The case of the Florida AIDS (acquired immunodeficiency syndrome) dentist continues to fascinate many because it was unprecedented when first reported in 1990 by the Centers for Disease Control and Prevention (CDC), because it remains unique today, and because no one has been able to put forth a likely mechanism for how the dentist infected his patients. It would have been irresponsible for me, as a journalist, to state that the dentist is blameless simply because the case is odd or because the CDC cannot pinpoint a route of infection. That is not what I have done. It also would have been irresponsible for me to propose a theory of infection based on the unsubstantiated assertions of a few individual persons, as have some news organizations, or based on certain hypothetical scenarios rather than on the facts of the case, as have some researchers. That, also, is not what I have done. As an independent investigative reporter, I was able to gain access to thousands of pages of documents related to the lawsuits brought by the infected patients against the dentist's insurers. These documents—medical records, legal depositions, scientific analyses, and more—contained information pertinent to the CDC's epidemiologic field work and molecular analyses, information that had never before been made public. This was the foundation for my articles in Lear's Magazine [3] and The New York Times [4] and for the 60 Minutes broadcast on 19 June 1994, for which I served as Associate Producer.

In those reports, I suggested a more plausible and more mundane explanation of this strange case: that the six patients were infected through well-documented routes of HIV transmission and not by the dentist. I found documentary evidence to show that some patients had known risk factors for HIV and that others were not completely truthful about their medical, sexual, and drug use histories. I also found evidence of potentially serious flaws in the phylogenetic analyses used by the CDC to conclude that the dentist and his patients had the same strain of HIV. Like the CDC, I cannot state with absolute certainty how the patients acquired HIV, but epidemiology is the science of "best fit" from an incomplete set of data. I believe that this new information expands the available set of data and raises the strong possibility of a different "best fit." More importantly, I believe that this new information deserves careful examination by eminent scientists and clinicians.

Ciesielski and colleagues [1] contend that my print and television work "omitted pertinent epidemiologic and laboratory evidence" because these facts would "contradict [my] conclusions." They cite only two apparent lapses, however, and neither lapse would suggest that my mission was obfuscation. Let us consider each one and its implications.

One lapse concerns the fact that 60 Minutes did not mention that the two boyfriends named by Patient A had tested negative for HIV. This detail was not only included in the article in Lear's Magazine [3] but is less relevant given the evidence I presented about Patient A's sexual past. This patient told federal and state health investigators the same thing she told the world: that she was a virgin. I brought to light two critical pieces of information that suggest that her sexual history was not consistent with that claim.

Internal CDC field notes written by Ciesielski herself indicate that she was skeptical that Patient A was being completely truthful; the notes even point toward a motive for why the patient might have lied: She may have feared "serious negative impact if her mother believed she had participated in risky behaviors" (Ciesielski CA. Unpublished communication). This does not prove that the patient was lying to investigators, but the other piece of new information that I presented seems to support Ciesielski's doubts. As part of her legal action against the dentist's insurers, Patient A was ordered by the court to have a gynecologic examination. The physical examination found that her hymen was consistent with having engaged in sexual intercourse. In addition, the gynecologist observed genital lesions, and specimens from Patient A's vagina and anus tested positive for human papillomavirus type 18, a sexually transmitted pathogen. In related research [5], CDC investigators have deemed that a history of sexually transmitted diseases is a potential risk factor for HIV.

Because of the limitations on its investigative powers, the CDC would have had no way to detect this information unless it had been contained in the patient's medical records. But in a deposition taken in November 1992 as part of lawsuits brought by two patients other than Patient A, Jaffe confirmed that he had been given a copy of the gynecologic report in late 1990, several months after the news about Patient A had become a worldwide sensation. At that time, Jaffe spoke privately with a colleague, who told him that human papillomavirus had been found in children in Africa who were infected with HIV, and Jaffe testified that, on that basis, he had determined that the gynecologic examination had no value to the CDC's still-active investigation. In turn, he gave the report to the CDC's general counsel, consequently shielding it from access through a Freedom of Information request.

Beyond questioning the scientific basis for Jaffe's actions, scientists should be concerned about his apparent lack of interest in information that might undercut the validity of the public position that he had already staked. Rather than do more follow-up to verify the gynecologic findings, he thought it was "entirely inappropriate," as he stated in his deposition testimony, that the examination results were sent to him in the first place. Ciesielski and colleagues still seem to be uninterested. Instead of using a scientific forum to rebut the information I presented, they did not even mention the gynecologic findings in their article [1].

Similarly, I reported on information that casts doubt on the claims of the five other patients that the dentist was their only possible source of HIV infection. Ciesielski and colleagues seem to suggest that the foundations of their epidemiologic field work are firm as long as no one can document some other specific, unassailable source of infection. This is an impossible standard, especially in a case in which all the patients after patient A were already aware of the CDC's dental transmission theory when they were first questioned about their risk factors and sexual partners, and therefore had a monetary stake in having the CDC find that the dentist infected them. For me to state that the patients might have lied would be unremarkable and could be construed as nitpicking. However, the evidence I presented goes beyond mere innuendo.

Take Patient I, who was the last patient to accuse the dentist. Investigators from the CDC had access to the insurance billing records of this patient's family, but apparently did not scrutinize them closely enough. These records directly contradict the patient's claims about the number of dental visits she made to the office and the kinds of dental treatment she received. In fact, they suggest that she may never have been treated by the dentist himself, which would mean that her risk for exposure was nonexistent. Ciesielski and colleagues' only comment on this information is quaint—that such records "do not document every patient visit" [1]. Not only does this unfounded statement seem out of place in a scientific forum, but it is the information in the records—not what is missing from them—that discredits the patient's version of the truth.

Patient I stated that she went to the dentist's office in the summers of 1987, 1988, and 1989 and had cavities filled. However, the extant insurance records (her dental charts were never found) show only a single visit for an initial oral examination and teeth cleaning by a hygienist in August 1988. That entry, in turn, carries an American Dental Association code indicating that the visit was the patient's first to the dental office. If the patient had visited the office in 1987, as she had claimed, the code would have indicated that the 1988 visit was a return visit. Indirectly, the records also counter Ciesielski and colleagues' suggestion that other patient visits were not logged by the insurance company; they show dozens of visits made to the dentist's office by the patient's parents and siblings during the same period, which makes it hard to imagine that visits by the patient herself would not have been submitted for reimbursement. Further more, during preparation of the 60 Minutes segment, I uncovered and authenticated a letter dated 23 May 1989, written by the patient's stepmother to the family's dental insurance provider. In it, the stepmother informed the insurer that the family had already started seeing a different dentist, who was part of their health plan. The letter, which Patient I did not make available to investigators, indicates that she also could not have visited the dentist in the summer of 1989, as she had asserted. (Because of time constraints, this information was not included in the 60 Minutes broadcast.)

As for the four other patients, I found it difficult to determine where truth and openness began and ended. Patient G, for instance, said that he had used intravenous drugs just once in 1973, and he reported having had only two female sexual partners since 1986, both of whom tested negative for HIV. However, I reported on sworn deposition testimony, given by an acquaintance of Patient G, which stated that the patient had frequented a local crack house three or four times a week in the mid-1980s, that he had traded crack for sex, and that he had had unprotected intercourse as many as 50 times with a prostitute who later died of AIDS. Ciesielski and colleagues ignored this evidence in their article [1].

In the case of Patient C, Ciesielski and colleagues [1] seem to bolster the validity of information I uncovered—that the patient had had anal intercourse on at least six occasions with another man. Although the patient denied that he had ever had homosexual contact, Ciesielski and colleagues reported for the first time that they had identified a male sexual partner who had tested negative for HIV [1]. They also disclosed that the patient had named 14 female sexual partners that he had had since 1978 and that only 9 of these partners had been located and tested. In related research, the authors categorized patients with multiple sexual partners as high risk [5], even if only some of the reported sexual partners could be located and those that were located tested negative for HIV. In this instance, however, the authors applied a different standard. They found not only that the patient had had multiple sexual partners, but also that his claim of never having had homosexual sex was untruthful. Nevertheless, they overlooked this inconsistency in their epidemiologic analysis and remain convinced that only the dentist could be blamed for the patient's infection.

The evidence surrounding Patient E is most perplexing. This patient learned that she was positive for HIV in 1988. She confirmed to me in an interview that, at the time of diagnosis, she believed that her boyfriend, who had known risk factors for HIV, had infected her. But in 1991, 6 months after the dental case became public, she came forward and told investigators that the boyfriend had actually tested negative for HIV when she had first tested positive, and that he had tested positive only later. However, it is unclear on what basis the authors have accepted Patient E's claims, because her 1992 deposition testimony does not concur with the information presented by the CDC as to when patient E and her boyfriend learned of their infections.

As for Patient B, when she was first tested for HIV in September 1990, she checked off "blood recipient 1975-85" as a potential risk factor on two different forms. Although she had had major surgery several times during this period, her hospital charts show no record of any transfusion, and Jaffe is shown asserting that on 60 Minutes. Although a patient's medical chart is a more telling record than a patient's memory, the discrepancy is curious. Even more curious is the evidence related to an extra-marital affair that Patient B had in the late 1970s. In a deposition given in May 1991, this patient named the man with whom she had had a sexual liaison, but there is reason to believe that his identity had not been disclosed to the CDC 7 months earlier, when the patient came forward. In an interview in 1993, the man named by the patient told me that he had never been tested for HIV or even contacted by the CDC, and he also denied ever having had sexual relations with the patient. If the man did have an affair with the patient, why was he never tested by CDC investigators, unless his name only first surfaced during the deposition? And if he did not have an affair, why would the patient lie under oath and name him as a sexual partner? Either way, these questions suggest that the patient was not completely truthful when investigators first asked her about her sexual history in 1990.

The reason that Ciesielski and colleagues have opted to believe the patients at every turn, even when confronted with evidence that undercuts the patients' credibility, is their absolute faith in the DNA sequencing work that they used to show that the dentist and his patients had similar strains of HIV. And in once again professing their faith in their laboratory data, they point to my other apparent lapse—the failure to mention, as part of 60 Minutes, the existence of a letter by Drs. David Hillis and John Huelsenbeck of the University of Texas [6]. This letter was published in Nature in May 1994, after the publication of the article in Lear's Magazine. Ciesielski and colleagues [1] present this work as the final arbiter of their original molecular analysis, because Hillis and Huelsenbeck "independently reanalyzed" all available sequence data from the CDC and from scientific critics of the CDC.

During preparation of the "60 Minutes" segment, I asked several of these critics for their comments on the work by Hillis and Huelsenbeck, and they pointed out two reasons why the new research should not be considered definitive.

As I have reported, these critics have claimed to have found other persons living in the community in which the dentist practiced who had no connection to the dentist and six patients but who were infected with similar strains of HIV. They had presented their preliminary findings in a letter that was published in Nature in 1993 [7], but a journal article presenting the critics' challenges in detail has never been published. This means that there are additional sequence data that are not yet available to the scientific community for further analysis. In addition, these critics questioned the statistical method used by Hillis and Huelsenbeck, arguing that it was a modification of another analytic approach and had never before been used to look at a phylogenetic problem. The new work may have some merit in helping scientists to better understand a new and rapidly evolving field of inquiry, but there was still no consensus that it was the optimal way to determine whether the dentist had truly infected his patients. Indeed, a more recent article that also supports the CDC's conclusions has described yet another method of analysis [8]. A debate about methods of phylogenetic analysis may be interesting for scientists to ponder, but it would not have enhanced a general television audience's understanding of a situation in which researchers fundamentally disagree on complex issues of molecular biology.

Similarly, because of time constraints, the 60 Minutes segment did not delve into additional information that casts doubt on the certainty of the CDC's scientific work—something the Lear's Magazine article did at length [3]. There is available research that questions whether the CDC used the best region of the HIV genome for its sequencing work, whether viral strains of HIV "converge" and become more alike over time, and whether it is appropriate to use phylogenetic analyses to conclude that one person transmitted HIV to another [9]. Pertinent questions have also been raised based on research that points to the lack of similarity in HIV strains of infected mothers and their infants [10]. In addition, a statistical analysis indicates that the prevalence of HIV-infected patients in the dental practice is analogous to the community at large. Finally, there is information from the lawsuits against the dentist's insurers that, on close scrutiny, would give the scientific community a chance to look broadly at the process by which CDC investigators reached their original conclusions and subsequently took steps to bolster them.

Some of this evidence, for instance, suggests that the CDC used faulty control groups at several turns in its investigation. The local controls used initially were drawn from a national database that included sequencing data contributed by scientists who said that these data should never be used for control purposes. At later stages of its investigation, the CDC used blood samples drawn from public HIV clinics in Florida cities with large, transient homosexual populations and not from the more stable, more heterosexual community where the dentist and patients lived. Evidence also suggests that the CDC retrofitted its analyses when confronted with answers that did not match its desired conclusions. While investigating the claims of the six patients, CDC researchers twice changed the eight amino acids they used in their signature pattern analysis and three times changed their method of determining the signature pattern. And the possibility of contamination in the CDC laboratory also cannot be ruled out, given that the blood samples from the dentist and Patient A were received on the same day and amplified on the same day. Interestingly, no independent laboratory has ever been given a sample of the dentist's blood in order to verify the CDC's original sequencing data. In addition, one of the CDC's own collaborators has reported on data that indicate the possibility of contamination among the controls used by the CDC in its dental study [11]. I would also be interested in an explanation of comments made by one of Ciesielski's co-authors in 1990, before any information about the case had become public. Specifically, in a letter to the CDC, Witte called the analysis linking the dentist's and the patient's viral strains "scientifically inconclusive" (Witte J. Unpublished communication). My calls to his office have never been returned.

In general, Ciesielski and colleagues have sought to dismiss the epidemiologic and molecular questions that I raised by arguing that I presented evidence "related to private litigation generated by the case." This is a red herring. Six of the eight scientists who have participated in research critical of the CDC's conclusions have received no money from the insurance companies and have had no financial stake in the outcome of their research. Several even claim that their affiliation with this research has affected their chances of getting grants. The authors' charge is also a red herring because the authors have no idea whether the evidence is credible. Take the finding that Patient A may have had a sexually transmitted disease: Investigators from the CDC did nothing to try to verify whether the gynecologist's diagnosis was valid.

As for the fact that the insurance companies settled all the legal claims in this case, that is another red herring. Defense attorneys have repeatedly stated that the settlements were based on monetary considerations—in particular, the high cost associated with refuting the CDC's sequencing analysis—and concern about finding open-minded jurors in a state saturated with publicity about the "AIDS dentist."

Finally, if the authors are going to dismiss the value of the evidence uncovered by scientists associated with the defense in these lawsuits, the same charge should apply to many who did work for the CDC. Los Alamos National Laboratory received funds from the CDC for analyzing the genetic data that concluded that the dentist and patients had the same strain of HIV, and the laboratory's chief investigator, Dr. Gerald Myers, testified on behalf of the plaintiffs as to the validity of his analytic findings. Dr. James Mullins of Stanford University, who was hired by the CDC to review its laboratory techniques, was later paid $1000 a day as an expert witness for several patients in their lawsuits against the dentist's insurance companies. In addition, Jaffe himself willingly agreed to testify on behalf of two of the patients. It is exceedingly rare for government scientists to take sides in private litigation, and Jaffe's participation gave the plaintiffs' attorneys a substantial legal advantage, because they did not have to spend millions of dollars to replicate the CDC's scientific work. However, it is unlikely that Jaffe had any interest in getting paid for his testimony; his involvement probably reflected concern that if the legal claims had not been settled out of court, the CDC's reputation as the world's premier public health organization would have been put on trial.

Given that the authors saw fit to include personal invective in a scientific article, I also feel compelled to address the criminal allegations against Dr. Lionel Resnick, one of the most outspoken critics of the CDC's work, that came to light in February 1995. Charged with fraud and embezzlement, Resnick has since resigned his post as chief of retrovirology and research at Mount Sinai Medical Center in Miami. No evidence of scientific impropriety has been found, but the charges make it easy to discredit the work done by Resnick and other CDC critics (none of the others have been connected with Resnick's alleged criminal activities). The news has also dampened the interest among the others in trying to get a fair hearing in a scientific forum. A major paper that contained detailed molecular analyses critical of the CDC has been shelved, even though Resnick, who was a co-author, agreed to take his name off of it. I find the whole situation unfortunate, not only because potentially important scientific information may never be made available for review, but also because the material I have presented goes far beyond Resnick's involvement in this case. Criticism of the CDC's epidemiologic and phylogenetic work does not begin and end with Resnick.

Ciesielski and colleagues seem content to describe this case as a medical mystery that is beyond resolution. They suggest that it should be left at that. From a public health perspective, that may be an acceptable position. Given that there have been no other cases of a health worker infecting patients, the CDC's contention that the risk for transmission in a health care setting is infinitesimal will probably stand up over time. But there are so many loose ends, bizarre details, unusual occurrences, and open issues that the case itself should not be simply forgotten. The very fact that there have been no other cases like it is a compelling enough reason to probe further and try to determine what exactly did happen in this single instance.

The authors' final statement [1]—"[O]ur inability to determine how these transmissions occurred does not mean that they did not occur"—is an appeal for trust that the CDC would not and could not have made a mistake. Although I would not expect scientists and clinicians to rely solely on media reports for drawing conclusions about the complex and controversial questions involved in this case, I also would not expect them to be content with the authors' assertions that their work is infallible or that it is unnecessary to continue to seek answers to this strange and troubling case. Rather than shutting off the scientific process, I encourage a full and open discussion of the evidence. That is something that has never happened.