Department of Epidemiology University of North Carolina School of Public Health Chapel Hill, NC email@example.com
To the Editor:
Just before taking my first sip from a cup of coffee, I read the case-crossover study by Baylin et al1 and its estimated relative risk of 1.5 for myocardial infarction in the hour after doing what I was about to do. I wondered what this result might mean if it were translated into more meaningful terms. Recalling the advice of Cornfield et al2 to use the risk difference for such a purpose, I realized I needed an estimate of the baseline risk: the risk my cup of coffee would hypothetically be increasing by an estimated 50%.
Schwartz et al found that baseline risks are hard to come by in journal articles in which estimates of ratio effect measures are reported.3 Sure enough, the paper by Baylin et al1 gave no hint as to what the 1-hour risk of myocardial infarction might be. I went to a previous paper the authors cited from the same study,4 then to the paper cited by that paper.5 No luck. It was a population-based study and myocardial infarction incidence rates had been calculated, but none had been reported.
Clearly, I had to take another tack. Browsing through the web site of the National Heart, Lung and Blood Institute (www.nhlbi.nih.gov), I stumbled across a document6 with an appendix giving a brief description of Framingham risk scoring. Although a “Framingham risk” is an estimated risk of myocardial infarction or coronary death and Baylin et al1 had studied only the former, I was unfazed. I needed only a ballpark value. After all, the decision in question was just a single cup of coffee for a target population of size N = 1.
Some illustrative tables in the appendix6 suggested that a 10-year risk of 10% would not be out of the question. That was the risk given for a man with a Framingham score of 9 or a woman with a score between 19 and 20.
With the help of my trusty calculator, I was able to take it from there. An average decade has 10(365.25)(24) = 87,660 hours. Hence, a 10-year risk of 10% would be a 1-hour risk of 0.1/87,660, or about 1 in a million. A 50% increase in such a risk would be an increase of about 1 in 2 million.
Finally, I had an answer to my question. What would it mean for a cup of coffee to increase the 1-hour risk of myocardial infarction by 50%? It would mean an expectation, on average, of about 1 extra heart attack in the hour immediately following the consumption of each 2,000,000 cups of coffee.
Schwartz et al3 suggested that researchers should routinely provide this kind of information to readers: the absolute differences that are implied by the ratios of risks, rates, and prevalences we typically estimate. I agree. Journals might think about encouraging or even insisting that this information be provided, even if only in very approximate terms like those in my calculations.
In the 2 hours it had taken me to work out that risk difference on my own, my coffee had gone cold. While weighing the risk-benefit tradeoff of getting a fresh cup, I was reminded of something a colleague once told me. He, too, had conducted a case-crossover study of myocardial infarction and coffee, but the results were never published. They were “negative.”
That made my decision even easier.
Department of Epidemiology, University of North Carolina School of Public Health, Chapel Hill, NC firstname.lastname@example.org
1. Baylin A, Hernandez-Diaz S, Kabagambe EK, et al. Transient exposure to coffee as a trigger of a first nonfatal myocardial infarction. Epidemiology
2. Cornfield J, Haenszel WH, Hammond EC, et al. Smoking and lung cancer: recent evidence and a discussion of some questions. J Natl Cancer Inst
3. Schwartz LM, Woloshin S, Dvorin EL, et al. Ratio measures in leading medical journals: structured review of accessibility of underlying absolute risks. BMJ
4. Baylin A, Kabagambe ED, Ascherio A, et al. Adipose tissue α-linolenic acid and nonfatal acute myocardial infarction in Costa Rica. Circulation
5. Campos H, Siles X. Siesta and the risk of coronary heart disease: results from a population-based, case-control study in Costa Rica. Int J Epidemiol
6. National Cholesterol Education Program, National Heart, Lung, and Blood Institute. Third report of the National Cholesterol Education Program (NCEP) Expert Panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III). Executive Summary. NIH Publication No. 01-3670. Bethesda, MD: NIH; 2001:25–27.