, resulting in chronic pain and disability. But she was at a perfectly healthy weight and had no medical problems until that car slammed into her.

Another example: A friend developed an incurable autoimmune disease. A couple of months of prednisone caused her to gain 30 pounds, which put her into the overweight category, and because the disease kept her from moving her hands, and her job was all about fine motor skills, she couldn't work for months. But the cause was the autoimmune disease, not being overweight.

So if you're looking at this and saying, "Wow, being overweight or obese causes you unhealthy", you need to think about the noise in the sample. The person's weight may be irrelevant to the actual health problems, or it may be the result of the health problems. It would have made far more sense for them to ask, "Do you have any chronic medical conditions? And *at the time those health problems started* (not today!), what was your weight?"



Of course there's noise in the sample, but when we filter out the noise, we find that there's a very large signal. Same as for driving while drunk or talking on a cell phone: not every drunk or cell-phoning driver causes an accident every time they drive, but that doesn't mean that they don't, overall, cause lots more accidents that the base population.


I'm thinking that if you took ANY 86% of the full-time working population, you'd show a drastic difference in combined missed days. I've been overweight all of my adult life, yet had five years of perfect attendance at one point--the best in the company!

Until we check on things such as, say, number of children, age, sex, and so forth, we can easily brush it off on obesity (and this is not to say that obesity is not a factor!). I'm betting it's more than that that is being represented here. We can tease it out more, perhaps finding that underweight men between the ages of 25-35, and who are unmarried, have a college education, and drive a Prius, are much more likely to be sick, etc.?

Henry Lahore

Economic Impact of Obesity in the United States - Aug 2010 PDF is on-line
$215 billion (vs Gallup $153 billion) due to
Medical Costs
Presenteeism (less productive if sick on the job)
Premature mortality
Additional cost of health insurance for the obese
Total indirect costs
Transportation costs
Reduced education