From Dose-Response to Cause-and-Effect: Statistical Reasoning and Cancer Etiology

• In Event: Cancer's Contested Causes: The Science and Politics of Research into Exposures, Genes, and Populations

English Abstract

As researchers were repeatedly stymied in the hunt for simple causal accounts of cancer over the first half of the twentieth century, government regulators and scientists increasingly turned to statistics as a tool for regulating putative carcinogens. Though the statistics of randomized experiments (such as clinical trials) have been well studied, the arguably far more influential concept of dose-response curves has been largely overlooked by historians of science. That’s perhaps because the curve’s statistical properties were settled with the publication of the maximum-likelihood solution in 1938. Between the 1940s and the 1970s, however, dose-response calculations became the central method by which the effects of various “risk factors” or etiological agents were assessed, particularly the low-dose exposure to known toxic compounds that eventually drew the attention of the nascent Environmental Protection Agency. These new approaches were required because traditional research methods (e.g., animal laboratory studies and epidemiological studies) failed to generate expert consensus around carcinogens. Dose-response curves instead provided a clear, step-by-step quantitative method—essential for the formation of regulations—for combining and translating evidence of carcinogenicity into “virtually safe” doses for human exposure. The paper concludes by arguing that the history of dose-response also helps reshape our understanding of statistics in the twentieth century, as a discipline transformed from a practice of description, to one of experimentation, to one of expert-consensus formation. The discipline’s methods enabled studies to be aggregated and practical responses to be established despite ongoing epistemic uncertainty surrounding the causes of cancer.

Author

• Christopher J Phillips, Carnegie Mellon University