National Center for Complementary and Integrative Health (NCCIH)
NIH…Turning Discovery Into Health

Información en Español

Health Topics A-Z

NCCIH Research Blog

Reduced Mortality Risks and Correlation vs. Causation

July 22, 2015
Josephine P. Briggs, M.D.
Josephine P. Briggs, M.D.

National Center for Complementary and Integrative Health

View Dr. Briggs' biographical sketch

Recently, I noticed news articles about a study on chamomile consumption and its potential effect on mortality. Researchers from the University of Texas Medical Branch, Galveston, analyzed data from a population-based study of older Mexican Americans in Southwestern states and reviewed 7-year all-cause and cause-specific mortality. They found a 29 percent decreased risk of death (all-cause mortality) among chamomile users compared with non-users, and concluded, for women at least, that this difference was statistically significant. It’s good news that using this herb is associated with longer life. Nevertheless, let’s remember that, even with statistical corrections, correlations don’t prove causality.

The authors reported that the decreased risk was statistically significant for women after adjusting for age, smoking, chronic conditions, and other known confounding factors. This is the right approach to the data. But it is not enough. Statistically correcting for other factors only captures the impact of measured differences. The women who use chamomile may differ in many ways from those who do not. For example, the use of herbal tea may be a marker of a “healthier” lifestyle. In their paper, the researchers note that “other unmeasured factors, such as frequency and duration of chamomile, level of physical activity, and quality of diet, which were not measured in the survey, could influence the results.”

This study illustrates the challenges of observational (non-experimental) research. This type of research helps us find patterns and signals and can raise new and interesting hypotheses. But let’s remember that there may be a big gap between correlation and causation.



Howrey BT, Peek MK, McKee JM, et al. Chamomile consumption and mortality: a prospective study of Mexican origin older adults. Gerontologist. April 29, 2015. Epub ahead of print.



Comments are now closed for this post.

Your comments on correlation and causation are of course correct.  Too bad the media doesn’t register a similar concern when an observational study yields a negative correlation for a dietary supplement.

A 29% decrease in risk of death is very important to me.  The comment that this is the right approach but not enough other factors are considered  is interesting.  Patients do not live in a test tube that allows observational research.  I  suggest Pubic Health recognize evidence based results to a greater degree than I have observed during my 16 year cancer journey.  I am less concerned about a gap between correlation and causation then the concept of “do no harm” within the medical profession.  Many doctors practice medicine by treating symptoms and few, in my opinion, are truly interested in healing.

Thanks for your intelligent comment. People eschew critical thinking and then blame others when this lack of rigorous analysis gets them into trouble. Journalistic science reporting is guilty of much of this intellectual laziness. That is a pity when they need only work a little harder at their trade——writing!

This page last modified February 10, 2016