The recent review [1] by Silverman and colleagues (employees of the Food and Drug Administration [FDA]) concluded, as has every review and epidemiologic study published in authoritative peer-reviewed journals, that silicone breast implants do not lead to any marked increases in well-defined connective tissue diseases. These epidemiologic studies have properly addressed the individual case reports and clinical series that initially raised some alarm about the safety of breast augmentation and reconstruction with silicone implants. The case reports drew attention to a possible problem that was not confirmed when submitted to more rigorous scrutiny [2-4].

Why, then, did the FDA review article draw attention only to the limitations of epidemiology without similarly faulting the anecdotal reports and causal speculations? Case reports and clinical series are not sufficient evidence for causal associations; only cross-sectional cohort or case–control epidemiologic methods can validate relations between exposure and effect. Epidemiology, which deals with human beings and is subject to potential inclusion bias, recollection bias, and other confounders, can reassure but cannot totally eliminate minor associations. However, it is the best we can do.

Why even consider "atypical connective tissue diseases" when these fail to define any one disease or group of verifiable symptoms and signs? In these "atypical" syndromes, the presence of a silicone implant was isolated and all subsequent diseases and symptoms were attributed to it. This is improper. Silverman and colleagues [1] themselves agree that the term "human adjuvant disease" has been discredited because of the lack of precise and reproducible criteria. Thus, the possible design flaws in epidemiologic studies cited in the FDA review are minor in comparison to the major flaw of lack of definition in anecdotal reports. Without such a definition, the "atypical" syndrome cannot be measured and studies cannot be proved or disproved.