On poor science communication in the media: a case study on the flu vaccine

This morning I was directed to a blog post by Bill Tieleman in The Tyee. The post dealt with the flu vaccine and was titled Time to End Expensive, Ineffective Forced Flu Shots. This is the latest of his, now yearly, tirades on the topic in the Tyee. His previous efforts being in December 2014 (Are Flu Shots as Effective as Billed?) and December 2013 (More Evidence Against Forced Flu Shots). These are in addition to his similarly themed articles in Vancouver 24 Hrs (2013 and 2014 ) and his blog (Oct 2013Dec 2013 and 2014). One could almost suggest that he is a one-man content provider for the anti-flu shot brigade. Mr. Tieleman’s yearly pieces are similar in form and content and provide excellent fodder for any class studying the communication of science.  As I have written numerous times, one of the roles of this blog is to point out examples of problematic science communication and for those not able to take a science communication class, I will now examine parts of Mr. Tieleman’s latest article for you.

Let’s start with a simple explanation of one of the critical terms used in these articles: “vaccine effectiveness”. Vaccine effectiveness is the “ability of a vaccine to prevent outcomes of interest in the real world”. To further clarify, if a vaccine has 50% effectiveness that means it reduces the likelihood of getting the flu by 50%. This does not represent a one-time deal, it is a seasonal effect. During the flu season you can be exposed to the influenza virus numerous times a day, numerous days a week, numerous weeks in the year and 50% effectiveness means that over that entire time the vaccine has reduced your likelihood of getting the flu by 50%. This is not like turning the key on your car and having a 50% chance it will start, which essentially represents the analogy Mr. Tieleman implies in his article.

Talking analogies, anyone who has read my writing knows one of the ways to make science more reader-friendly is to use analogies and personal anecdotes. Of course the risk with analogies is that a bad analogy can distract from your narrative. In his articles Mr. Tieleman uses inapt analogies (typically comparing vaccines to consumer goods) to address vaccine effectiveness and in doing so he miscommunicates the actual effectiveness of the influenza vaccine. In his most recent article he asked: “Would you buy a car or a television that had a 50 per cent chance of working at best and at worst only seven per cent”? As I point out above, vaccine effectiveness is not some one-time event like turning the key in your ignition. Individuals are exposed to the influenza virus repeatedly over the course of the flu season. Instead a more apt analogy would involve some medical device or public safety innovation that had a comparable effectiveness in reducing a negative outcome. Happily for our discussion such an innovation exists, it is called “the seatbelt”. A properly used seatbelt reduces your likelihood to be injured in the event of an automobile accident. Seatbelts aren’t perfect, however; and they won’t prevent all injuries. No seatbelt in the world will save your life if you get t-boned by a semi but seatbelts reduce serious crash-related injuries and deaths by, yes you guessed it: about 50%. Thus using the same descriptive criteria as is used for vaccines the effectiveness of seatbelts would be defined as 50%. As presented by the CDC, the flu vaccine since 2004 has varied in effectiveness between 10% and 60% (not the 50% as indicated in his article). Funny, I don’t see Mr. Tieleman suggesting that we should be giving up on seatbelts in automobiles because they only have an effectiveness of 50%.

Now sticking with our seatbelt analogy, as I pointed out above the best seatbelt in the world will not save your life if you get T-boned by a semi. Similarly, the best vaccine in the world won’t work if it is designed for the wrong strain of influenza. The problem is that there is not one single human influenza virus, rather there are dozens of strain/subtype variations and experts must pick which viruses to include in the vaccine many months in advance in order for vaccine to be produced and delivered on time. Sometimes they get it wrong and like last year’s vaccine you end up with lower effectiveness. That is not a reason to abandon a good program. Even in a bad year (like last year where we saw 23% effectiveness) the result is a substantial reduction in illness rate. This brings me back to Mr. Tieleman’s article. In it he emphasizes that last year’s effectiveness was seven percent. That number appears to be drawn from interim estimates for vaccine effectiveness published by Dr. Skowronski (et. al. Eurosurveillance Jan 2015). Since that publication was produced prior to the end of the 2014-2015 flu season I would tend want to trust the CDC result.

Returning to Mr. Tieleman’s article, it starts with a quotation from Glasgow Dr. Margaret McCartney taken from an article in the British Medical Journal. Dr. McCartney’s article relies heavily on several “Cochrane Reviews”. I have written about Cochrane Reviews in the past and have pointed out that they represent one of the higher quality meta-analyses out there but are deliberately very limited in scope. In Dr. McCartney’s article she cites three Cochrane Reviews including one restricted to healthy adults and children. Its conclusion, as Dr. McCartney points out in her article, was that for every 7 healthy children vaccinated 1 case of influenza was avoided. From a medical perspective she views this as a wasted effort. I, as a policy type, view it as no such thing. Rather to my mind this is a terrific result that pretty much justifies the childhood flu vaccine program all on its own.

Consider the Cochrane Review number from a policy perspective. Take a class of 21 kindergarten kids. Vaccinating that class would avoid 3 cases of influenza. Now consider that a typical case of the flu can usually last up to 7 to 10 days. That represents missing 5 days of school. Those 5 days are 5 days when the sick child has to be at home under the care of a care-giver; in my family’s case that means me (using sick time from work) or my wife (using her family leave from her school). Each day our child is sick costs either my employer, or our provincial government (my wife’s ultimate employer) money. Consider that an average teacher gets paid about $200 per actual school day. Missing those 5 days to take care of a sick child represents a direct cost to our government of around $1000. Considering the flu shot costs about $20/shot  those 7 shots cost the government around $140. In return they can generate a direct reduction of employment costs by about $1000. That represents $860 in reduced government costs. Don’t even get me started on the cost savings when you also include hospital and emergency room admissions. Can you show me any other health care intervention that saves the government over 7 times the cost of the program?

Getting back to confusion about numbers, Mr. Tieleman uses interesting arguments regarding some of the numbers used in the reporting of influenza statistics. In his article he writes:

The Public Health Agency of Canada states on its website: “It is estimated that, in a given year, an average of 12,200 hospitalizations related to influenza and approximately 3,500 deaths attributable to influenza occur.”

But PHAC also says elsewhere on the same website that in the past four years the range of flu-related reported fatalities was 100 to 600 annually, while hospitalizations ranged from 2,000 to just under 8,000.

And in the 2015-16 season, the number of deaths with statistics to Jan. 9 is 10 while hospitalizations are 201.

The Public Health Agency puts out a number of documents, a summary document that includes summary information and the research basis for their decisions and then actual statistics on “Influenza Hospitalizations and Deaths”. Unfortunately, you cannot compare the two sets of statistics (as Mr. Tieleman appears to have done) because the qualifiers associated with the data.

The problem with influenza is that it only becomes an official case if a sample was collected; submitted to the lab; and characterized by strain/sub-type. Conducting a laboratory test for influenza is both expensive and time-consuming and typically has no effect on how the illness will be treated. As a result, in the vast majority of emergency room visits, the emergency room doctors do not bother to request this expensive test. Rather they report the patient as having an influenza-like illness (ILI) and leave it at that.

As for the “flu-related fatalities” not matching between the two PHAC sections, that is because the flu is not the sole (or even major) cause of death in most cases, rather it is a contributing cause and thus the numbers do not match up perfectly. Instead they rely on “regression modelling” (note the scare quotes in his article) which allows scientists to peel out the effect of the influenza on increased death rates. In the old days they used to call the influenza the “old man’s friend” because it was the disease that ultimately weakened the severely ill enough to allow them to die of their diseases rather than lingering on with a debilitating ailment in an era prior to the development of effective palliative care. As with the emergency-room visits, in our modern era doctors don’t necessary see the need to actually submit samples for confirmation of flu strain when an elderly patient with a pre-existing condition dies. Thus, only a small percentage of the deaths “attributed to influenza” are actually confirmed as being a caused by influenza. So when we are talking deaths from influenza, we are talking only about cases where a person has died and because there was not an underlying condition a test was undertaken to confirm the diagnosis of influenza.

Finally, by using the numbers presented above it is possible to understand the cost of all these influenza cases to our medical system. Recognize that the first number presented in his quotation is for “hospitalizations” only. That means that the patient was admitted into the hospital ward. It does not consider outpatient and emergency room visits. Consider that in the last two weeks of December 2015, the proportion of visits to BC Children’s Hospital Emergency Room (ER) attributed to ILI represented, 16% and 18%  of all visits. That is a huge number. An even marginally effective vaccine can cut the number of hospital visits dramatically resulting in less crowded emergency rooms and bundles of saved government money. Taking a look at this resource from the CDC shows how incredibly effective the flu vaccine has been at reducing hospital admissions and thus reducing our national medical bill.

In reading Mr. Tieleman’s writing on the flu vaccine it is clear that his major complaint is that he does not believe that health care workers should be forced to be vaccinated. The current research supports him in this respect. The most recent Cochrane Review on the subject: Influenza vaccination for healthcare workers who care for people aged 60 or older living in long-term care institutions supports the idea that forced vaccination does not appear to reduce death rates in long-term care institutions, but notes the absence of good double-blinded research on the subject.

To be clear, I question the professionalism and empathy of any health care provider who chooses to forgo the flu shot. The data is clear that the flu shot reduces incidences of infection. Logically this decreases the likelihood that they will unknowingly transmit the virus to their often immuno-compromised and fragile patients. That being said, at this time conclusive data does not appear to be present to justify coercing health care providers, with threats to their continued employment, into taking the flu shot prior to encountering patients. As for Mr. Tieleman’s articles, it is a pity that in fighting a battle for his perceived constituency (union members who do not want to be coerced into getting vaccinated) he has decided to side-swipe a legitimate program that has a demonstrable effect of protecting the young and the elderly and the added benefit of potentially saving the government a lot of money.

This entry was posted in Canadian Politics, Chemistry and Toxicology, Risk, Uncategorized. Bookmark the permalink.

5 Responses to On poor science communication in the media: a case study on the flu vaccine

  1. DHR says:

    Perhaps Mr. Tieleman would decline a lottery ticket if it had only a 50% chance of winning??


  2. Albi says:

    Another great article Blair. Have you ever thought of offering your services to the Tyee?


  3. Old Bruno says:

    It may be of interest to Canadians that, after a long PR campaign of misinformation by people opposed to vaccinations for children, the Australian government announced that, effective January 2016, it would suspend all family & child support payments in cases where children were not immunised. Opposition collapsed in one month.


  4. Alexis says:

    If vaccines are so effective, why don’t the companies that manufacture them stand behind their product by offering full liability? All other pharmaceutical products come with liability–how come vaccines are excluded if they are so safe and effective?


  5. Pingback: Flu shots save both money and lives | A Chemist in Langley

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.