We agree that anecdotal reports and case series are insufficient to establish the existence of a new disease entity or determine its cause. The purpose of our article was not to discuss such reports, which have been summarized elsewhere and do not meet the criteria for epidemiologic studies. Still, as Dr. Whysner points out, such reports may be the first clues to a previously unrecognized disease entity. Although many "new" syndromes are in fact variants of known disorders, new diseases such as the eosinophilia-myalgia syndrome and Lyme disease are occasionally reported [1]. Crucial to the study of such "new" entities is the establishment of a case definition based on case reports. For instance, although the eosinophilia-myalgia syndrome had some "scleroderma-like" features [2, 3], it was recognized that the syndrome differed from true scleroderma. One wonders whether the remarkably rapid progress in associating L-tryptophan with the syndrome would have been possible if researchers had dismissed early case reports as anecdotal or had confined their studies to persons with defined connective tissue disorders rather than formulating a case definition for a new syndrome.

Several studies that claim to have ruled out an association of silicone breast implants with connective tissue diseases have relied on data collected for other purposes, such as surveys or administrative records. As such, the outcomes that could be studied were limited to known physician-diagnosed syndromes. Despite claims to the contrary, these studies left little room for the detection of patients, with or without implants, who had syndromes that did not meet established diagnostic criteria. Although it is incorrect to conclude that all adverse health events can be attributed to the receipt of a breast implant, it is equally incorrect to assume, for reasons of convenience, that adverse health events that do occur will meet criteria for known disorders. Therefore, we do not agree that these studies "properly addressed the individual case reports and clinical series that initially raised some alarm" about silicone breast implants.

Dr. Whysner is correct that, despite the potential for bias and confounding, epidemiology is the "best we can do" to determine whether the problems reported by so many women are truly associated with breast implants. We disagree, however, with the conclusion that the epidemiologic studies to date represent the best that can be done. These studies have done a fair job of ruling out a markedly increased risk for certain defined disorders. However, because of various methodologic problems described in our paper, these studies failed to address the possibility that anecdotal reports of silicone-associated disease point to a previously unrecognized disease entity.