the center. “At least 40 of the fi rst 248 gay men diagnosed with GRID in
the United States . . . either had had sex with Gaetan Dugas or had had
sex with someone who had,” he wrote, incorrectly, apparently not notic-
ing the additional degrees of separation between Dugas and patients on
the outer edges of the cluster. “The links sometimes were extended for
many generations of sexual contacts, giving frightening insight into how
rapidly the epidemic had spread before anyone knew about it.” Shilts
went on to describe these sexual connections in a fashion that was dis-
tinctly reminiscent of the example of Voltaire’s Pangloss presented in
chapter 1:
Before one of Gaetan’s Los Angeles boyfriends [“LA 3”] came down with
Pneumocystis, for example, he had had sex with another Angelino [“LA 2”]
who came down with Kaposi’s sarcoma and with a Florida man [“FL 1”]
who contracted both Kaposi’s and the pneumonia. The Los Angeles contact
[“LA 2”], in turn, cavorted with two other Los Angeles men [“LA 1” and
138. Later research would provide a much better understanding of the relative risk of
contracting HIV per sexual exposure and per type of exposure. Recent estimates place
insertive and receptive oral sex as very low risk, and per- act transmission rates for in-
sertive penile– vaginal intercourse at 4 per 10,000 exposures to an infected person, recep-
tive penile– vaginal intercourse at 8 per 10,000 exposures, insertive anal intercourse at
11 per 10,000 exposures, and receptive anal intercourse at 138 per 10,000 exposures. Fac-
tors such as concurrent genital ulcers and high viral load can increase the risks of trans-
mission, while using condoms and taking antiretroviral medications (as preventive or as
treatment) can reduce these risks. See Pragna Patel et al., “Estimating Per- Act HIV Trans-
mission Risk: A Systematic Review,” AIDS 28 (2014): 1509– 19.
134
chapter 2
“LA 4”] who later came down with Kaposi’s, one of whom [“LA 4”] infected
still another southern California man who was suffering from KS [“LA 5”].
The Floridian, meanwhile, had sex with a Texan who got Kaposi’s sarcoma
[“TX 1”], a second Florida man who got Pneumocystis [“FL 2”], and two
Georgia men, one of whom got Pneumocystis [“GA 2”] and another who
soon found the skin lesions of KS [“GA 1”]. Before fi nding these lesions, how-
ever, the Georgian [“GA 1”] had sex with a Pennsylvania man [“PA 1”] who
later came down with both Pneumocystis and KS.
He summarized: “From just one tryst with Gaetan, therefore, eleven
GRID cases could be connected.”139 Interpreting the cluster image with-
out access to information about each case’s dates of sexual contact and
symptom onset, Shilts drew a false assumption based on his perception
that “Patient 0” was the most important individual in the cluster and that
certain cluster patients had infected others. Judging by his later com-
ments, Shilts may also have interpreted a public health offi cial’s disclo-
sure—which linked a French Canadian fl ight attendant to the fi rst two
reported symptomatic AIDS patients in New York or to their lovers—as
a statement suggesting that the cluster’s patient numbers indicated the
earliest cases in each city.140 This too was erroneous— this numbering
was not absolute but rather relative to the cases linked to the cluster. Ear-
lier reported cases existed, particularly in New York, but Darrow and his
colleagues had not been able to link them to the men in the cluster, and
thus these were left out of the ordinal series employed in the study.
Some of these misconceptions would be clarifi ed when the cluster
study later reached a welcoming audience in the related fi eld of network
dynamics in the mid- 1980s, albeit after a slightly bumpy start.141 When
Alden Klovdahl, a sociologist with experience in network analysis and
computer- aided graphic design, published a letter criticizing some of the
cluster study’s limitations, Darrow recruited him to expand on his work.142
In an article published in 1985, Klovdahl used the study’s information to
demonstrate the strengths and weaknesses of a network approach to the
spread of infectious diseases; in the process, he revisualized the cluster
139. Shilts, Band, 147 [cluster labels added].
140. Sipchen, “AIDS Chronicles,” V9.
141. Douglas A. Luke and Jenine K. Harris, “Network Analysis in Public Health: His-
tory, Methods, and Applications,” Annual Review of Public Health 28, no. 1 (2007):77.
142. Darrow, e- mail to author, August 8, 2013.
The Cluster Study 135
in a way that destabilized the conventional centrality of “ Patient 0.”143
To manipulate the cases in the cluster diagram, Klovdahl employed
ORTEP— short for Oak Ridge Thermal Ellipsoid Plot— a computer pro-
gram which had been designed in the mid- 1960s as a means of repre-
senting crystalline structures in three dimensions. (Fittingly, the internal
newsletter that announced the program’s creation began with the phrase
“A picture is worth a thousand words.”144) Using ORTEP, Klovdahl ar-
rayed the cases by date of symptom onset, rather than the roughly geo-
graphic approximation depicted in the original diagram, to demonstrate
what he called “time sequence anomalies” (see fi g. 2.10). With the inclu-
sion of an axis for time in the model to account for symptom develop-
ment, cases LA 1, NY 1, NY 2, NY 3, and NY 4 near the bottom of the
diagram take on a new prominence. When one considers that the aver-
age number of partners reported by these patients was 227 per year (with
a range of 10 to 1,560), one begins to realize the very likely absence of
important links to earlier cases.
To explain the “time sequence anomalies,” Klovdahl noted that the
“possibility was that there were ‘missing nodes’ in this network, for ex-
ample, individuals who were exposed to an infectious agent by LA1, who
had not developed symptoms at the time of the original CDC study, and
who transmitted an infectious agent to ‘0’ directly, or to others who did.
It follows that good network data can signal the existence (and identity)
of individuals with subclinical infections.”145 Others would not be as dip-
lomatic. In addition to Andrew Moss’s 1988 letter, Duncan Campbell, a
British journalist, published strong critiques of Randy Shilts’s use of the
cluster study in response to the author’s British book tour for And the
Band Played On that year. In his work Campbell included excerpts from
a recent interview he had conducted with Darrow, in which the CDC so-
ciologist stressed that the 1982 hypotheses of incubation could no longer
be maintained. This statement was taken by some observers to suggest
that he dismissed his earlier work, and many would later clumsily inter-
pret the cluster study out of its original context. For example, on Octo-
143. Klovdahl, “Social Networks,” 1206.
144. Bill Felknor, “Laboratory Scientist Draws ‘Atoms- In- Depth’ Using Computer-
Oriented Graphic Technique,” The News [Oak Ridge National Laboratory
], April 2, 1965,
1, accessed November 3, 2014, http:// www .umass .edu/ molvis/ francoeur/ ortep/ ortepnews
.html.
145. Alden S. Klovdahl, “Social Networks and the Spread of Infectious Diseases: The
AIDS Example,” Social Science and Medicine 21, no. 11 (1985): 1208– 9.
136
chapter 2
Figure 2.10 A rotated reconfi guration of the “AIDS cluster,” using the Oak Ridge Ther-
mal Ellipsoid Plot Program (ORTEP), with date of symptom onset represented on the z-
axis; 11.6 × 11.0 cm. Reprinted from Alden S. Klovdahl, “Social Networks and the Spread
of Infectious Diseases: The AIDS Example,” Social Science and Medicine 21, no. 11
(1985): 1206, © 1985, with permission from Elsevier. With “0” off- center and cluster pa-
tients arrayed from bottom to top by date of symptom onset, other individuals are able to
assume more visually prominent roles in Klovdahl’s diagram. “LA1” and “NY1,” for ex-
ample, both reported having more than one hundred partners in the year before they ex-
perienced symptoms, yet their inability to recall many names for the investigators means
that they are surrounded by white space. This diagram suggests the new directions being
pursued by network analysis researchers and foreshadows the far more complex represen-
tations of sexual networks that would be made possible with the assistance of computers.
ber 31, 2007, in the light of recently announced scientifi c work which sug-
gested that HIV had fi rst arrived in the United States via Haiti in 1969,
two journalists for the Miami Herald wrote that the new evidence “de-
bunks the original ‘Patient Zero’ theory that said the HIV virus came to
Los Angeles via a gay Canadian fl ight attendant named Gaetan Dugas.
The Cluster Study 137
That theory was created by Dr. William Darrow and others at the CDC
and turned into the 1987 book And the Band Played On, by journal-
ist Randy Shilts. Darrow later repudiated his own study.”146 It would ap-
pear that the journalists had briefed themselves by consulting one of sev-
eral online articles that offered this misleading history.147 In response,
Darrow— who by this time had left the CDC and was a public health
professor at Florida International University leading community- based
AIDS prevention programs— wrote his own letter of frustration regard-
ing the cluster study in 2007 to the Miami Herald, complaining that
I have never repudiated the fi ndings of the L.A. Cluster Study. Nor, to my
knowledge, has anyone else.
Some people who have never taken the time to read our initial report in
Morbidity and Mortality Weekly Report (June 18, 1982), subsequent articles
published in the American Journal of Medicine (and other peer- reviewed jour-
nals), and chapters (that have been published in books about AIDS) have mis-
interpreted the purpose of the L.A. Cluster Study, methods, fi ndings, and con-
clusions. To be counted among those who have misrepresented the signifi cance
of our work and our conclusions are two reporters for the Miami Herald.148
Darrow’s letter was never published.
* * *
The cluster study that gave rise to the epidemiological phrase “Patient 0”
in the early years of the AIDS epidemic must be understood in the con-
text of a longer history of shifts in epidemiological practice and venereal
disease control. Framing the discussion in this way allows us to see how
the US federal government’s response to VD from the 1950s onward laid
146. Fred Tasker and Jacqueline Charles, “Disease Research: Scientists Trace AIDS
through Haiti; Findings Draw Anger,” Miami Herald, October 31, 2007, 1A. The report-
ers quoted a professor of medicine who, while expressing his doubts on the study’s fi nd-
ings, pointed to the topic’s perpetual intellectual appeal: “People love to play history, and
it would be great to fi gure out who Patient Zero was.”
147. Their history bears a strong resemblance to the one available on Wikipedia .com,
the free encyclopedia website, at that time: Wikipedia, s.v. “index case,” October 28, 2007,
http:// en .wikipedia .org/ w/ index .php ?title = Index _case & oldid = 167559312.
148. William Darrow, “Re: Disease Research— Scientists Trace AIDS through Haiti,
Stirring Ire,” unpublished letter to the editor, n.d. [2007], Darrow Papers.
138
chapter 2
the groundwork for the training and working culture for the VD Branch
at the CDC, and in turn for some of the early investigations into AIDS.
Public health investigators were taught to think of themselves as detec-
tives, to suspect homosexuality in cases of syphilis, and to trust the clus-
ter method of contact investigation to allow them to reach the source of
an outbreak. Such a working style laid the groundwork for potential con-
fl icts with their colleagues working on chronic disease, some of whom
appear to have viewed contact epidemiology as a quaint, outmoded ap-
proach from a simpler past. In communicating their research for the
cluster study, Darrow and his colleagues faced signifi cant challenges in
terms of representing complex human relationships. Linguistic and vi-
sual choices unintentionally served to infl ect the cluster study with over-
tones of origins, connotations that were later adopted and elaborated by
other scientists, the media, and members of the public.
As we have seen, further developments in understanding the natu-
ral history of AIDS eventually led to the gradual receding of the clus-
ter study’s importance in most scientifi c circles. Because the main thrust
of the study— that a sexually transmissible agent caused AIDS— was on
the “winning” side of history and helped redirect research efforts to-
ward this consensus, the outmoded hypotheses that underpinned it
would escape more careful reevaluation at a later date. In 1982, it had
seemed reasonable to hypothesize that sexual contact could transmit an
infectious agent that would result in an illness appearing within nine to
twenty- two months. This belief, which was necessary to convey the sig-
nifi cance of the network being represented by the cluster, was increas-
ingly diffi cult to sustain over time. Later, when investigators were in-
terviewed for a popular history of the epidemic, they would recall the
study’s importance for the work it accomplished— reorienting a research
consensus— rather than scrutinize how well its constitutive elements had
withstood the test of time. Thus, as part of the edifi ce that sustained this
consensus, “Patient 0” and the cluster study that generated this phrase
would continue to retain their explanatory power for popular audiences
for many years to come.
Chapter Three
“Humanizing This Disease”
As a gay person myself I wasn’t thrilled about Gaetan’s behavior. I don’t see him as any
more typical of a gay man than Jack the Ripper was of the heterosexual— but it did happen.
— Randy Shilts, 19881
On the afternoon of March 15, 1986, Randy Shilts addressed an au-
dience of Canadian journalists at a conference in Vancouver, Brit-
ish Columbia. The Ottawa- ba
sed Centre for Investigative Journalism
(CIJ) was holding its annual conference at Vancouver’s Pan Pacifi c Ho-
tel, and Shilts, on the strength of his AIDS reporting for the San Fran-
cisco Chronicle, had been invited to join four other panelists for a dis-
cussion on AIDS in the media. The reporter had accepted the invitation,
though it was not the sole reason for his Vancouver visit. A month later,
the Vancouver- based conference coordinator wrote to thank Shilts for
his participation, adding that she hoped “your other business in Vancou-
ver worked out well.”2 She could have had little idea of how successful
his research expedition had been. During his twenty- four- hour Vancou-
ver stay, Shilts had managed to gather nearly all of the background in-
formation he would need to tell his version of the “Patient Zero” story, a
tale that would ultimately bring the author and his history of the Ameri-
can AIDS epidemic to international attention.
In front of his peers at the conference, Shilts delivered what would
become his standard criticism of the mainstream media’s response to
1. Philip Young, “Patient Zero: Man Who Gave the World AIDS,” Northern Echo
[High Wycombe, UK], April 9, 1988, 6.
2. Anne Mullens to Randy Shiltz [ sic], 24 April 1986, folder 23, box 34, Shilts Papers.
140
chapter 3
AIDS, complaining about the lack of coverage the disease had received
when it had appeared to affect mostly homosexuals and about the “most
shameless press release journalism that’s existed since the Vietnam war.”
In addition to outlining the Freedom of Information requests that he
had fi led to interrogate the US government’s response to the epidemic,
Shilts explained that “the other focus of, of the reporting that I was in-
terested in, was in terms of humanizing this disease.” He slowed his typ-
ically rapid delivery to emphasize these last three words, before pick-
ing up his pace once more: “I mean these people are, who we’re dealing
with, are human beings, and, and, and I am not so cynical as to believe
that the fact that these people, that, that because these people are gay or,
or where essentially all we have this number of a percent of gay men in
San Francisco getting it, I think that people can relate to anybody if you,
Patient Zero and the Making of the AIDS Epidemic Page 23
