A recent article in Neurology suggests that the combination of coronary artery bypass surgery with carotid endarterectomy may not be a good idea.
Dr. Wes takes this article on in his blog and provides a very insightful analysis of the issue and delves into the problems that are involved with this type of retrospective analysis of administrative data. This is the type study that I like to call a "coarse grain" study, a satellite view of the forest providing no essential data about the individual trees.
Dr. Wes says it well when he speaks of the study as:
"... a retrospective chart review of computer-code data set....in an attempt to glean a flicker of data with which to draw a glimmer of a trend-NOT a conclusion."
In regard to the authors's disclaimer regarding "the limitations to the use of administrative data sets" his translation nails it:
"we know the data are poor and we fixed them a bit to clean them up and erased some data, but don't worry about that."
When the odds ratio is corrected for the various things that epidemiologists like to think they can correct for it shrinks from a value of 2.25 to an unimpressive 1.38. The uncorrected odds ratio is the one that will get the emphasis in the lay press and will serve to worry and frustrate patients and their physicians particularly since this study seems to not support the current position of the American College of Cardiology and American Heart Association. One retrospective, administrative coding analysis should not have the evidentiary weight to effectively contradict the ACCP/AHA position. Calls for a randomized clinical trial to "settle" this issue appeared quickly.