Follow The Prince
twitter icon rss icon

Download &
print (pdf)

Confounders and the Confounding Array of Conflicting Studies



“Without a randomized, double-blinded, intervention trial controlled for confounding variables, the debate as to the health benefits of alcoholic beverages as part of a healthy lifestyle will continue.”

Research on wine and alcohol has been confined to in vitro studies on lower life forms, or in vivo cohort or case-control studies or meta-analyses (statistical analysis) which are epidemiological studies subject to bias and confounding factors, and lead to false positive and false negative conclusions. There have been no well executed, randomized, double-blinded, intervention trials controlled for all possible confounding variables. A confounding variable, also known as a confounding factor or hidden variable, is an extraneous variable in a statistical model that correlates positively or negatively with both the dependent variable and independent variable.

Arthur L. Klatsky, M.D., a prominent researcher on the health benefits of alcohol and wine, has explained the major flaws in much of the research to date. “The major issue is residual confounding by diet, smoking and other traits, problematic statistical modeling, and under estimation of drinking amount which lowers the apparent threshold of risk. The association of a behavioral trait such as alcohol drinking with a health outcome in any study could be due to chance, due to bias (confounding), or causal. While a chance relation is always possible, statistical evaluation can render the likelihood vanishingly small. The accepted standard for elimination of confounding and establishment of likely causality is the double-blinded randomized controlled trial. Problems in blinding and prescription of long-term behavior, plus the perceived risk of harmful effects have so far precluded this type of long-term study with chronic disease endpoints. As a result, the issue of how well we can infer causality from observational data remains a formidable challenge. While even well performed observational studies cannot completely exclude possible genetic or environmental predilections to health outcomes....criteria exist that can establish a very high probability of causality in these data.”

Beyond confounding inherent in epidemiological studies, a single study may tout a result, but the research often proves to be flawed. John P.A, Ioannidis, chief of the Stanford Prevention Research Center, works with colleagues all over the world to scrutinize treatments and has found many are costly and even worthless or harmful. He concluded in a 2005 PLoS paper, “For most designs and settings, it is more likely for a research claim to be false than true.” Joan O’Connor Hamilton, (Stanford Magazine, May/June 2012, “Something Doesn’t Add Up), quoted Ioannidis who concluded, “Journalists pursue stories about cures and progress but they are not likely to diligently explain the fine points of clinical trial bias and why a first splashy result may not hold up.”

A statistical analysis of 228,220 medical research trials reported in the journal JAMA (Volume 308 Issue 16, 2012) showed that the effect of large medical interventions rarely held up under scrutiny when other research teams tried to duplicate the findings. For example, for 75 years scientists have found that rats and mice that eat 30% to 40% fewer calories than normal live 15% to 40% longer than is typical for their species. A recent study in monkeys showed no increased longevity when calories were restricted to 30% below normal. So what might be true for rodents and other animals may not hold true for primates including humans.

Valuable critiques of currently published research is now available by members of the International Science Forum on Alcohol Research, a joint undertaking of the Institute on Lifestyle and Health of Boston University of Medicine and Alcohol in Moderation (AIM) of the United Kingdom. The forum consists of an international group of invited scientists and physicians who are specialists in their fields and committed to balanced and wellresearched analysis regarding alcohol and health. The Forum provides timely critiques and comments by Forum members on emerging scientific publications and policy statements related to alcohol and health, and provides an opportunity for all to seek expert opinion on topics related to alcohol and health. Visit the website at www.bu.edu/alcohol-forum/. Some commentary by the Forum is included on a few of the abstract reviews that follow.

In the pages to follow beginning on page 12, I will summarize major, peer-reviewed, scientific studies published in 2012 and early in 2013 that contribute to our current understanding of the roles that alcohol and wine play in our health. For those wishing to seek out the studies in their entirety, visit PubMed, which offers more than 21 million citations from the biomedical literature (www.pubmed.gov), or visit the websites of the journals named.

Share