Snowball in a Blizzard

by Steven Hatch

  Couric’s history with cancer coverage began with the heartbreaking loss of her husband to colon cancer at age forty-two. Following his death, she assumed an activist stance toward colon cancer, using her considerable journalistic muscle at the Today show to promote cancer screening awareness. She underwent an on-air colonoscopy in 2000, and the number of colonoscopies in the United States was estimated to have increased by 20 percent the rest of that year following the episode. In 2005 she underwent a mammogram as part of Breast Cancer Awareness Month. She cofounded the group Stand Up to Cancer, whose aim is to provide grants to mainstream scientists focused on innovative research and has thus far raised more than $200 million. Her advocacy for cancer research and public health measures such as screening—no controversies were in evidence as part of this journalism—won her the Award for Distinguished Public Service by the American Association for Cancer Research in 2013. It was harder to find a journalist friendlier to the cause of cancer eradication than Couric. So it must have come as a surprise to the cancer research establishment later that year when Couric, who by then had become the host of an afternoon talk show called Katie, decided to air a segment titled “The HPV Vaccine Controversy.”

  It was surprising principally because there was no controversy, at least among public health officials. HPV is a sexually transmitted virus with dozens of strains. Four of these strains are associated with the majority of cases of cervical cancer; having immunity to these cancer-causing strains should reduce the total number of cervical cancer cases, and Gardasil works by providing this immunity. HPV is also associated with other genital cancers as well as genital warts, and a vaccine rollout should reduce the incidence of these diseases as well. Gardasil is well tolerated; a study of nearly 300,000 young women from Sweden published in the British Medical Journal in 2013 found no serious adverse events associated with its administration. So there was fairly strong consensus that this was a big step forward in the field of cervical cancer. There simply wasn’t a raging debate going on among specialists.

  What is controversial, however, is teen sex, and the recommendations are to provide vaccination for HPV at age eleven or twelve, prior to the onset of presumed sexual activity in mid- to late teen years by many or even most kids. That recommendation has angered socially conservative groups who believe premarital sex is morally abhorrent, and further believe the recommendation for HPV vaccination encourages such behavior. (No evidence has ever supported this claim.) These groups have found allies with antivaxxers, and the rollout of the vaccine has generated a backlash against Gardasil and its maker, Merck, as well as GlaxoSmith-Kline, the maker of another HPV vaccine called Cervarix, which is likewise accepted by the CDC and other professional groups as part of the recommended vaccinations. Regardless of the social policy discussion, in terms of the risk-to-benefit analysis, the HPV vaccine was a slam dunk.

  Because there was no actual scientific controversy, Couric was obligated to spin a tale of an indifferent scientific medical establishment that was paying no attention to the bodies accumulating outside the clinic door. The centerpiece of the Katie episode featured Lauren Mathis, who claimed that Gardasil nearly killed her. Her story is the mirror image of Timothy Ray Brown’s: a medical dystopia in which ignorant doctors push her, jab by jab, toward death. Lauren’s story, and that of another girl named Christina Tarsell whose mother claimed Gardasil had killed her, were given the lion’s share of time. Toward the end, about three minutes were set aside for a pediatrician from Massachusetts General Hospital, Dr. Mallika Marshall, to try to counter the claims that the Katie episode had so clearly accepted at face value. Dr. Marshall’s eloquence and poise notwithstanding, it was largely a failure because the millions of viewers who had tuned in throughout the segment would no doubt have already concluded that Gardisil unleashes horrors aplenty. Indeed, given the overall tenor of Katie’s teach-the-controversy approach, Marshall’s presence simply allowed the producers of the show to provide window dressing to the notion that the show was somehow being objective and representing “both sides of the debate.”

  Why can we be so certain that HPV vaccination is safe, and that Katie Couric not only misrepresented all the reliable evidence about its effectiveness, but essentially thrust a set of nonsensical claims at a credulous public, aimed directly at her afternoon TV-watching demographic group, with the apparent intent to stop moms from vaccinating their kids for HPV? Why should we think that her skepticism should have been applied at least with equal, if not more, force to the claims of these two mothers than to the vaccine manufacturers and the public health officials who work with them?* What evidence allows us to conclude that she used the uncertainty inherent in claims of vaccine safety into a counterproductive exercise in sophistry?

  In asking this question, I don’t mean to imply that skepticism isn’t warranted for public health officials or drug manufacturers or doctors at prestigious academic institutions. The problem with the Katie episode on HPV vaccines, and other reportage in this vein, is that the skepticism being applied to “the establishment” is totally out of proportion to the existing data indicating that “the establishment” got this one right. She allowed the accusations of HPV harm by her guests to stand without even the slightest inclination to signal to her audience that they might want to treat those accusations with some caution. Therefore, in the unspoken logic of the show, if these claims are good enough for Katie, why shouldn’t they be good enough to persuade a member of the audience?

  Before I answer those questions, I should return to the idea of the spectrum of certainty. As I will show, HPV vaccination is a practice whose value is not completely at the far, positive end of the spectrum of certainty, but that’s mainly because we have yet to count in reality how many lives have been saved from preventing cervical cancer, which will take some time to do, as cervical cancer takes decades to manifest itself. Yet, as we will see, the risks of HPV vaccination are negligible, so even potential but not quite fully proven benefits, at the cost of vanishingly small risk, move us away from the absolute middle of the spectrum.

  The answers to the questions above can be found in a variety of places, but a good example of how data is misunderstood by people who have already reached their desired conclusion can be found in a national database known as the Vaccine Adverse Events Reporting System, or VAERS. Some of the key claims made by HPV antivax groups rely heavily on data from VAERS, and, at first blush, the evidence they marshal would appear to be damning, especially because VAERS is a database not run by some antiestablishment alternative medicine group, but by the US federal government itself, in a joint effort of the CDC and the FDA. In the case of HPV, those hostile to the vaccine cite VAERS data indicating that, by the end of 2013, nearly 25,000 adverse events were associated with the vaccine, which included an even more shocking number of approximately one hundred deaths (the number can vary depending on the source). Why won’t public health authorities sound the alarm, or even recognize the threat, of a vaccine that kills so many?

  To appreciate the answer, one must have a nodding familiarity not only with the guts of the VAERS system, but also with the process by which a drug or vaccine receives approval from the FDA. In the vast majority of cases, a so-called Phase III study is required of manufacturers to demonstrate that whatever they propose to bring to market is not only reasonably safe but also that it is effective at doing whatever the manufacturer claims it can do. In the case of Gardasil, that meant that Merck had to enroll nearly 12,000 young women, provide the vaccine, and follow them for three full years, during which time Merck repeatedly evaluated them to see what side effects they might have suffered from and whether the vaccine prevented the condition of cervical intraepithelial neoplasia—the forerunner to cervical cancer. Those results, which did indeed show that the vaccine prevented a significant number of cases of this protocancer, were published in the New England Journal of Medicine in 2007.

  The safety of Gardasil was also studied in this paper, and the results have direct b
earing on how one can interpret the VAERS data. Of the roughly 6,000 women in the vaccine arm of the trial, there were forty-five “serious adverse events”—the kind that would reflect the sort of data seen in VAERS. But because a Phase III study performed for FDA approval is very carefully controlled, there is an equal arm of subjects who don’t get the actual vaccine (they receive a dose of fluid lacking the virus-like particle that produces immunity, essentially getting saltwater). Therefore the researchers were able to look at the number of serious adverse events in that group as well. Because unusual and serious illnesses can occur in a small number of people at random, and because Merck was following 12,000 women for three full years, the placebo arm allowed them to compare the total number of serious adverse events between the groups.

  Similar to the vaccine group, the number of serious adverse events in the placebo group was fifty-four. In raw numbers, this amounted to a higher percentage of cases, but in terms of statistical analysis, it was well within the range that one would expect random variation if a small number of serious medical problems were randomly distributed between two groups—that is, forty-five versus fifty-four is different, but not so different when the total number of each group is 12,000, and is probably due to pure chance. The lack of significant difference meant two things: first, because there was no evidence of additional harm in the vaccine arm, it almost certainly meant it was a very safe vaccine; second, not only were there an equal number of serious adverse events in both groups, it was also a very small number of people who had such problems. That is what one would expect when following a cohort of thousands of young women for three years. The vast majority of them will be fine, but a very small percentage will have something terrible—and even inexplicable—happen to them, whether they received a vaccine, took a pill, worked a particular kind of job, lived in a certain part of the country, or any number of other factors that could in theory be measured.

  The Phase III trial is about as carefully constructed a scientific experiment as can be done on large numbers of humans, and the data that is produced is significantly more reliable than that produced by other ways of investigating the benefits and harms of a drug or vaccine. We saw this in the previous chapter when we looked at how hormone replacement therapy seemed to be a wonder drug when analyzed through the lens of other kinds of studies; when various Phase III–type trials were done and the numbers between those treated and those not were compared, those medications not only didn’t achieve the benefits everyone was so certain would be there, but they actually came with a small amount of harm. Viewed in this light, the evidence of the overall safety of Gardasil and Cervarix (whose Phase III trial included 18,000 subjects followed over the same time) was about as good as it could get.

  Compared to the data generated in a Phase III trial, the VAERS database so frequently invoked in the pseudodebate about HPV vaccination is a different beast altogether. VAERS is a component of the regulatory oversight of medications and vaccines after they have been approved, known as postmarketing surveillance. The means by which data gets entered into VAERS bears no resemblance to the strict rules of evidence required of drug companies performing Phase III trials. Anyone can file a report to the VAERS database, at any time, regardless of when the vaccine was administered.* The database makes no judgment as to which claims might raise genuine concern for vaccine safety because no supporting evidence is required to link the timing of the vaccine to the reported symptoms. Moreover, there is no control group that can be used for purposes of comparison. In the words of one medical scientist who blogs under the name of Orac, VAERS is the medical database equivalent of an unmoderated Internet discussion board, where pretty much anything goes. The hope, as the VAERS website itself notes, is that the system can be used to enable “the early detection of signals that can then be more rigorously investigated.” In other words, the CDC and FDA do not believe that it should be used as a source of reliable information in its pure, raw form.

  More than two-thirds of VAERS reports are entered by vaccine manufacturers themselves as well as health-care providers. The remainder come from a variety of sources including state immunization programs and the vaccine recipients themselves, the latter group accounting for about 10 percent of total reports.

  Reviewing a specific report in VAERS allows us to see how it actually works. Let’s look at the brief story of case number 29281 from the 2013 VAERS report, which is freely available to anyone wishing to download it, because it serves as a reasonably good example of the difficulties interpreting this database. Because this is already public information, and because we do not know who this person is, I haven’t altered any details here. The report, submitted in late December of that year, describes a young woman whose life had gone awry in the back half of her high school years, and whoever wrote the report—the use of “we” below could indicate a parent was the author—overtly links this decline to Gardasil:

  Patient got the first Gardasil shot 7-25-12, the second 9/26/12, and the third shot 1/28/13. Im not sure the exact onset date except to say she got the second and third shot during her junior year in high school and she was continually telling me how fatigued, tired, dizzy, light headed and tired she was. She also had continued to complain about her heart racing, short of breath, heart pains, sleeping difficulty, chest pain, rib pain under her breast, brain fog, headaches, back aches, stomach pain, limb heaviness, neck pain random pains everywhere, seeing different shapes and colors in her vision, severe hip pain, electrical shock pain, and numerous other pains. In Aug 2013 (the beginning of her Senior year in high school) she was walking slowly like she had rode a horse to long, Sept 2013 she needed a walker because she was starting to trip and had trouble standing, Oct 2013 she had to use a wheelchair and unable to walk by herself. We have had MRI, bone scans, CT scans, xrays and numerous bloodwork [sic].

  What’s going on here? It’s very difficult to know. We can make a reasonable guess that this young woman has become the victim of a tragic neurologic disorder, but the precise diagnosis cannot be divined from this snippet. Whether this is due to an oversight on the part of an evaluating physician or the manner in which the patient has attempted to obtain a diagnosis (are all these tests from the emergency room, for instance?), we have no idea. Could this be multiple sclerosis, a rare but not unheard-of condition in teenage girls? If it is—in fact, regardless of what it is—there is no way to know from the report whether her condition is a consequence of receiving the vaccine. This is another version of the correlation/causation problem we saw in the previous chapter.

  We also cannot know whether this report is an instance in cognitive bias, because perhaps the author has been psychologically primed to link the two events in a similar manner to that of the psychiatrists of the Rosenhan experiment we witnessed in the first chapter, who themselves linked harmless neurologic phenomena to deep psychiatric illness. Had the author just watched the Katie episode, and under the power of the dark suggestions from the show, or even by encouragement from friends who had seen the show and planted the idea, suddenly and belatedly “realized” that the patient had received Gardasil a few months before the onset of her symptoms?* If so, can this really be considered reliable data?

  The episode aired December 4, and the report was filed December 29.

  At the time case number 29281 was entered in the database, more than 10 million girls and boys had received the vaccine. Whenever you do anything to 10 million people—just give them a glass of water—some of those people will develop strange medical symptoms, and an even smaller subset will develop serious medical problems. That’s because they were going to develop them anyway, so the default assumption that a report like 29281’s implies causation indicates a deeply naïve view of the VAERS data.

  Despite this, virtually all of the literature put out by groups opposed to the HPV vaccine, whether online or on paper, trumpets the number of serious adverse events, and deaths, in the VAERS database as proof of the giant cover-up perpetrated on an unsuspecting pu
blic.

  None of this was discussed in the Katie episode. There simply wasn’t enough time for it because all the oxygen available was devoted to two horror stories by mothers who are convinced of the vaccine’s evils, data analysis be damned. Given her powerful platform, and the fact that she brought major credibility to the subject of cancer reporting, Couric could have taken a chance to use her program to ask a simple question on her audience’s behalf: How do we know when a vaccine is associated with harms? Surely she and her producers were aware of the rumors about HPV vaccination that were circulating among the kinds of people likely to tune into her show. They were probably likewise aware that this demographic was increasingly suspicious of the HPV vaccine, a fact that Katie Couric herself in all likelihood realized was due to pseudoscientific misinformation of the kind illustrated by the VAERS story. She could have used the opportunity to explain why the theory that HPV vaccination is dangerous is almost certainly nonsense, but she didn’t, going all in for an approach celebrating credulousness and making a virtue of misrepresentation, an exercise in Kool-Aid consumption that had all the appearance of a desperate grab for the approval of a viewing audience different in nature from those who so admired her during her stint at Today.

  Backlash

  As a testament to the perceived influence of a show like Katie, an eruption of criticism ensued within hours of the show’s airing, and the episode itself became a news story in its own right. Whether it was the kind of publicity the producers of Katie were seeking is unclear, but the mainstream media coverage wasn’t favorable, as the (entirely justifiable) story line became “Couric does irresponsible journalism.” A CBS news story written the following day documented the consensus opinion that Couric had engaged in innuendo of the worst sort. It quoted bioethicist Arthur Caplan, whose thoughts were representative: “The show was kind of inexcusable in terms of damage done. . . . [T]he problem in TV and all media, [is that] human interest drives the story. In science and public health, it doesn’t.”