, ,

So, this has been making the rounds in fat spaces, especially on Tumblr:

Someone posted this in a Fat Studies group on yahoo.

I don’t have permission to use their name so I’m going to keep them anonymous, but this BLEW MY MIND.

“I’ve been mulling over where and when I should post this as, while I think it is an important personal testimony that gives insight into the politics of BMI research, I also still work for a related organisation and there would certainly be severe personal consequences to an overly public critique at this time.
Over the last few years, my research management career (previously researching ecological toxicity thresholds) moved into the sphere of public health and back at the end of 2010 I found myself being offered an opportunity to coordinate the latter stages of a prestigious global metabolic risks project. The project was the culmination of over a decade of concerted efforts to collect data from across the globe on blood pressure, cholesterol, blood sugar and BMI. It was (and is) hosted at a major UK research institution and is staffed by an all-star team of public health statisticians, modellers, and demographers combined with several clinical research fellows with various non-communicable disease specialisations. As research coordinator, I oversaw all aspects of grants, budgets, contracts and reporting of progress to funders. I also tracked progress internally toward specific research objectives and helped ensure all the various researchers remained on track.
So…as the data collection phase came to a close, the analysis phase ramped up a gear. Our institution has excellent access to national (and international) mortality and disease incidence data and the researchers in the team went about the process of applying statistical models to examine the data trends by region and over time. They were able to extrapolate with a very high degree of certainty the trends in each of these metabolic factors. In the case of BMI, the trends were of a general and very slow increase worldwide. Various countries had differing results and there was variation within countries by sex and by age but the general trend was clearly shown to be that the increase in BMI is slowing and will continue to slow over time and then level out.
This was not what our primary research lead had expected to find and so, where there had once been talk of publishing findings to verify the gravity of the soon-to-be-apocalyptic `obesity epidemic’, instead the directive that came down to me was that we would not seek to publish our predictions. The researchers I worked with were serious people who had invested a lot of life into their work and there was quite a lot of frustration at this decision.
The next phase of our research plan was to put our global data set to use to compare country metabolic risk (BMI, blood pressure, cholesterol) trends over time with incidence of cardio vascular disease (CVD) over time. To the greatest extent possible, all other factors were controlled for within the models used for this with the aim of demonstrating the direct relationship between CVD and each of the risk factors on their own.
In the case of cholesterol and blood pressure, there was a clear relationship between those countries with higher recorded data and their heart disease outcomes and mortality. In the case of BMI there was a very weak negative correlation. That is to say that the countries with the highest BMI levels showed lower rates of CVD than those with the lowest BMI levels… and that, generally speaking, the relationship between BMI and CVD wasn’t anything to write home about once other factors were controlled for.
This was when the primary research lead (my boss) started to get really concerned. The statistics were sound. The models were sound. The results of the models went as expected for blood pressure and cholesterol. He wanted to publish but, in his words; “the BMI lobby will destroy us if we publish a negative correlation between CVD and BMI”. I asked him what he meant by `destroy’ and his explanation was that we would be blacklisted for research money… and worse than that become a target to be discredited by `some people with a lot of money and power’ with the result that his credibility as a researcher would take a serious hit.
Kudos is everything in the prestige pyramid that is the world of medical research (and, I suspect, all high-level research). Driven by the ever-increasing scarcity of research grants, the way that such senior researchers often behave toward one another could probably be most politely described as `Darwinian’… it certainly isn’t the kind of collegial or collaborative way in which you might imagine science to be conducted.
So, from the perspective of my boss, there was a problem. He commanded the researchers to change the research specification whereby instead of just looking at the relationship between BMI and heart disease, they were to only consider the BMI data in conjunction with diabetes prevalence. This essentially created a forced relationship in the data between BMI and diabetes that ended up demonstrating a correlation that wasn’t specific to BMI. The actual relationship was one between diabetes and heart disease.
Again… for clarity… the largest data set in the world shows us that the relationship between BMI and heart disease is weak and negative. The relationship between diabetes and heart disease is positive. Countries with higher BMIs often (not always) had slightly higher rates of diabetes but there is no statistical rationale to lump the two together as a conflated variable.
No rationale… unless you have a pre-existing bias to confirm and your future research funding to secure I suppose?
End result… ten years of data… and many years of constructing a statistical model is thrown out of the window. No results are published. The world carries on assuming that there is a positive causal relationship between BMI and heart disease.
I was appalled by this. However, while I am a scientist, I wasn’t qualified to the same level of my colleagues and was aware that my own bias could have been stoking a fury in me that wasn’t entirely justified.
A week later, one of the most senior statisticians (who had lovingly – there is no other word – spent the previous 2 years building the over-ruled statistical model) decided to leave the team on account of the `farce’ that was the senior researchers decision not to jeopardize his career by not publishing the ‘negative’ result with regards to BMI. I admired her sense of principle and followed suit
(albeit one month later as soon as I had secured another job!)”

And I thought bringing this up here was appropriate, especially after my last post. Now, this is unsourced and possibly unprovable (although, if it’s real, it would probably be possible to figure out what study it’s referring to, check the published results {if they’re out} and see if they match up with this report, but I’m not going to do it, because I really don’t have time or energy), and as such we can’t simply trust it right away. But it certainly does fit the pattern. Again and again, we see studies on fat, diet, health, and lifespan all jiggered to fit the preconceived notions of what’s true and what isn’t, whether it’s to fit researchers’ ideas or the ideas of grant givers or institutions.

This story is unsourced, but both the findings and the cover-up are part of a general pattern. Again and again studies find the same sorts of things. Ones that actually publish results showing that fat isn’t bad are regularly ignored. Fat people are less likely to die of diabetes than thin people (except for OMGDETHFATS people, and the study explicitly says that hypertension accounts for it, and when you control for that, very fat people with diabetes have the same mortality rates as thin people with diabetes); fat people who have heart attacks are less likely to die of them than thin people (and there are more studies with the same results, but they’re on protected sites); and fat people have less geriatric depression and better overall quality of life in old age than thin people. And it’s been demonstrated repeatedly and in different ways that being fat is not bad for you, but repeated dieting is.

Expectations warp science, and money warps science. No scientist should go into an experiment or study determined to get a particular answer. No scientist should ever change or have to change the results of a study to match expectations, whether to support their own opinions or to ensure that they can still get grants. This is not science, this is lying with numbers.

Over and over, in articles about these studies, researchers who work in these fields express surprise at outcomes like these despite the fact that there are lots of studies out there showing that fat isn’t bad. This information is out there, and it’s findable, and it’s verified, and people still just keep ignoring it and/or lying about it and telling us that being fat is bad for you, worse than anything else, and OMG WE’RE GOING TO DIE IF WE’RE FAT OH NOES. But if they were right, they wouldn’t need to lie.