Correlation and Causality: Alcohol and Diabetes to BPA and Estrogen Disruption

Alcohol and Diabetes:

From a purely health-oriented perspective, alcohol is one of those things where a little seems to be good and more than a little puts you at risk for a bad health outcome. One of the positive things small regular intake can do for you is protect you from diabetes. There is good epidemiological evidence that this is the case. Light drinkers are less likely to develop diabetes than those who abstain.

Samples prepared for analysis. 


As most people know, just because two things occur together does not mean that one caused the other. Correlation and causality are two different things. It is important to keep this in mind when reading about human health studies in the news. Epidemiology is a wonderful tool however, and provides clues to problems that might have gone unnoticed. In addition to identifying potential problems correlational (and epidemiological) studies are important in evaluating how a problem identified in the lab may be harming people in the real world. Are light drinkers more likely to eat better, exercise better, be healthier than those who drink more alcohol or even none at all?

An important next step is figuring out why the correlation exists.  Here is a case where epidemiology observed something (low alcohol = less diabetes, high alcohol = more diabetes) and experimental study identified a physiological reason that explains those observations.    He et al. (2007) discovered that light alcohol intake activates a biochemical pathway that has a positive influence on removal of blood sugar from the blood stream. Higher alcohol intake disrupts glucose management by activating an additional protein that blocks the positive effects seen with light alcohol intake.  This discovery helps explain why you can see two different outcomes with alcohol.  At first light it may have seemed that scientific studies conflicted each other.  The appearance of contradiction may lead some people to throw up their hands and decide to disregard health news or science in general.

Sometimes we need to keep working at the puzzle.

There may be a similar issue with environmental contaminants.  Sometimes a relationship is observed and someone publishes on it, and there there are a flurry of other studies that confirm or don’t confirm the relationship, expand it or limit it.  Identifying the mechanism and understanding how it works under varied circumstances becomes essential.  There are now many correlational studies that show that Bisphenol A (BPA) intake is associated with health problems.  There are also mechanistic studies that show how BPA interacts with estrogen receptors and causes things to go arwy.  One could stop here and take steps to remove BPA from use.  Environmentalists and children’s health advocates could claim victory.  This might result in fewer birth defects, less obesity, less diabetes, fewer behavioral problems in children etc.  But it might not.

Plastics are more complex than they seem.  Even without BPA many of the chemicals that go into or are released from different plastics may be may activate estrogen receptors.  Would people be exposed to less estrogenic chemicals if BPA is banned?  We still don’t know.  While many studies have used data on concentrations of BPA or BPA-metabolites, to the best of my knowledge no one has looked at total exposures to estrogenic chemicals originating from plastics.  (There are also estrogenic chemicals from sources other than plastics too).   Does removing BPA from a plastic actually make it less estrogenic?  Is BPA, rather than other chemicals coming off of plastics, the main source of dietary exposures to estrogenic compounds?

Plastics are amazing, really, and have made many positive and important contributions to our economy and quality of life.  For information on how plastics benefit the environment and contribute to quality of life take a look at this pdf from the British Plastics Federation.  Its taken tremendous effort and ingenuity to develop polymers.  We can figure out where they might cause problems in the environment or in human health by continuing polymer research and including within, cell and molecular studies.  We should not have to leave epidemiology to discover problems that could have been avoided after the fact. 

He L, Marecki JC, Serrero G, Simmen FA, Ronis MJ, & Badger TM (2007). Dose-dependent effects of alcohol on insulin signaling: partial explanation for biphasic alcohol impact on human health. Molecular endocrinology (Baltimore, Md.), 21 (10), 2541-50 PMID: 17622585

Nguyen KH, Lee JH, & Nyomba BL (2012). Ethanol causes endoplasmic reticulum stress and impairment of insulin secretion in pancreatic β-cells. Alcohol (Fayetteville, N.Y.), 46 (1), 89-99 PMID: 21840159

Submit a Comment