J Thorac Cardiovasc Surg 2008;136:799
© 2008 The American Association for Thoracic Surgery
Reply to the Editor:
Thoralf M. Sundt, MD
Division of Cardiovascular Surgery, Mayo Clinic, Rochester, Minn
I appreciate Dr Royston's kind comment regarding my editorial. I am quite certain that he agrees with me that we are "worse off without [aprotinin] in our arsenal." Indeed, I feel this sentiment particularly this evening, as I wait for a call from the operating room to start repair of an acute dissection in an 80-year-old patient who is receiving warfarin 5 years after coronary bypass and aortic valve replacement. I am also sure that he agrees that, in the best of all possible worlds, the risks and benefits determining the use of a drug should pertain to the welfare of the patient as judged by physicians and not to the litigation risks of a pharmaceutical company as judged by lawyers.
I also appreciate Dr Royston's comments regarding bias. The biases of which he speaks have not (entirely) escaped me; all studies have biases. Randomized studies are of necessity biased at entry. Rigid eligibility criteria are necessary to define a population with sufficient precision to permit analysis, and the demands of equipoise encourage inclusion of low-risk patients for whom harm is the least likely–but so is benefit. Consequently, few such studies truly reflect the spectrum of disease that we face in clinical practice. The populations included in observational studies are more representative of practice; however, the bias introduced by the clinical judgments made in the application of a therapy or administration of a drug impose considerable challenges to balanced interpretation, as so beautifully demonstrated in Dr Royston's letter. I could not agree more. As noted in his comments, understanding the appropriate application and interpretation of propensity analysis demands a learned understanding of the methods as well as the aims of the matching. Unfortunately, few of us (certainly not I) are so statistically sophisticated. I know that in this regard Dr Royston can run circles around me. No contest.
In the end, where we differ, it would appear, is regarding just where the rest of the medical community is struggling. What are the real risks associated with aprotinin? Is it "a potentially harmful drug"? If so, what is the magnitude of that risk? Personally, I remain amazed that today (May 23, 2008), with more than 7100 citations now retrievable on a PubMed search for aprotinin, there is still room for debate.
