The CDC’s questionable estimates of annual flu hospitalizations and deaths align with its fear marketing strategy to increase demand for flu vaccines.
The US Centers for Disease Control and Prevention (CDC) claims that its recommendation that everyone aged six months and up should get an annual flu shot is firmly grounded in science. The mainstream media reinforce this characterization by misinforming the public about what the science says.
A New York Times article from earlier this year, for example, in order to persuade readers to follow the CDC’s recommendation, cited literature reviews of the prestigious Cochrane Collaboration to support its characterization of the influenza vaccine as both effective and safe. The Times said the science showed that the vaccine represented “a big payoff in public health” and that harms from the vaccine were “almost nonexistent”.
What the Cochrane researchers actually concluded, however, was that their findings “seem to discourage the utilization of vaccination against influenza in healthy adults as a routine public health measure” (emphasis added). Furthermore, given the known serious harms associated with specific flu vaccines and the CDC’s recommendation that infants as young as six months get a flu shot despite an alarming lack of safety studies for children under two, “large-scale studies assessing important outcomes, and directly comparing vaccine types are urgently required.” The CDC also recommends the vaccine for pregnant women despite the total absence of randomized controlled trials assessing the safety of this practice for both expectant mother and unborn child. (This is all the more concerning given that multi-dose vials of the inactivated influenza vaccine contain contain the preservative Thimerosal, which is half ethylmercury by weight. Ethylmercury is a known neurotoxin that can cross the blood-brain barrier and accumulate in the brain. It can also cross the placental barrier and enter the brain of the developing fetus.)
The Cochrane researchers also found “no evidence” to support the CDC’s assumptions that the vaccine reduces transmission of the virus or the risk of potentially deadly complications—the two primary justifications claimed by the CDC to support its recommendation.
The CDC nevertheless pushes the influenza vaccine by claiming that it prevents large numbers of hospitalizations and deaths from flu. To reinforce its message that everyone should get an annual flu shot, the CDC claims that hundreds of thousands of people are hospitalized and tens of thousands die each year from influenza. These numbers are generally relayed by the mainstream media as though representative of known cases of flu. The aforementioned New York Times article, for example, stated matter-of-factly that, of the 9 million to 36 million people whom the CDC estimates get the flu each year, “Somewhere between 140,000 and 710,000 of them require hospitalization, and 12,000 to 56,000 die each year.”
What is not being communicated to the public is that those numbers do not represent known cases of influenza. They do not come directly from surveillance data, but are rather based on controversial mathematical models that may greatly overestimate the numbers.
Consequently, the public is routinely misinformed about the impact of influenza on society and the ostensible benefits of the vaccine. Evidently, that’s just the way the CDC wants it, since the agency has also outlined a public relations strategy of using fear marketing to increase demand for flu shots.
The CDC’s “Problem” of “Growing Health Literacy”
Before looking at some of the problems with the CDC’s estimates, it’s useful to examine the mindset at the agency with respect to how CDC officials view their role in society. An instructive snapshot of this mindset was provided in a presentation by the CDC’s director of media relations on June 17, 2004, at a workshop for the Institute of Medicine (IOM).
In its presentation, the CDC outlined a “‘Recipe’ for Fostering Public Interest and High Vaccine Demand”. It called for encouraging medical experts and public health authorities to “state concern and alarm” about “and predict dire outcomes” from the flu season. To inspire the necessary fear, the CDC encouraged describing each season as “very severe”, “more severe than last or past years”, and “deadly”.
One problem for the CDC is the accurate view among healthy adults that they are at not at high risk of serious complications from the flu. As the presentation noted, “achieving consensus by ‘fiat’ is difficult”—meaning that just because the CDC makes the recommendation doesn’t mean that people will actually follow it. Therefore it was necessary to create “concern, anxiety, and worry” among young, healthy adults who regard the flu as an inconvenience rather than something to be terribly afraid of.
The larger conundrum for the CDC is the proliferation of information available to the public on the internet. As the CDC bluntly stated it, “Health literacy is a growing problem”.
In other words, the CDC considers it to be a problem that people are increasingly doing their own research and becoming more adept at educating themselves about health-related issues. And, as we have already seen, the CDC has very good reason to be concerned about people doing their own research into what the science actually tells us about vaccines.
One prominent way the CDC inspires the necessary fear, of course, is with its estimates of the numbers of people who are hospitalized or die each year from the flu.
The Problems with the CDC’s Estimates of Annual Flu Deaths
Among the relevant facts that are routinely not relayed to the public by the media when the CDC’s numbers are cited is that only about 7% to 15% of what are called “influenza-like illnesses” are actually caused by influenza viruses. In fact, there are over 200 known viruses that cause influenza-like illnesses, and to determine whether an illness was actually caused by the influenza virus requires laboratory testing—which isn’t usually done. Furthermore, as a the authors of a 2010 Cochrane review stated, “At best, vaccines may only be effective against influenza A and B, which represent about 10% of all circulating viruses” that are known to cause influenza-like symptoms. (That’s the same review, by the way, that the Times mischaracterized as having found the vaccine to be “a big payoff in public health”.)
While the CDC now uses a range of numbers to describe annual deaths attributed to influenza, it used to claim that on average “about 36,000 people per year in the United States die from influenza”. The CDC switched to using a range in response to criticism that the average was misleading because there is great variability from year to year and decade to decade. And while switching to the range did address that criticism, other serious problems remain.
One major problem with “the much publicized figure of 36,000”, as Peter Doshi observed in a 2005 BMJ article, was that it “is not an estimate of yearly flu deaths, as widely reported in both the lay and scientific press, but an estimate—generated by a model—of flu-associated death.”
Of course, as the media routinely remind us when it comes to the subject of vaccines and autism (but seem to forget when it comes to the CDC’s flu numbers), temporal association does not necessarily mean causation. Just because someone dies after an influenza infection does not mean that it was the flu that killed him. And, furthermore, many if not most people diagnosed with “the flu” may not have actually been infected with the influenza virus at all, given the large number of other viruses that cause the same symptoms and the general lack of lab confirmation.
The “36,000” number came from a 2003 CDC study published in JAMA that acknowledged the difficulty of estimating deaths attributable to influenza, given that most cases are not lab-confirmed. Yet, rather than acknowledging the likelihood that a substantial percentage of reported cases actually had nothing to do with the influenza virus, the CDC researchers treated it as though it only meant that flu-related deaths must be significantly higher than the reported numbers.
The study authors pointed out that seasonal influenza is “associated with increased hospitalizations and mortality for many diagnoses”, including pneumonia, and they assumed that many cases attributed to other illnesses were actually caused by influenza. They therefore developed a mathematical model to estimate the number by instead using as their starting point all “respiratory and circulatory” deaths, which include all “pneumonia and influenza” deaths.
Of course, not all respiratory and circulatory deaths are caused by the influenza virus. Yet the CDC treats this number as “an upper bound”—as though it was possible that 100% of all respiratory and circulatory deaths occurring in a given flu season were caused by influenza. The CDC also treats the total number of pneumonia and influenza deaths as “a lower bound for deaths associated with influenza”. The CDC states on its website that reported pneumonia and influenza deaths “represent only a fraction of the total number of deaths from influenza”—as though all pneumonia deaths were caused by influenza!
The CDC certainly knows better. In fact, at the same time, the CDC contradictorily acknowledges that not all pneumonia and influenza deaths are flu-related; it has estimated that in an average year 2.1% of all respiratory and circulatory deaths and 8.5% of all pneumonia and influenza deaths are influenza-associated.
So how can the CDC maintain both (a) that 8.5% of pneumonia and influenza deaths are flu-related, and (b) that the combined total of all pneumonia and influenza deaths represents only a fraction of flu-caused deaths? How can both be true?
The answer is that the CDC simply assumes that influenza-associated deaths are so greatly underreported within the broader category of deaths coded under “respiratory and circulatory” that they dwarf all those coded under “pneumonia and influenza”.
In his aforementioned BMJ article, Peter Doshi reasonably asked, “Are US flu death figures more PR than science?” As he put it, “US data on influenza deaths are a mess.” The CDC “acknowledges a difference between flu death and flu associated death yet uses the terms interchangeably. Additionally, there are significant statistical incompatibilities between official estimates and national vital statistics data. Compounding these problems is a marketing of fear—a CDC communications strategy in which medical experts ‘predict dire outcomes’ during flu seasons.”
Illustrating the problem, Doshi observed that for the year 2001, the total number of reported pneumonia and influenza deaths was 62,034. Yet, of those, less than one half of one percent were attributed to influenza. Furthermore, of the mere 257 cases blamed on the flu, only 7% were laboratory confirmed. That’s only 18 cases of lab confirmed influenza out of 62,034 pneumonia and influenza deaths—or just 0.03%, according to the CDC’s own National Center for Health Statistics (NCHS).
Setting aside pneumonia and looking just at influenza-associated deaths from 1979 to 2002, the annual average according to the NCHS data was only 1,348.
The CDC’s mortality estimates would be compatible with the NCHS data, Doshi argued, “if about half of the deaths classed by the NCHS as pneumonia were actually flu initiated secondary pneumonias.” But the NCHS criteria itself strongly indicated otherwise, stating that “Cause-of-death statistics are based solely on the underlying cause of death … defined by WHO as ‘the disease or injury which initiated the train of events leading directly to death.’”
The CDC researchers who authored the 2003 study acknowledged that underlying cause-of-death coding “represents the disease or injury that initiated the chain of morbid events that led directly to the death”—yet they coupled pneumonia deaths with influenza deaths in their model anyway.
At the time Doshi was writing, the CDC was publicly claiming that each year “about 36,000 [Americans] die from flu”, and as just seen with the example from the New York Times, the range of numbers is likewise presented as though representative of known cases of flu-caused deaths. Yet the lead author of that very CDC study, William Thompson of the CDC’s National Immunization Program, acknowledged that the number rather represented “a statistical association” that does not necessarily mean causation. In Thompson’s own words, “Based on modelling, we think it’s associated. I don’t know that we would say that it’s the underlying cause of death.” (Emphasis added.)
As Doshi noted, Thompson’s acknowledgment is “incompatible” with the CDC’s “misrepresentation” of its flu deaths estimates in its public relations messaging. The CDC, Doshi further observed, was “working in manufacturers’ interest by conducting campaigns to increase flu vaccination” based on estimates that are “statistically biased”, including by “arbitrarily linking flu with pneumonia”.
More “Limitations” of the CDC’s Models
While the media present the CDC’s numbers as though uncontroversial, there is in fact “substantial controversy” surrounding flu death estimates, as a 2005 study published in the American Journal of Epidemiology noted. One problem is that the CDC’s models use virus surveillance data that “have not been made available in the public domain”, which means that its results are not reproducible. (As the journal Cell reminds, “the reproducibility of science” is “a lynchpin of credibility”.) And there are otherwise “significant limitations” of the CDC’s models that potentially result in “spurious attribution of deaths to influenza.”
To illustrate, when Peter Doshi requested access to virus circulation data, the CDC refused to allow it unless he granted the CDC co-authorship of the study he was undertaking—which Doshi appropriately refused.
In the New York Review of Books, Helen Epstein has pointed out how the CDC’s dire warnings about the 2009 H1N1 “swine flu” never came to pass, as well as how “some experts maintain that the CDC’s estimates studies overestimate influenza mortality, particularly among children.” While the number of confirmed H1N1-related child deaths was 371, the CDC’s claimed number was 1,271 or more. To arrive at its number, the CDC used a multiplier based on certain assumptions. One assumption is that some cases are missed either because lab confirmation wasn’t sought or because the children weren’t in a hospital when they died and so weren’t tested. Another is that a certain percentage of test results will be false negatives.
However, Epstein pointed out, “according to CDC guidelines at the time”, any child hospitalized with severe influenza symptoms should have been tested for H1N1. Furthermore, “deaths in children from infectious diseases are rare in the US, and even those who didn’t die in hospitals would almost certainly have been autopsied (and tested for H1N1)…. Also, the test is accurate and would have missed few cases. Because it’s unlikely that large numbers of actual cases of US child deaths from H1N1 were missed, the lab-confirmed count (371) is probably much closer to the modeled numbers … which are in any case impossible to verify.”
As already indicated, another assumption the CDC makes is that excess mortality in winter is mostly attributable to influenza. A 2009 Slate article described this as among a number of “potential glitches” that make the CDC’s reported flu deaths the “‘least bad’ estimate”. Referring to earlier methods that associated flu deaths with wintertime deaths from all causes, the article observed that this risked blaming influenza for deaths from car accidents caused by icy roads. And while the updated method presented in the 2003 CDC study excluded such causes of death implausibly linked to flu, related problems remain.
As the aforementioned American Journal of Epidemiology study noted, the updated method “reduces, but does not eliminate, the potential for spurious correlation and spurious attribution of deaths to influenza.” Furthermore, “Methods based on seasonal pattern begin from the assumption that influenza is the major source of excess winter death.” The CDC’s models therefore still “are in danger of being confounded by other seasonal factors.” The authors also stated that they could not conclude from their own study “that influenza is a more important cause of winter mortality on an annual timescale than is cold weather.”
As a 2002 BMJ study stated, “Cold weather alone causes striking short term increases in mortality, mainly from thrombotic and respiratory disease. Non-thermal seasonal factors such as diet may also affect mortality.” (Emphasis added.) The study estimated that of annual excess winter deaths, only “2.4% were due to influenza either directly or indirectly.” It concluded that, “With influenza causing such a small proportion of excess winter deaths, measures to reduce cold stress offer the greatest opportunities to reduce current levels of winter mortality.”
CDC researchers themselves acknowledge that their models are “subject to some limitations.” In a 2009 study published in the American Journal of Public Health, CDC researchers admitted that “simply counting deaths for which influenza has been coded as the underlying cause on death certificates can lead to both over- and underestimates of the magnitude of influenza-associated mortality.” (Emphasis added.) Yet they offered no comment on how, then, their models account for the likelihood that many reported cases of “flu” had nothing whatsoever to do with the influenza virus. Evidently, this is because they don’t, as indicated by the CDC’s treatment of all influenza deaths plus pneumonia deaths as a “lower bound”.
For another illustration, since it takes two or three years before the data is available to be able to estimate flu hospitalizations and deaths by the usual means, the CDC has also developed a method to make preliminary estimates for a given year by “adjusting” the numbers of reported lab-confirmed cases from selected surveillance areas around the country. It does this by multiplying the number of lab-confirmed cases by a certain amount, ostensibly “to correct for underreporting”. To determine the multiplier, the CDC makes a number of assumptions to estimate (a) the likelihood that a person hospitalized for any respiratory illness would be tested for influenza and (b) the likelihood that a person with influenza would test positive.
Once the CDC has its estimated hospitalization rate, it then multiplies that number by the ratio of deaths to hospitalizations to arrive at its estimated mortality rate. Thus, any overestimation of the hospitalization rate is also built into its estimated death rate.
One obvious problem with this is the underlying assumption that the percentage of people who (a) are hospitalized for respiratory illness and have the flu is the same as (b) the percentage of those who are hospitalized for respiratory illness, are actually tested, and test positive. This implies that doctors are not more likely to seek lab confirmation for people who actually have influenza than they are for people whose respiratory symptoms are due to some other cause.
Assuming that doctors can do better than a pair of rolled dice at picking out patients with influenza, it further implies that doctors are no more likely to order a lab test for patients whom they suspect of having the flu than they are to order a lab test for patients whose respiratory symptoms they think are caused by something else.
The CDC’s assumption thus introduces a selection bias into its model that further calls into question the plausibility of its conclusions, as it is bound to result in overestimation. In a 2015 study published in PLoS One that detailed this method, CDC researchers acknowledged that, “If physicians were more likely to recognize influenza patients clinically and select those patients for testing, we may have over-estimated the magnitude of under-detection.” And that, of course, would result in an overestimation of both hospitalizations and deaths associated with influenza.
Caveats such as that, however, are not communicated to the general public by the CDC in its press releases or by the mainstream media so that people can make a truly informed choice about whether it’s worth the risk to get a flu shot.
In summary, to avoid underestimating influenza-associated hospitalizations and deaths, the CDC relies on models that instead appear to greatly overestimate the numbers due to the fallacious assumptions built into them. These numbers are then mispresented to the public by both public health officials and the mainstream media as though uncontroversial and representative of known cases of influenza-caused illnesses and deaths from surveillance data. Consequently, the public is grossly misinformed about the societal disease burden from influenza and the ostensible benefit of the vaccine.
It is clear that the CDC does not see its mission as being to educate the public in order to be able to make an informed choice about vaccination. After all, that would be incompatible with its view that growing health literacy is a threat to its mission and an obstacle to be overcome. On the other hand, a misinformed populace aligns perfectly with the CDC’s stated goal of using fear marketing to generate more demand for the pharmaceutical industry’s influenza vaccine products.
This article is an adapted and expanded excerpt from part two of the author’s multi-part exposé on the influenza vaccine. Sign up for Jeremy’s newsletter to stay updated with his work on vaccines and receive his free downloadable report, “5 Horrifying Facts about the FDA Vaccine Approval Process”.