Men’s HealthWomen’s Health

Interpreting Epidemiologic Studies

By March 29, 2019 No Comments
Download PDF

Epidemiologic findings form the basis for much of the U.S. Dietary Guidelines. But sometimes, common sense isn’t applied to the interpretation of the results of epidemiologic studies. That clearly is the case for a recent cross-sectional study by Marcelo et al.,1 that identified a positive link between urinary genistein and mortality.

The limitations of epidemiologic studies are well-known. For starters, their findings don’t allow causal relationships to be drawn. That’s because it’s impossible to identify all potentially confounding variables. If for example, a study finds that higher soy intake is associated with a lower risk of dementia, is it because of soy per se, or because soy-consumers lead an overall healthier lifestyle? Another important limitation for studies involving diet, is the challenge of accurately assessing dietary intake.

Of course, epidemiologists are aware of these difficulties and limitations and go to great lengths to mitigate their effects, for example, by statistically controlling for a wide range of confounding variables and by using validated food frequency questionnaires. These steps allow the field of epidemiology to continue to provide important insight into the diet and health relationship.

Unfortunately, while the above-referenced study by Marcello et al.1 might have been well designed and conducted, the interpretation of the results misses the mark. For this study, data were analyzed from the National Health and Nutrition Examination Survey (NHANES) from 1999–2010. Mortality data were obtained from the National Death Index and matched to the NHANES participants. Mortality rates were then compared to urinary genistein concentrations.

Of 11,497 participants, 944 died during the 64,443 person-years of follow-up. The all-cause mortality rate was significantly lower in the lowest quartile of urinary genistein compared to the highest quartile (incidence rate ratio = 2.14, 95% CI = 1.76 to 2.60). In other words, those 25% of the participants with the highest urinary genistein levels were about twice as likely to die from any cause in comparison to the 25% of the participants with the lowest urinary genistein levels.  The authors of this study noted that their finding is contrary to popular opinion on the health benefits of genistein and called for further research.

This study did not measure dietary intake although it is possible to come up with an estimate based on urinary genistein levels and understanding about the absorption and excretion of isoflavones. But that really isn’t necessary. Using NHANES (1988-1994) data, Bai et al.2 estimated adult isoflavone intake was 2.35 mg/d, and Chun et al.,3 (based on NHANES 1999-2002) estimated intake to be a little more than 1.0 mg/d. More recently, Rohrmann et al.4 estimated isoflavone intake to be less than 1.0 mg/d among caucasian, latino and black participants in the multiethnic cohort, which includes individuals residing in Hawaii or Los Angeles. Approximately one-half of the isoflavones in soybeans are comprised of genistein.

While many other surveys could be cited, there is no need, as they all show the same thing, isoflavone intake by Americans is almost negligible. Negligible, because one cup of soymilk provides 25 mg of isoflavones. So, the average American consumes the amount of isoflavones found in about ½ ounce of soymilk. For comparison, most clinical studies intervene with 50 to 100 mg/d isoflavones and it is thought to take 50 to 60 mg/d to alleviate hot flashes.5

The point is that it is inconceivable that 0.5 mg genistein per day could have any effect on anything, let alone mortality, and let alone increase risk of death by twofold. Despite this, when discussing the limitations of their research, Marcello et al.1  never once mentioned the negligible isoflavone intake.

Fifteen years ago, I commented that epidemiologic studies involving the general population of non-Asian countries are unable to provide insight into the health effects of soy.6 This is why, regardless of the findings, I don’t cite studies like the one conducted by Marcello et al.1 when discussing the health effects of soy. Unfortunately, studies like this continue to be published. For meaningful insight, it is necessary to study soy-consuming populations, be they Asians or vegetarians or flexitarians.  


  1. Marcelo C, Warwick M, Marcelo C, et al. The association between urinary genistein levels and mortality among adults in the United States. PloS one. 2019;14(1):e0211368.
  2. Bai W, Wang C, Ren C. Intakes of total and individual flavonoids by US adults. Int J Food Sci Nutr. 2014;65(1):9-20.
  3. Chun OK, Chung SJ, Song WO. Estimated dietary flavonoid intake and major food sources of U.S. adults. J Nutr. 2007;137(5):1244-52.
  4. Rohrmann S, Shvetsov YB, Morimoto Y, et al. Self-reported dietary flavonoid intake and serum markers of inflammation: the multiethnic cohort. Cancer Causes Control. 2018;29(6):601-
  5. Taku K, Melby MK, Kronenberg F, et al. Extracted or synthesized soybean isoflavones reduce menopausal hot flash frequency and severity: systematic review and meta-analysis of randomized controlled trials. Menopause. 2012;19(7):776-90.
  6. Messina M. Western soy intake is too low to produce health effects. Am J Clin Nutr. 2004;80(2):528-9.
Dr. Mark Messina

Author Dr. Mark Messina

PhD in Nutrition, Executive Director, Soy Nutrition Institute. Expert in soyfoods and isoflavones.

More posts by Dr. Mark Messina

Leave a Reply