Blog | Tuesday, March 9, 2010

Coffee and the Heart - Researchers are Getting Paid Way Too Much to Rehash Old Data

This post by Westby Fisher, MD, originally appeared at Better Health.

This week, coffee seems to be good for the heart: "People who are moderate coffee drinkers can be reassured that they are not doing harm because of their coffee drinking," said Arthur Klatsky, the study's lead investigator and a cardiologist at Kaiser's Division of Research.

These "surprising" data were presented at the American Heart Association meeting on March 5th.

Valentine’s Day Coffee by Damian Cugley via FlickrBut a quick Google search on Dr. Klatsky's earlier studies using the same questionnaire database shows the problems with using questionnaire data to make such sweeping conclusions. Take, for instance, these findings from 1973: Coffee drinking is not an established risk factor for myocardial infarction.

And yet a bit later, in 1990, there's a flip flop: Because of conflicting evidence about the relation of coffee use to coronary artery disease, the authors conducted a new cohort study of hospitalizations among 101,774 white persons and black persons admitted to Kaiser Permanente hospitals in northern California in 1978-1986. In analyses controlled for eight covariates, use of coffee was associated with higher risk of myocardial infarction (P=0.0002). (By the way, British researchers failed to find a similar correlation in instant coffee drinkers.)

So what, really, do these data from the Kaiser questionnaire data regarding heavy coffee consumption and the heart say?

What they say is:
1) Questionnaire data crunched to suggest correlations are insufficient to mean causation, irrespective of how the media parses it.
2) Questionnaire data are subject to significant sampling and reporting biases.
3) Rehashing the same old questionnaires using the same samples with newer data can dramatically alter prior findings.
4) Researchers are getting paid way too much to keep rehashing the same data for large health systems.
5) On the lighter side, college undergrads and medical students should note that they could use these types of questionnaire data to justify significant caffeine consumption along with alcohol to protect themselves from developing cirrhosis.


This post originally appeared on Better Health, a network of popular health bloggers brought together by Val Jones, MD. Better Health's mission is to support and promote health care professional bloggers, provide insightful and trustworthy health commentary, and help to inform health policy makers about the provider point of view on health care reform, science, research and patient care.