Author + information
In an effort to evaluate the relationship between the use of fenfluramines as diet drugs and the prevalence of mitral valve and aortic valve regurgitation, Burger et al. (1) compared measurements from a study conducted for another purpose to those described by Singh (2). Burger observed that the prevalence of mitral valvulopathy in his study was comparable with the Framingham study (1.3% vs. 1.6% from Framingham) and aortic regurgitation (6.6% vs. 4.8% from the Framingham study). Burger surmised that the valvulvar regurgitation seen in his patients may not be due to fenfluramine but to age-related degenerative changes. Schiller (3), in an accompanying editorial, seconds this point of view stating (page 1161), “It would seem then that as studies have become more scientifically rigorous, the role of fen/phen in valve disease appears to be approaching the vanishing point.” However, there are two important additional observations concerning Burger’s methodology that undermine these conclusions.
There were 591 patients in Burger’s study. Of these patients, only 226 (38.2%) returned for an echocardiogram. The remaining 365 patients who were also exposed to the fenfluramines, for unknown reasons, did not undergo echocardiography. Since only three of the 226 patients who returned had mitral regurgitation, and only 15 of the 226 patients had aortic regurgitation, the fate of the remaining 365 patients is critical in a proper assessment. The absence of the echocardiograms in over 60% of the cohort makes this study especially vulnerable to selection bias.
A second concern involves Burger’s simple comparison between regurgitation prevalence in his population and Framingham. The difference in the mean ages between that of Burger’s cohort (mean age 46.9, standard deviation 8.9) and that of the Framingham population (mean age 55, standard deviation 10) suggests that a coarse comparison of the crude prevalences from these two populations is inappropriate and misleading. Fortunately a more appropriate adjustment is available through an examination of Singh’s data (4). Given both the mean age (standard deviation) and the gender distribution provided by Burger one can, assuming the normal distribution, approximate the distribution of age and gender in the Burger study. From Singh (4) the prevalence of each of mitral regurgitation and aortic regurgitation is available (Table 1).
From Table 1 one can compute the expected prevalence in the Burger cohort based on the gender and age-specific mitral and aortic valve prevalence in the Framingham study. If the 10.2% of patients whose ages are outside the 26 to 83 age range (based on a normal distribution with mean age 46.9 and standard deviation 8.9) fall in the upper age range (greater than 83) and these patients have the same prevalence of valvular regurgitation as those in the 70 to 83 age range, the computations reveal that the expected prevalence for mitral regurgitation (Food and Drug Administration criteria) is 1.0% and for aortic regurgitation is 3.3%. This conservative computation provides mitral and aortic prevalences that are less than those observed by Burger. Based on these age- and gender-adjusted prevalences, the prevalences seen by Burger et al. are greater than would be expected from degenerative changes alone.
These two observations substantially weaken the explanation provided by Burger. Since Schiller chose to build his conclusions on Burger’s results, this editorial’s foundation is now effectively removed. The data collected by Burger, while representing an incomplete assessment, support, rather than refute, the association found between fenfluramine and cardiac valvulopathy.
↵1 Both Dr. Moyé and Dr. Annegers have been retained by counsel as experts for the plaintiffs in ongoing fenfluramine litigation.
- American College of Cardiology
- Burger A.J.,
- Sherman H.B.,
- Charlamb M.J.,
- et al.
- Schiller N.B.
- Singh J.P.,
- Evans J.C.,
- Levy D.,
- et al.