December 7, 2009
Piling on Formaldehyde
By Michael D. Shaw
Climategate is showing the world that scientists are not above fudging their results, and even using all available means to crowd out those who would dissent from their point of view. After all, there is a fortune in research grants to protect, and these days, sadly, there is not much difference between science and politics.
An equivalent story is playing out regarding formaldehyde and its supposed role in causing leukemia. Only here, dissent is not crowded out. Rather, crummy “science” is simply signed, sealed, and delivered, forcing an instant change in policy.
The International Agency for Research on Cancer (IARC) is part of the World Health Organization. IARC’s mission is to coordinate and conduct research on the causes of human cancer, the mechanisms of carcinogenesis, and to develop scientific strategies for cancer prevention and control. In a monograph published in 2006, formaldehyde was classified as “carcinogenic to humans,” putting it in IARC’s Group 1. This classification was based primarily on findings regarding cancer of the nasopharynx—a rare condition.
Those familiar with the politics of scientific research will attest that it is far easier to get grant money for what they call “sexy” projects. In this case, that means findings for more common forms of cancer, such as leukemia.
During its meetings in October, 2009, IARC updated cancer assessments for numerous compounds, including formaldehyde. Sure enough, a finding emerged for leukemia, that was widely promoted in the media. IARC itself noted, though, that “The Working Group was almost evenly split on the evaluation of formaldehyde causing leukemias in humans.” As it happens, the study that pushed this finding is controversial—to say the least—and according to IARC, “The authors [of the study] and Working Group felt this study needed to be replicated.”
Good luck finding out why IARC, and our own National Cancer Institute, breathlessly promoted the leukemia finding, inasmuch as the paper (the Zhang paper) had only been accepted for publication rather than published, and had thus not had a chance to be widely peer-reviewed. (Published paper).
Presumably, Dr. Zhang and her 33 co-authors provided the necessary blood chemistry mechanism explaining how formaldehyde could cause leukemia. The attraction of Zhang’s paper was the supposed demonstration of how earlier epidemiological data—itself extremely sketchy—could be explained at the cellular level.
It is worth noting that the only way a statistical significance for leukemia and formaldehyde exposure could be shown in the epi study was to create—out of whole cloth—the notion of “peak exposure.” Essentially, peak exposure is a self-assessed number based on some exposure event remembered by one of the thousands of workers in the study. That is, if a worker posited that on any day of his employment this peak value was, say 4 parts-per-million, whether that occurred on his first day of employment, or his last day of employment, and maybe never again in his entire term of employment, that was nonetheless his “peak exposure.”
Some people might call this data dredging, especially since more statistical trickery had to be used to develop a finding even with peak exposure. Bear in mind that extensive use of statistical analysis in scientific research calls results into serious question. The purest scientific findings do not require elaborate statistics. For example, no statistics would be necessary to show that an object dropped off a building will fall to the ground.
Zhang’s paper is based on a small cohort of Chinese workers exposed to formaldehyde, along with a non-exposed cohort. Her key results include comparison of various blood cell counts and the incidence of two chromosomal abnormalities in both groups. Fair enough.
However, you might be surprised to discover that while there were differences in the blood counts, both the exposed and non-exposed groups were well within the normal range. Yet, according to Zhang, this meaningless difference was “statistically significant.” The difference is indeed “meaningless” because if any of these workers had individually submitted to a blood test, their doctor would tell them that there is no problem, since they are in the normal range.
Good luck finding out how a normal result in an individual patient is somehow significant, if that same person is part of this cohort.
As to the chromosomal abnormalities, which occurred more frequently in a subset of “highly exposed subjects” (10 people), this again sounds impressive until you realize that these conditions (Monosomy 7 and Trisomy 8) also appeared in the non-exposed group. Moreover, Zhang deliberately reversed cause and effect here as these abnormalities are often observed in people who HAVE leukemia, rather than as some sort of predictive marker.
To further exemplify the poor quality of science that occurs in this field, Zhang references a paper in which it is suggested that exposure to formaldehyde reduced white blood cell counts in a group of 50 nurses. But, how seriously should we take a study in which certain concentration data (in parts-per-million) was reported as 0.231 ppm, 0.054 ppm, and 0.082 ppm for mean, lowest, and highest, respectively. The mean was 2.8 times higher than the highest?
Further, the exposures in this study were low enough that the “exposed” group about equaled the control group in Zhang’s study!
Some think that Zhang’s newfound interest in formaldehyde has something to do with the benzene/cancer gravy train reaching the end of the line. Perhaps she wants to create a big formaldehyde/cancer express version to replace it.
Finally, for those of you who think that any of this work is actually related to protecting public health, consider EPA’s incandescently stupid risk assessment on formaldehyde, based on its ludicrous linear no-threshold model:
EPA estimates that, if an individual were to continuously breathe air containing formaldehyde at an average of 0.065 parts-per-billion (ppb) over his or her entire lifetime, that person would theoretically have no more than a one-in-a-million increased chance of developing cancer as a direct result of breathing air containing this chemical.
Considering that human breath contains 1-2 ppb of formaldehyde, based on natural metabolism and not pollution, you might ask how this would be achievable, or what relevance such nonsense has to public health.