Rarely a day goes by that I do not scan the headlines collected from various blogs and sources by Google Reader. Smashing a recent lull in AIDS news, some pretty outrageous headlines have been breaking through lately. Last week, it was Baby AZeTa, the little girl in Mississippi who researchers claimed was cured of AIDS with AZT and other dated AIDS drugs. This week’s headline seems to be: “No increase in risk of death for patients with well-controlled HIV.” Clearly, the message to be taken away by superficial news consumers is that the pharmaceutical treatments for AIDS have not only rendered the disease manageable, but are harmless, to boot.
Needless to say, my Skeptic Alert went off like a klaxon, and off I went to read more.
This “news”—which has been languishing since it was first published more than four months ago—is suddenly blazing across the Internet only after a meme seed was planted by a official sounding organization that doesn’t even feel the need to disguise its purpose as a front for Big Pharma. What started as a single press release reported to Google Reader yesterday, has become a flood of headlines from dozens of “sources” today, all of them verbatim repeats of the press release, down to the punctuation used in the headline. Don’t make the same mistake I made, thinking that a site called Science Blog is really about science. Its actual stated purpose is to promote press releases from research institutions, under the guise of science.
It took awhile to discern that the source for these reports can be traced to a single press release, which I have not yet been able to track down. However, since the sole original attribution to a presumably “credible authority” in the story is to Veronica Miller, PhD, Director of the Forum for Collaborative HIV Research, I’m willing to risk my own editorial credibility and venture to guess that the Forum is also the source of the press release. Who is this Forum for Collaborative HIV Research? Their Mission statement, as found on their website, sounds admirable enough (emphasis in original):
Founded in 1997, The Forum for Collaborative HIV Research is a public/private partnership at the University of California, Berkeley Washington Campus. The Forum’s mission is to enhance and facilitate HIV research and this is accomplished by bringing together all relevant stakeholders to address emerging issues in HIV/AIDS. Through our work, we identify gaps and impediments, frame issues and help set research strategy. The goal is to optimize care and treatment of those affected by HIV/AIDS and our scope includes research addressing prevention, treatment strategy, health services utilization and health policy.
The Forum includes representatives from government, industry, patient advocates, health care providers, academia and foundations.
To their credit, the Forum does not make it too difficult to peek behind the curtain to see where the funding for this “public/private” partnership comes from. Just click the “Sponsors” link: 
The list reads like a veritable Who’s Who of pharmaceutical AIDS drug manufacturers. At this point, any reader who is uncomfortable with me referring to The Forum as a pharmaceutical front group is welcome to go back to scanning headlines. Now that we know who wants to promote this meme, what about the underlying data that says people taking so-called highly active antiretroviral therapy (HAART) for the rest of their lives can expect to live as long as Average Joe? The source for this news is a report in the November, 2012 issue of AIDS, the Official Journal of the International AIDS Society. Mere mortals are not allowed to read articles in AIDS without a subscription ($779/year), or first purchasing the single article for $59. Needless to say, like most Average Joes, I’ll have to content myself with information available from the abstract available online.
Like so much “research” about people with AIDS, the results being published here are merely a rehash of results previously published in ongoing trials; in this case the ESPRIT and SMART trials. Despite the headlines suggesting otherwise, the authors did not report on all patients taking antiretroviral therapy (ART), but rather a subset of those people. For example, intravenous drug users are excluded from one side of the equation (those taking ART), but not from the control group, aka “general population”. The study group was further pared down to include only those patients on ART who had low viral loads and high CD4 counts. Unfortunately, there is no way to know how many people taking ART failed one or both of these arbitrary criteria for inclusion. At least we now know the definition of “well-controlled” HIV.
The next step was determining the death rate among this select sub-group of ART patients who happen to be performing well on surrogate markers, after three years. Not a lifetime. Not even a decade. Three years. During the three years, there were 62 deaths, out of 3,280 study participants, resulting in a death rate of 18.9 per thousand for three years, or 6.3/1000 per year. This number was then compared, apparently, to overall death rates, according to the Human Mortality Database (HMD). Perhaps the authors provide more clarity in the study report about what data they were actually comparing, to come to the conclusion they did. All of the data on death rates that I am able to find in the HMD, is based on age. Using the raw data to calculate an overall death rate is more work than I’m willing to do, especially since the Centers for Disease Control offers a much more user-friendly interface to mortality data. According to the CDC, the overall death rate in the U.S. is just a hair shy of 8/1000. If this is what the study authors are comparing, it is small wonder that ART patients fared so well after only three years of drugging.
However, the Grim Reaper is not an equal opportunity practitioner. Age is no doubt the single most significant co-factor for death. The median age of those studied was 43 years, according to the abstract. The death rate for 35-44 year-olds in 2007 was only 1.84/1000. Based on this, trial participants were more than three times as likely to die than their counterparts in the general population. Even when compared to a rate of 4.21/1000 for the next oldest age group, 45-54, patients with “well controlled HIV” were 1.5 times more likely to die after only three years on HAART.
Regardless of the actual numbers acquired, the real manipulation is how these findings have been projected into the mass consciousness, via the media. What is the basis for extrapolating three years experience on “highly active” drug treatment for three years over a lifetime? What part of “long term” adverse effects is not understood here? How can cumulative toxicities be studied, without allowing time for accumulation? HAART has been the standard of care for HIV since 1996. The AIDS research literature since then has been filled with reports of increasing “non-AIDS defining illnesses” in people with “well controlled” HIV. Any sane, rational mind has to wonder what role drugs that are known to be toxic (why else do so many of them sport the “Black Box Warning” required by the FDA?) is playing in this new epidemic of illness?
The kind of data mining and cherry-picking found in this latest AIDS “research” reminds me of my days as a Database Specialist, working for the Marketing Department of an area health insurance company 15 years ago. One of my jobs was to query member data and provide results to upper management, as well as to complete RFIs (requests for information) and RFPs (requests for proposals) from large employers and from health care consulting firms. The primary goal of my team was to present the company in as favorable a light as possible, without crossing the line into outright falsification. Fabrication, on the other hand, was not only acceptable, but admirable. I knew that even slight changes to the query parameters would change the results of any information request. For example, changing the age range for a request of data regarding women with cancer by even a few years, would raise or lower the return rate considerably. So, when a RFI asked, for example: “What percentage of women over the age of 60 are receiving care for breast cancer?” we might respond with inflated numbers, followed by an asterisked comment: “We only track this data for women aged 65 and older.”
Just so, these “researchers” have proved to my satisfaction that similarly mining the data with a variety of variables could result in a number that supports a desired message.