02 January 2011

Unhealthful News Day 2 – pudgy babies and deciphering insider code

The usual health scare reporters have been hyping a not-quite-junk but grossly-over-interpreted study about heavy babies.  As is often the case, the reported hype about how this is a public health crisis is garbage, but some general lessons lie in the details.  There is no evidence in the research about bad health outcomes, or even a reason to predict future overweight adults.
Obesity research is probably one of the most junk-filled areas of junk public health science.  (Completely random aside:  It looks like “junk” has the inside track to be chosen the word of the year for 2010.  I will do my part, by continuing my current practice of using it far more than I ever did before 2010.)  

Here are a few extra clues that bode ill for the quality of the study:  It was published in the “American Journal of Health Promotion”.  If you do not know, “health promotion” is a label often adopted by the activist political faction within public health that is the most nanny-ish (or health-nazi-ish, if you want to be more combative).  The people who use that label tend to be the ones who give public health a bad reputation among most of the population (and they are so self-involved that they do not even realize it).  So when you read the term “health promotion”, hold on to your wallet – or more to the point, guard your liberty and watch for a lack of scientific integrity, though you might want to hold your wallet too, because the “health promotion” types also suck up far more of your tax dollars than most public health researchers.
In addition, the lead author is associated with the Wayne State University School of Social Work.  If you think public health schools are rife with political activism pretending to be science (and you would be correct), you should see social work.  Note that I say “associated with” because he self identified as adjunct there, but a search reveals nothing about him online, at that school or elsewhere, since he completed his dissertation at the same school in 2002.  Thus it is difficult to figure out what his political angle might be, or whether we should trust his judgment for that matter.
It is almost certain that none of the reporters hyping this story have any understanding of what obese means in this context.  Basically it is an arbitrary point chosen based on the distribution that existed among children at some point in the past.  None of the reports I saw even mentioned which of the various historical distributions were used in the study, which is pretty crucial (the U.S. from the 1970s and some global numbers from the 1990s are common, and both are populations that were much less well nourished than 21st century Americans).  I remember trying to help a student refute the use of these charts to exaggerate the rate of stunted children in Latin America, where the racial phenotype just runs smaller than white Americans.  Also, the cut off points from those past distributions were not chosen because we knew that they demarcate particularly elevated risk for anything.  They were just based on what happened to exist in the data someone had and the fact that we evolved with five fingers on each hand (i.e., picking numbers just because they happen to divide by five or ten).  Like I said, this subfield is mostly junk.  (Careful readers might wonder why I did not go to the original article to figure out which historical growth chart they were using.  The answer is that this crucial information was absent from the free abstract and I'll be damned if I am going to spend $25 to buy access to a junk science article, and the journal is obscure enough that it was not available through the online service I have access to.)
At least one commentator picked up on an incredibly obvious criticism of what was written, though he was wrong about what was wrong.  The news stories have reported that obese means “above the 95th percentile” and “at risk for being obese” is “above the 85th percentile” and thus – the commentator pointed out – it is obviously nonsense to claim that over 30% of the kids were in this category.  By definition, exactly 15% of the population is above the 85th percentile.  The problem is that “above the percentile” in this case refers to the percentiles on the historical distributions, and thus has nothing to do with actual percentiles – it is just an arbitrary line and the terminology is dumb.  The fact that the researchers in this area and the reporters blithely use terminology that is arithmetic nonsense speaks rather badly of them.  It is pretty clear that they are following a recipe or blindly repeating what they are told without pausing to think “wait, does that even make sense?”
Also on the topic of jargon, it is worth noting that the core science of public health, epidemiology, uses the term “at risk” to mean what it really means in natural language:  it could happen to this person in the future.  I.e., someone is at risk of prostate cancer if he has a prostate and does not already have the cancer.  Thus, everyone is at risk of obesity unless they are already obese.  “At risk” was adopted as a badly-chosen euphemism for “high risk” in some politically charged areas, presumably because it sounded nicer to say someone was “at risk” of winding up in prison rather than he has a high risk of that occurring.  But in the case of the obesity research, this is even worse than using politically correct terminology rather than proper scientific terminology, because it is not even clear that those in the “at risk” category really are at particularly high risk of moving up to the top category.  A proper term for the category would have been “heavy” or “almost obese”, or my personal favorite, “who's a pudgy wudgy little guy”, but apparently those do not sound scary enough.
It should be clear, then, that knowing that many babies are in these arbitrary categories tells us nothing.  The observed pattern might represent a health problem.  But as far as we know, it might represent a growing hardiness of our kids thanks to better nutrition, and they will be healthier in life as a result.  Sometimes more weight is bad, sometimes it is good.  Of course, this is not how the news reporters (or the author) spun it.
Finally, the main study result is something that is simply incredibly obvious:  That babies that are at the high end of the scale at 6 months are somewhat more likely to be at the high end of the scale at 24 months.  Shocking!   Who would have ever thought that being fat one day means that you are more likely to be fat the next day?  Or that generally being a hefty kid tends to persist rather than changing from month to month?  The study did not show that babies are getting heavier right now because it only looked at one birth cohort.  And that cohort was 2001, so it is not even that this is current information.  It certainly did not predict any real health problems.
So here is a good general lesson:  Think about whether the actual study result that is stated qualitatively (i.e., that something exists or that one quantity is bigger than another, without putting numbers to it) is actually information that you did not already have.  A study that shows that some chemical can be found in people’s bodies is uninformative, because a little bit of basically every chemical that exists free in the world is in everyone’s body.  The “discovery” that people who have a particular characteristic are likely to have closely related characteristics (overweight in January and overweight in February; tries one drug and tries another drug) is not a discovery.  If what the study "discovered" is something we already knew, then whatever the researcher or media try to hype, it is a political agenda not the finding of the study.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.