Thursday, October 24, 2013

Who will remember, Big Boy?

Old guys taking testosterone to help stiffen up (their muscles) is a new fad, reports the NY Times.  And even not such old guys.  And why not, if it can fix their "Low T", make them young and full of energy again, play sports like they're 30, return their sex drive to their studly days?
 
Well, not so fast.  The effectiveness and use and advertising and all that are now being called into question.  Testosterone products "are increasingly being sold as lifestyle products, to raise dipping levels of the male sex hormone as men age. . . . .The market for testosterone gels evolved because there is an appetite among men and because there is advertising" and so on, as one might guess.  That is, manufacturers have once again created a disease for which they can sell a high-priced cure.  Has this lead to overuse?  Needless cost on something that doesn't work?

Virile Jupiter, statue at the Louvre (Wikipedia)

These are questions that are being asked, to resistance by the manufacturers and promoters of testosterone-for-everyone, of course.  But they are separate from the issue we'd like to raise here.

Who will remember?
We are not particularly qualified to judge the questions of current use or abuse patterns, nor the effectiveness of T-therapy, nor the personal or vested interests involved in all the great claims about these substances.  There's a lot one could say about those topics.  But instead, our point for mentioning the story here on MT is a different one.  It is one that applies to epidemiological inference, and hence our understanding of genetic and general public health causation.

Testosterone therapy is a current fad, that according to the story is taking many forms.  Some are through the usual channels of the medical system and may be documented, though how thoroughly is not clear.  But apparently much of the T-therapy usage is of the informal, over-the-counter type. Regardless of the hormone's actual properties or uses, legitimate or otherwise, much of that usage will largely be undocumented or under-documented.  Yet it can have substantial effects on the health and physiology of the users--if the claims are true.

However, lifestyle factors other than motorcycling without a helmet do not cause disease right away. If they did, hardly anyone would expose themselves to those factors.  Instead, these behaviors and exposures cause no or few immediate symptoms (or may give good immediate outcomes as the story suggests many believe), or -- and mainly -- they cause symptoms or disease decades later, combining with the accumulation of all sorts of behaviors and exposures.  Even smoking takes decades for its major health effects to arise.

Years from now, given the epidemiological research fads of that time, how well will exposure to testosterone be remembered or documented?  If their effects are dramatic, we will know about them--though even then it is hard to disentangle cause and effect.  But if they are subtle, as most lifestyle factors are, will they even be included in survey studies of exposure to risk-factors?  Even if they prove to be a substantial risk factor, how accurately will the timing, duration, and level of doses be remembered?  Particularly if users were on the young end when they partook. 

Big Data resources, regardless of their privacy-invading potential, will document what's known, and Big Data miners with future NIH grants to find whatever they can without having to think much about it in advance, will generally only be able to examine the ore that's in the mine.   Unreported, unremembered, transitory, low-level, or inadvertent exposures will be statistical 'noise' in the system.  Or, worse, exposure will be correlated with all sorts of other behaviors and lifestyles--confounders, the bane of epidemiology--and supposed causal associations will be spurious.

Epidemiology often relies on interviews or what was measured (and how it was measured) in widespread clinics and offices over decades.  Asking about your current diet may tell us something, but dietary risks are usually manifest much later and only if your current diet reflects what you ate long ago, even as a child, will diet questionnaires be veryinformative .  This is widespread or even typical experience in epidemiology, generating serious data reliability issues that lead to the common ephemeral pronouncements about risk that we see in the news media almost every day.

In the 1950s, hula hoops made their debut and were very popular for a few years, far more than they are now.  The fad pretty much died out, though it may be resurging temporarily these days.  The kind of hip and spine rotation that hula hooping involved could, in principle, have later impact on arthritis or other bone disease (we're just being hypothetical, here).  But who would include hula-hoop use patterns in a lifestyle survey?  Instead, we'd look for genes!

For those of us who have been around for a considerable time, we have seen fads like testosterone salves come and go, often very quickly.  But if physiological set-points, mutations, or other sorts of lasting effects result from the exposure, it will to a great extent disappear from memory, often leaving the field to other, even lesser, confounding but measured factors that happen to be current at the time of the study, decades after the relevant exposures.

This is another way to illustrate the rather deep issues we face when it comes to important aspects of scientific knowledge, in a wide array of areas. 

No comments: