Friday, May 21, 2010


In episode 233 of Phedippidations, Steve Runner focused on an abstract from the 2010 ACC Scientific Sessions titled Marathon Runner Have Increased Aortic Stiffness by Dr. Kardara from Greece. While I do not personally do research on pulse wave velocity of the aorta, I do conduct science and actually presented my own research at the same meeting. I wanted to take the opportunity to share my analysis of the study for anyone who is interested.

  • Overall, I thought Steve made an excellent summary in describing the physiology of systole, diastole, and arteriosclerosis. Furthermore, he asked the critical questions about the SIGNIFICANCE of the findings.
  • The media coverage of this, like so much of the coverage of medical research is hyperbole. They have taken the most preliminary of findings and jumped far beyond the conclusions, extrapolating well beyond the real scope of this study.
  • I should provide some context about scientific meetings such as the ACC Scientific Sessions. In short, think of these meetings as “rough draft science”. The abstracts are purposefully short because THOUSANDS are presented at each meeting in the span of just a few days. Some are presented as 4’ by 6’ posters and some as 10-15 minutes oral presentations. Furthermore, the abstracts have been reviewed and graded by other scientists to pick which ones will be presented, but the abstracts have not yet undergone full peer review (a lengthy process where scientific papers are reviewed by unbiased third party experts to make sure that all the science is on the up-and-up). This could happen for many reasons. For example, the authors may not have the time and energy to write a full paper. However, it could also be that the study fails to meet the high standards of publication after peer review. We do not yet know whether or not Dr. Kardara’s research will ultimately be published. Also, sometimes research at these meeting is presented in rough draft form to get feedback from other scientists so the researchers can go back and retool the project. Sometimes it takes a fresh set of eyes to point out glaring errors in your methodology or interpretation of data.
    • For example: I presented an abstract at the same meeting. Based on feedback from this session and the peer review process, I have written a paper that will be published in a major journal and it looks very different from the findings that I presented at the meeting.
  • Ok, I’m gonna geek out for a moment, so feel free to skip this section if you want. But, let’s look at the findings and ask some questions.
    • Where were the controls were recruited from? Were they on the couch of doom? 
    • The abstract reports incomplete data on all the baseline variables (what risk factors were controlled?)
    • When were the controls measured? At the race? Were they spectators?
    • As Steve's twitter responses point out, just prior to a race the excitement of getting ready to run could certainly be a confounder that would not likely also have been present in the controls.
    • While the sample size is small, the statistics may still be appropriate
      • Statistical power (beta) is a measure of whether or not your sample size is big enough, typically researchers want their power to be >80% 
      • I calculated the statistical power for each of the outcomes described in the abstract
      • Aortic systolic pressure: power=98.4%
      • Brachial systolic pressure: power=97.7%
      • Heart rate: power=90%
      • Pulse wave velocity: power=80%
      • Mean age: power=10.4%
      • So, except for the one baseline variable they described, the study was powered (ie: large enough sample) to detect the differences they described
  • “Significant” is a scientific code word for “the numbers were different using the statistical methods we chose”. The word significant has absolutely no bearing on the relevance or importance of the findings. An important maxim to remember in reading all medical literature is: “Statistically SIGNIFICANT does not necessarily mean clinically RELEVANT.”
    • Example: If I develop a cancer drug that SIGNIFICANTLY shrinks your tumor, but you die of metastases at the same time as you would have without the drug, would that drug be RELEVANT?
  • In terms of clinical relevance, none is provided by this study. There is no measurement of clinical outcomes. They did not actually follow the runners and see if they had more heart attacks. The statistics are complicated, but to do so would take thousands of people followed for years.
  • Remember, the authors did not actually measure aortic stiffness directly. They used a surrogate marker, the pulse wave velocity. Furthermore, they assume that increased pulse wave velocity measured in marathon runners is the same process as it is in older people.
    • Example: What if the increased pulse wave velocity is due to another mechanism, such as hypertrophy (muscle growth) in the aorta? Athlete’s hearts grow to account for the increased work of constant exercise, so maybe the aorta can do the same. If so, then that is completely different from the “hardening” of arteries that occurs with age due to breakdown of elastic fibers. If the cause of the increased velocity is different then it stands to reason that the consequences of that increased velocity will be different.
  • Ultimately, remember this. The overwhelming body of evidence says that you live longer if you exercise. In fact, one of the best predictors of your personal mortality is your aerobic capacity, or your ability to maintain sustained exercise. Steve mentioned that ‘there is no science to say that becoming a runner will increase the duration of your life’. In my search, I could not find a randomized trial that took nonrunners and randomized them to starting running vs. sitting on the couch of doom. Such a study would be very hard to design and carry out, but instead I selected these studies. At least these studies actually measured clinical outcomes (ie: they counted and compared the number of heart attacks and deaths as opposed to measuring a velocity and extrapolating the results). Of note, I did not find any studies that measured clinical outcomes and found that exercise increased the risk of cardiovascular events.

Happy running!

No comments:

Post a Comment