Today’s blog is a re-post of a blog that I did about a year ago with some slight edits. I wanted to repost it because I discussed it at my talk at Google today (see how I worked in that I gave a talk at Google today…) and I really think that it’s very representative of a huge problem that we have getting true information about our health and I have a bunch of new readers since a year ago so I wanted to give it a little more attention:
I read the headline “Obesity’s Hidden Job Costs – 73 Billion”. I’m ginormous and as or more productive than anyone I know so my first thought was “How did they come upon that number?” I looked up a bunch of different articles online to make sure that they were all reporting the same basic thing, and they were.
So I went to the Journal of Occupational and Environmental Medicine and paid $20 for the article, which was titled “The Costs of Obesity in the Workplace.”
Allow me to attempt to elucidate, it won’t be easy because this study is kind of a cluster F&$#. I’ll start with some basics, and include the longer more detailed nerdy explanation (that I love so much) at the end.
The study looked at three factors: Medical Cost, Absenteeism (not showing up to work) and Presenteeism (being at work but being unproductive). According to my spell check, presenteeism isn’t a word but we’re going to go with it for now.
According to the articles that I read online, the study was out of Duke University. That’s true – the lead author is Eric Finkelstein, an associate professor at Duke-National University of Singapore.
Under “acknowledgments” it says “This study was supported by Allergan, Inc.” It lists that as an acknowledgment, and that’s a problem for me because it should be listed as a conflict of interest. Why?
“Supported” here has the meaning of “funded by”.
Allergan is a pharmaceutical company. They produce Botox, Latisse (it will grow your eyelashes and don’t worry, that eye discoloration is probably temporary) and…wait for it…the LapBand. The item used to constrict your stomach as a weight loss surgery option.
Allergan is currently using this study from the good people of Duke Singapore to convince health insurance companies to encourage and pay for lap band surgery because it’s “cheaper than the loss productivity”. Astonishingly [sarcasm meter is a 9 out of 10] according to this study, the cost of obesity per person was more than the cost of lap band procedures. Let me muster up some shock… Sorry – I’ve got nothing.
The study used two sources:
The 2006 Medical Expenditure Panel Survey where BMI data is self-reported, and the 2008 National Health and Wellness Survey which is a series of self-administered internet-based questions fielded by 63,000 members of an internet based consumer panel. Every piece of information is self-reported and unverified. The $73 billion is an estimated projection based upon statistics that were created by doing computations with statistics and estimates, and statistics of other statistics. There are issues in the collection of data, the control of variables, the use of data (the study authors feel that the words “obesity” and “health problems” are scientifically interchangeable, that fat people’s health problems may be assumed to be caused by fatness, and utilizes BMI which has any number of problems as a statistic in and of itself which I talk about more here), and the conclusions that they drew.
The fact that news agencies reported this information as true without bringing up the limitations and issues is deplorable. Studies are suggesting that obese people already make less than their peers and are turned down for jobs based on prejudices about weight. Now companies may think that they have scientific proof to back up their bias. How many of them are going to pay $20 to read and understand a complicated study? [Edit: Since this was published friend of mine in HR have sent me e-mails from their bosses linking to an article about the study and asking "is it legal to just not hire fat people?"]
As always, may I suggest that you DO NOT need to believe everything that you are told is science, and DO NOT need to take this personally or allow it to affect how you feel about yourself, your health, or your productivity in any way.
Here come the details:
The calculation of this is statistically complicated because of the data. To speak in the vernacular of the peasantry, the data sucks. Basically they used a two part estimate that created four categories of overweightness (hey, if they can make up words so can I) based on the self reported weight. “Normal weight” were the omitted reference group They controlled for race, household income, education, insurance coverage, marital status and smoking. They subtracted the average predicted medical expenditures for obese individuals in each category from the average predicted expenditures for those of normal weight. Then they multiplied that estimated number times the number of people in each category and added them up and extrapolated based on the estimate of obese Americans.
First of all, notice the number of times that the words estimate, average, and predicted appear in that explanation. If I had more free time I would be doing a word count to give you an exact percentage of the number of words that are used in this study that essentially mean “um, maybe…”, there are many.
They also didn’t control for any genetic health issues, or health issues that aren’t even correlated to weight. They appear to have assumed that any medical problems that obese people had over and above what normal weight people had were due to fatness. They appear to have assumed that normal weight people’s health issues weren’t related to the same things that cause weight problems in overweight and obese people. That’s just embarrassingly bad science.
Absenteeism and Presenteeism
This is my favorite. These were measured based on two questions that asked people “During the last seven days, how many hours did you miss from work because of your health problems?” and “During the past seven days, how much did your health problems affect your productivity while you were working?” Participants indicated their level of work impairment via a rating scale ranging from 0 to 10. Each response was assumed to represent a percentage reduction in productive work. Then they took a week’s data and extrapolated it to a year, multiplied it by the estimated number of obese working people, and then monetized the predictions using wage data from the bureau of labor and statistics.
Respondents weren’t talking about how much work they missed or productivity they lost due to their weight, they were answering about their health problems. What they can reasonably conclude here is that people with health problems have more absenteeism and presenteeism than do people without health problems. The study’s authors are basically substituting “obesity” (and by that they actually mean high BMI) for “health problems” . You can do that I guess, but you probably shouldn’t do it while calling yourself a scientist.
Then, they computed statistics using statistics, and statistics of statistics. Dude. They used a 7 day sample to calculate a year’s worth of data. Once again, they assumed that any absenteeism or presenteeism over and above what normal weight people had was due to fatness. Except that overweight men reported less presenteeism than normal weight men. That was not reported in any news outlet that I could find EXCEPT the study itself and they gloss over it.
If I had turned this work in for my very first intro level freshman research methods class I would probably have failed the assignment and possibly been asked to leave the program because of specific incompetence and general stupidity. You don’t have to feel bad about yourself – you are fine. Feel embarrassed for the scientists who put their name on this. I hope that they are the laughing stock of Singapore.