Thursday, February 26, 2009

Anger Management

By now, there must be better than twenty-five "articles" in the so-called "press" about the effects that the emotion anger might have in predicting "heart risks," especially in patients with defibrillators. Each of these journalistic articles point to a recently published manuscript in the Journal of the American College of Cardiology as their authoritative reference, and report the "facts" without any journalistic effort to determine the study's credibility or scientific rigor. Certainly, I do not blame the Journal of the American College of Cardiology for publishing the work - the topic is one that warrants evaluation and critical review. (Of course, what the reviewers of the manuscript were thinking is another issue, but more on that later.)

First, for the lay reader of these press stories, it's important to realize that whenever there are so many stories published simultaneously about a single manuscript like this, the press are almost almost always tipped off about the results by a press release with a really interesting photograph. Certainly if it's about being angry, then the press release shows someone really angry. (It's hard to get your point across otherwise, I guess).

Second, one should always look for a medical device company or academic center that wants to promote the technology or researcher (or both). Interestingly, these articles fly around the internet faster than most because these larger subliminal sponsors have PR teams with friends in high places in the press corps. One doesn't have to look too far to note that the second author of the study, Vladimir Shusterman, MD, PhD, is the President of PinMed, Inc., the maker of the T-wave alternans technology used in this study. But that should not sway us, because we also note that Dr. Shusterman is also "a Faculty Member and Director of Noninvasive Cardiac Electrophysiology Laboratories at the Cardiovascular Institute, University of Pittsburgh."

Interesting that a researcher on such a study can wear two hats like this without batting a single conflict-of-interest eye.

But that's not enough. A few more clicks to follow the research (money?) trail, and we find that the lead author of the study and her home institution (Yale) have been working with PinMed, Inc. for quite a while, so long in fact that the lead author has been quoted saying some nice things about the company on their website:
"As a practicing electrophysiologist, I can envision multiple clinical uses for the PELEX system once it is approved by the FDA... I think this technology will be extremely useful, both for my own research applications, and also likely for clinical practice."
Not that this is bad, mind you. Certainly all of us are entitled to our opinions and biases. But it never ceases to amaze me how gullible the press can be when it comes to scientific PR spin and how this little point was not mentioned in the study's limitations, but rather, Dr. Lampert's conflicts mention only a "minimal" conflict with St. Jude Medical.

Finally, there are the issues with the study itself. I find it remarkable that few have questioned a basic premise of the study, that ICD shocks occur more often when people who have a propensity for a tiny increase in T-wave alternans when they get angry, are more likely to have an implantable defibrillator fire. But on the surface, without any fancy schmancy technology, it should be obvious to the casual observer that any defibrillator is more likely to go off when the heart rate is faster than someone whose heart rate is slower: that's how these gizmos work. Certainly, people's hearts might beat faster when they are angry. But any other cause of fast heart rhythm might make them fire as well. In fact, all that has to happen in the most basic defibrillator detection algorithm is the heart rate exceed a programmed rate for a sufficient amount of time and, BAMMO!, the patient gets shocked or might receive anti-tachycardia pacing. Might other things like atrial fibrillation, exercise, sex, anxiety poosibly cause similar therapies? You betcha.

But when one wants to report results for a new-fangled T-wave alternans test that we want the world to use (and maybe on angry people) to predict of their ICD will fire, well we get 62 of them, look for T wave alternans, then follow them for a year to see what their ICD does. Sure enough, 10 of them have some therapies. Yep, t-e-n. Just enough to place these patients into "quartiles" (2 -3 three a quartile???). Then we compare the top and bottom quartile and, yep, there's a statistical difference! Quick, get out the press release!

But alas, if the press had read the paper, they'd find these limitations admitted by the authors:
"Also, the sample size and number of events was small, and these data should be viewed as hypothesis generating. Whether patients who agreed to participate may differ from the overall population, creating a selection bias, cannot be determined. Larger studies may more definitively establish the predictive value of anger-induced TWA. Further, the programming of ICDs was not uniform. Although the rate cutoffs did not differ between high- and low-anger TWA groups, the possibility that differences in detection times or programmed treatment may have contributed to the findings cannot be excluded."
But these limitations don't make headlines, do they? Nope, better to gloss over these fine points and go to press anyway. After all, it's all improving that impact factor, right?

Sheesh, something like this might make someone angry...

.... uh, oh.


1 comment:

Anonymous said...

Thanks for doing the background work. I read the headline, thought there must be some underlying BS. But did not make an effort to find it.
BTW, never heard of NON-INVASIVE EP before!