Chemicals that cause cancer can’t be accurately studied

Nicholas Kristof in the New York Times writes about a new report from the President’s Cancer Panel calling attention to the role common chemicals may play in the development of cancer.

The overarching message is that we should be exercising much more caution in our trust of chemicals.

I am not familiar with the nuances of regulatory policies for chemicals, but (as Kristof points out) the “existing regulatory presumption [is] that chemicals are safe unless strong evidence emerges to the contrary.”  This approach seems insane to me.

We don’t approach the regulation of drugs this way, why are other chemicals different?  Unfortunately, changing this attitude is going to be nearly impossible given the heavy influence of chemical company lobbyists.

As important as I think the message of Kristof’s article is, I’m actually linking to this story to highlight an epidemiological problem inherent in almost any medical research attempting to link a chemical and cancer.  All such studies are observational, not experimental.

Ethically, we can’t randomize people to ingest what we think may be toxic chemicals.  Additionally, the development of cancer is slow.  Even if we ignored the ethical concerns, participants would need to be followed for decades.  Such follow-up is expensive and frequently suboptimal (participants are often “lost” when they move or simply drop-out of the study for various reasons which can bias results).

Fortunately, case-control studies were designed to overcome these very problems.  The drawback to case-control studies is that it’s difficult to measure a person’s exposure to a given chemical.  It’s generally impossible to measure it directly and asking somebody if they used a product years ago is obviously not very accurate.

Research has had success in linking some chemicals to specific diseases.  Such success has been largely limited to instances of occupational exposure.  Why?  Simply because it is easy to measure somebody’s exposure to a chemical if it happens in the work place.  Employment records can pinpoint when I person worked in a specific position and by re-creating their working conditions you can measure how much daily exposure they received.  Multiply this by the amount of time they worked there and you can easily create a relatively accurate picture of their exposure.  These studies have also been successful because workers generally receive huge doses of these chemicals and manifest severe symptoms much more quickly than the general public which may have a tiny fraction of the exposure.

This report from the President’s Cancer Panel and Kristof’s subsequent op-ed are likely to engender what may become a vitriolic debate about chemicals and their health consequences.  Undoubtedly, chemical proponents will point out that the research linking chemicals and cancer has methodological problems.  They will not be incorrect in this criticism.  When deciding for yourself, realize there are some inherent problems in trying to establish a direct, clear-cut, causal link between a given chemical and its possible impact on human health.

We can’t do a randomized clinical trial injecting one group with BPA to see if they develop cancer.  Also realize that we didn’t have to do a randomized clinical trial of smoking to establish its link to lung cancer.  As more researchers tackle this challenging problem, more evidence will emerge and recommendations will be revised.

In the meantime, I will be trusting the advice of some accomplished doctors and not that of big chemical companies.

Josh Herigon is a medical student who blogs a Number Needed to Treat.

Submit a guest post and be heard.

Comments are moderated before they are published. Please read the comment policy.

  • David

    The title of this blog post seems to belie your personal view of the dangers of chemical exposure. It may have done your viewpoint a disservice to title it thus.

    Simply because one cannot conduct a placebo controlled, randomized, double-blinded, prospective study does not mean that credible data cannot be obtained. I rebut, chemicals that cause cancer CAN be accurately studied. We need not fall further into the trap that the study method I described is the only one or the most valid. In fact, studies of this sort have provided cover for bad data in some very prominent cases recently. It shows that the process can be corrupted at many levels.

    Fortunately, as you say, more attention is being given to this matter. Intuitively, 80,000+, largely unregulated, chemicals, commercially available, in the U.S. can’t be very good for us.