Questioning the CDC 36,000 Deaths from Seasonal Flu Figure

Tuesday, 03 November 2009 12:30

It’s cold and flu season and the news is filled with just how deadly influenza can be. A figure that is frequently repeated in most news outlets when talking about the flu is that 36,000 people die each year from the flu.

Fox News

“H1N1 has been getting top line publicity and it’s important to know that seasonal flu takes a toll year in and year out,” Sebelius said. “But each year, 36,000 Americans die from the seasonal flu.”

New York Times

“… seasonal flu, which kills about 36,000 people annually.”

Wall Street Journal

“The seasonal flu sends an average of 200,000 Americans to the hospital each year, and 36,000 of them die, according to the CDC”

Orlando Sentinel

“According to the Centers for Disease Control and Prevention, 5 percent to 20 percent of the U.S. population is infected with seasonal flu each year… about 36,000 people die — mostly elderly.”

All these news reports leave the reader with a very specific and definitive figure of 36,000 people dying each year directly from the flu. But where did this figure come from? Most people would probably assume that careful statistics are kept to arrive at this figure. However, this 36,000 number that is repeated so authoritatively is actually based on a single study published in 2003. A mathematical model was applied to estimate flu-related deaths by using death certificates where the cause of death was listed as respiratory or circulatory disease.

From the CDC website:

“CDC estimated that about 36,000 people died of seasonal flu-related causes each year, on average, during the 1990s in the United States. This figure includes people dying from complications of seasonal flu. This estimate came from a 2003 study published in the Journal of the American Medical Association (JAMA), which looked at the 1990-91 through the 1998-99 flu seasons. Statistical modeling was used to estimate how many flu-related deaths occurred among people whose underlying cause of death on their death certificate was listed as a respiratory or circulatory disease. During these years, the number of estimated deaths ranged from 17,000 to 52,000.”

The fact is the 36,000 number of deaths is really an estimate and not an exact number as often it is represented. More importantly it is a mathematical estimate based on an assumption that if a death certificate had “respiratory or circulatory disease” listed as a cause of death that it should be counted as a “flu-related” death.

Just how many deaths are directly associated with the flu? According to Peter Doshi in the British Medical Journal:

‘… according to the CDC's National Center for Health Statistics (NCHS), “influenza and pneumonia” took 62,034 lives in 2001 - 61,777 of which were attributed to pneumonia and 257 to flu, and in only 18 cases was flu virus positively identified. Between 1979 and 2002, NCHS data show an average 1348 flu deaths per year (range 257 to 3006).’

So the real statistics show a far lower number of people dying from the flu with only a handful of cases being definitely identified as even having the influenza virus. 257 deaths from the flu in 2001 is only a fraction of the often cited 36,000. The CDC directly admits to using this model because most cases aren’t even tested to see if the flu virus was even involved.

‘In a written statement, CDC media relations responded to the diverse statistics: “Typically, influenza causes death when the infection leads to severe medical complications.” And as most such cases “are never tested for virus infection ... CDC considers these [NCHS] figures to be a very substantial undercounting of the true number of deaths from influenza. Therefore, the CDC uses indirect modeling methods to estimate the number of deaths associated with influenza.”’

An article published in Journal of American Physicians and Surgeons in 2006 is highly critical of the CDC methods.

“The CDC and news media frequently proclaim that there are about 36,000 influenza-associated deaths annually. Review of the mortality data from the CDC’s National Vital Statistics System(NVSS) reveals these estimates are grossly exaggerated. The NVSS reports preliminary mortality statistics and distinguishes between influenza-related deaths and pneumonia-related mortality. When the final report is issued, influenza mortalities are combined with the far more frequent pneumonia deaths, yielding an exaggerated representation of influenza deaths. Pneumonia related mortality due to immunosuppression, AIDS, malnutrition, and a variety of other predisposing medical conditions is therefore combined with seasonal influenza deaths. The actual influenza related deaths for the years 1997 to 2002 ranged from 257 to 1,765 annually. These values are further overestimated by combining deaths from laboratory-confirmed influenza infections with cases lacking laboratory confirmation.”

Looking at how the CDC website has changed over the years and how they report on the flu shows that there was a shift in 2002 from “influenza associated” to an absolute “die from the flu” statement. In 2003, the number of 20,000 was adjusted based on the discussed mathematical model to 36,000 “die from influenza”. The CDC website now currently and slightly more correctly notes that the deaths are “flu-related”.

• Dec 16, 2001 - "In an average year, influenza is associated with more than 20,000 deaths nationwide"
• Jan 24, 2002 - "An average of about 20,000 people per year in the United States die from the flu"
• Feb 9, 2003 - "An average of about 36,000 people per year in the United States die from influenza"
• Nov 02, 2009 - "about 36,000 people die from flu-related causes."

Peter Doshi concludes in his paper that, “If flu is in fact not a major cause of death, this public relations approach is surely exaggerated. Moreover, by arbitrarily linking flu with pneumonia, current data are statistically biased. Until corrected and until unbiased statistics are developed, the chances for sound discussion and public health policy are limited.”

The figure of 36,000 deaths each year from the flu as reported is clearly not based on careful analysis of actual cases that are tested for the influenza virus. Instead, the number is a mathematical model estimate based on a single 2003 study that links death certificates that list “respiratory or circulatory disease” as a cause of death as proof of influenza as causing the death.

Source:

Orlando Sentinel, “Flu 101: What you need to know about seasonal and swine strains”, November 2009

Fox News, “Health Officials: Get Your Seasonal Flu Vaccine Today”, September 10, 2009

New York Times, “As Flu Vaccine Arrives for the Season, Some Questions and Answers”, October 9, 2009

“Two Flu Vaccines, Lots of Questions”, Wall Street Journal, October 18, 2009

Questions and Answers Regarding Estimating Deaths from Seasonal Influenza in the United States, http://www.cdc.gov/flu/about/disease/us_flu-related_deaths.htm

Peter Doshi , Are US flu death figures more PR than science? BMJ, http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1309667

Influenza Vaccination During Pregnancy: A Critical Assessment of the Recommendations of the Advisory Committee on Immunization Practices (ACIP), Journal of American Physicians and Surgeons, Volume 11,Number 2,Summer 2006, http://www.jpands.org/vol11no2/ayoub.pdf