What impact could such “failures” of observational study have on public health today?

What impact could such “failures” of observational study have on public health today?

 http://www.nytimes.com/2007/09/16/magazine/16epidemiology-t.html

For those of you who have been enlightened and impressed about how large a role epidemiology plays in the world we live in, be prepared for a skeptic’s viewpoint.

One of the most important features of epidemiology is that much of it is observational. When researchers conduct an observational study, there are possible sources of bias and confounding which may explain an observed relationship between exposure and disease. This is inherently the most challenging aspect of epidemiology – determining whether an observed relationship is a true one, or if any threats to validity are untenable. You have learned about how bias and confounding can affect the interpretation of study findings, and the possible sources of bias specific to experimental and observational studies. Taubes outlines some of the key concerns about bias in modern epidemiological studies, including the hormone replacement therapy debate, and presents a case for why experimental designs should be the “gold standard” in epidemiological research.

As you read the editorial by Taubes, keep in mind all that you have learned about confounding and bias and why epidemiology is limited or not limited by its observational nature.

Things to think about – this editorial characterizes some common criticisms of epidemiology. How might some of Taubes’ points be countered? What impact could such “failures” of observational study have on public health today?