Errors in Medical Studies


The Harvard Medical Center and Aetna maintain a very interesting web site known as Intelihealth®. One particular article, by Dr. Robert H. Shmerling, M.D., Beth Israel Deaconess Medical Center, deals with the issue of errors in medical studies. The doctor laments the fact that many hyped "groudbreaking studies" end up nowhere, and the stories are replaced with other "breakthroughs". We agree with Dr. Shmerling, and have dedicated this site to providing "unhyped" information.

One well-known "error" involved the recommendation that menopausal women continue with Hormone Replacement Therapy - an incorrect recommendation resulting from the lack of inclusion of all the "confounding factors" (factors other than HRT, that impact women's health). The doctor notes incomplete knowledge of counfounding factors as one major reason for errors in studies.

Another error is due to the finite duration of studies. People often take medication a lot longer than the duration of typical studies. Safety problems may become evident only after many years of use. Likewise, people change over time, their medications (and their interactions) change, and consequently the impact of the original medication might be different than planned.

The last error listed by Dr. Shmerling involves the limitations of the question asked. For instance, researchers might ask whether a medication reduces cholesterol, because of the known correlation between cholesterol and heart disease. However, the fact that the medication reduced cholesterol does not necessarily mean that it impacts heart disease. Other studies involve animals, and then the question is whether the results can be applied to human beings. Last but not least, there may be subtle biases due to the funding - a pharmaceutical company may phrase the question in a way that will favor use of the drugs they have developed.


Last Modification - November 20, 2005