In 1995, journalist Gary Taubes published an article in Science titled “Epidemiology faces its limits”, which questioned the utility of nonrandomized epidemiologic research and has since been cited more than 1000 times. He highlighted numerous examples of research topics he viewed as having questionable merit. Studies have since accumulated for these associations.
In July of 1995, journalist Taubes published an influential news article in Science titled, “Epidemiology faces its limits.” The main emphasis was that epidemiology was “straying beyond the limits of the possible no matter how carefully the studies are done” because nonrandomized studies are so “plagued with biases, uncertainties, and methodological weaknesses that they may be inherently incapable of accurately discerning... weak associations.” The science of epidemiology, Taubes argued, studied a “mind-numbing array of potential disease-causing agents,” yielding an onslaught of “constitutionally contradictory” results.
Nonetheless a new study, Epidemiology beyond its limits, in the same magazine (Science: https://www.science.org/doi/10.1126/sciadv.abn3328 ), systematically evaluated current evidence of 53 example associations discussed in the article. Approximately one-quarter of those presented as doubtful are now widely viewed as causal based on current evaluations of the public health consensus. They include associations between alcohol consumption and breast cancer, residential radon exposure and lung cancer, and the use of tanning devices and melanoma. This history should inform current debates about the reproducibility of epidemiologic research results.
In this article, the authors highlight that Epidemiology has experienced tremendous innovation over the past few decades. The proliferation of large prospective cohorts and cohort consortia has allowed for better study of rare exposures and outcomes with adequate power rather than relying on retrospective case-control studies, which were the target of much of Taubes’ criticisms. These cohorts and consortia also allow for replication of studies across independent populations and in different contexts, which can help inform the impact of confounding (for example) on associations (confounding by health care access would presumably be less concerning in countries with universal health care).
Advances in technology have resulted in an explosion of data and the ability to readily link data across sources such as electronic health records, biobanks, vital records, disease surveillance registries, claims and administrative data, geospatial data, and mobile data. These can be used on their own for conducting research (e.g., studies using nationwide health registries or administrative databases) or linked to enrich cohort or case-control study data.
[...]
In recent years, causal inference methods have come to the forefront of epidemiology with increasing use of causal diagrams [e.g., directed acyclic graphs] and study designs [e.g., negative control and Mendelian randomization (MR) studies, difference-in-difference methods, and regression discontinuity designs] to rule out noncausal associations.
[...]
Thus, casual evaluation of risk factors, especially for those where randomized controlled trials are infeasible or unethical, will continue to require synthesizing evidence across various sources [see textbox 3 from Krieger and Davey Smith for triangulating evidence from eight different epidemiologic study designs for the example of smoking and low birthweight]. This triangulation of evidence is necessary to overcome the shortcomings of all study types from mechanistic studies and cohort studies to randomized controlled trials and meta-analyses.
Many of the associations selected by Taubes as examples to denigrate epidemiologic research have proven to have important public health implications—as evidenced by policy recommendations from reputable national and international agencies to reduce risks arising from the associations. The utility of epidemiologic research in this regard is all the more impressive when one remembers that the associations were selected because Taubes thought they would prove to be false positives. Twenty-five years later, epidemiology has reached beyond its limits. This history should inform current debates about the rigor and reproducibility of epidemiologic research results.
References:
G. Taubes, Epidemiology faces its limits. Science 269, 164–169 (1995).
Lauren E. McCullough, Maret L. Maliniak, Avnika B. Amin, Julia M. Baker, Davit Baliashvili, Julie Barberio, Chloe M. Barrera, Carolyn A. Brown, Lindsay J. Collin, Alexa A. Freedman, David C. Gibbs, Maryam B. Haddad, Eric W. Hall, Sarah Hamid, Kristin R. V. Harrington, Aaron M. Holleman, John A. Kaufman, Mohammed A. Khan, Katie Labgold, Veronica C. Lee, Amyn A. Malik, Laura M. Mann, Kristin J. Marks, Kristin N. Nelson, Zerleen S. Quader, Katherine Ross-Driscoll, Supriya Sarkar, Monica P. ShahIris Y. ShaoJonathan P. SmithKaitlyn K. Stanhope, Marisol Valenzuela-Lara, Miriam E. Van Dyke, Kartavya J. Vyas, Timothy L. Lash. Epidemiology beyond its limits. Sci. Adv., 8 (23), eabn3328. • DOI: 10.1126/sciadv.abn3328
Comments