A correspondent in the MedStats email discussion group (RR), mentioned an interesting example of problems in defining groups in observational studies. The actual publication is
- Kaye Fillmore et al. “Moderate alcohol use and reduced mortality risk: systematic error in prospective studies.” Addiction Research and Theory. Advanced online publication March 30, 2006.
I cannot find this particular article yet, but there are several news reports about this study that are on the Internet.
- http://sciencenow.sciencemag.org/cgi/content/full/2006/330/1
- http://www.eurekalert.org/pub_releases/2006-03/uoc--isq032706.php
- http://www.emaxhealth.com/63/5236.html
- http://www.medpagetoday.com/PrimaryCare/DietNutrition/tb/2959
- http://news.monstersandcritics.com/lifestyle/consumerhealth/article_1151367.php/Study_dilutes_alcohol%60s_heart_benefits
The basic finding of the original research is that much of the data that appears to show that moderate drinking is mildly protective might be flawed in that the comparison group, those who abstain from alcohol, includes a significant number of people who are only abstaining because they are ill and their doctor advised them to stop drinking. Thus, it might be the ill health that precipitated the abstention rather than the abstention leading to slightly worse outcomes. If you look at studies that define abstention as never having consumed alcohol, the difference in risk compared to moderate drinkers disappears. Of course, both groups always look much better than the heavy alcohol consumption group.
I’ll try to find the original research article and produce a summary here. It is an interesting example of how defining a control group without the benefit of randomization can sometimes lead to faulty conclusions.
You can find an earlier version of this page on my website.