Is the flu vaccine really only 10% effective? A sojourn into immunization epidemiology

If you work in IT, you must be familiar with the instances of well-meaning family asking you to fix their computers – even if you develop enterprise Java apps and they have a broken video card, they expect you to be able to do something about it. For epidemiologists, the equivalent is the flu shot. Over the last few weeks, I have been asked by so many of my friends to explain why they should even consider getting the flu shot if it’s, reportedly, only 10% effective. Add to that, of course, the sheer amount of crazy on the internet on the subject (the flu shot will give you cancer/autism/cancertism, here, have these herbs, open your chakras, meditate on these crystals and drink some colloidal silver until you look like Stan Jones).

“I mean, how hard could it be? We have the flu every year,” my friend, a very intelligent and accomplished lady, pointed out to me the other day. “And surely something like the flu is pretty simple, right? You’d think that by now we would have figured out what works and what doesn’t.”

In fact, it’s pretty complicated – and the simplicity of the flu virus makes matters worse, not better.

How vaccines work

Consider the example of the first vaccinated illness, smallpox, caused by an orthopoxvirus called variola virus. In 1796, the English physician Edward Jenner realized that people who have had the relatively harmless zoonotic virus cowpox were resistant to the much more severe smallpox. We know today that this was because as they fought off cowpox, they developed antibodies that identified surface proteins that are shared between orthopoxviridae and thus they had antibodies that were already ‘ready to go’ when they were exposed to smallpox. To this day, the general idea of vaccination is the same: administering a combination of something that has the relevant surface markers and an adjuvant. The surface markers can be provided in a number of different ways, ranging from live pathogens through inactivated pathogens (killed off using radiation or heat, usually), attenuated pathogens (live strains of pathogens that are not pathogenic or less pathogenic, e.g. for tuberculosis, a special strain called BCG is used, which is much less pathogenic than regular M. tuberculosis) to subunit vaccines that contain small but characteristic and recognizable bits of surface proteins or a conjugate of polysaccharide cell walls joined to immunogenic proteins in the case of bacterial vaccines where the bacteria have polysaccharide cell walls that are not very immunogenic, i.e. not very likely to elicit an immune response. The adjuvant is typically a non-toxic irritant that stimulates local immune cells to recognize the administered pathogen or pathogen fragment as foreign by, among others, causing a mild inflammatory reaction that releases cytokines helping to draw B and T lymphocytes to the injection site. The end result is a lasting immunity through the immune system’s ability to ‘remember’ past pathogens it was exposed to (memory B-cells).

The problem with flu

If you have a complex organism like a human or a wombat or a fox, you have a relatively constant form. Most humans look pretty much the same, right? That’s because complex organisms tend to have longer lives, a lower reproduction rate and a lower mutation rate. Simpler organisms, such as bacteria and viruses, mutate all the time. This is behind the problems with antiviral and antibiotic resistance – because of the fast reproduction rate and fast mutation rate, viruses can develop mutations that give it a degree of resistance against particular antivirals. But resistance is not only relevant for antivirals, it is relevant for antibodies, too.

The problem is that a flu vaccine protects only against strains that are sufficiently similar to what the vaccine is specifically targeted against. Flu viruses mutate so fast, they have their own nomenclature. The two influenza viruses A and B are divided into subtypes by looking at the variant of two surface proteins, neuraminidase and haemagglutinin. Each of these variants is denoted by a number, so e.g. H1N1 means the subtype of influenzavirus A that has the type 1 haemagglutinin and type 1 neuraminidiase on its surface, while H3N1 would differ in that it would have type 3 haemagglutinin on its surface. But that’s not all – even within the subtypes there is sufficient variation, and so we get to the strain level. For instance, A/Hong Kong/4801/2014 (H3N2) is a strain of A(H3N2) influenzavirus isolated from Hong Kong in 2014 with the unique strain identifier 4801. This has the same neuraminidase and haemagglutinin configuration as the A(H3N2) influenzavirus strain A/Victoria/361/2011, which was among those vaccinated for in 2012, but is still sufficiently different so that chances are the 2012 vaccine’s H3N2 element would have little efficacy against it.

The problem is that the flu virus changes so rapidly that new strains emerge almost daily. At this point, it’s worth pointing out an important difference between antigenic typing and genetic typing. Antigenic typing looks at surface markers exhibited by a particular pathogen, while genetic typing looks at their genetic material. Strains are all genetically different but may be antigenically similar. For instance, we know that cowpox and smallpox must be antigenically similar despite being genetically very different – much, much more than two flu virus strains! – because the antibody generated by exposure to one is triggered by the other as well. This is good, because there’s an empirical limit as to how many strains can be targeted in a vaccine, and that limit is pretty low – most flu vaccines are tri- or quadrivalent (sometimes people butcher Latin and call it ‘tetravalent’, which is bad Latin but means the same thing), i.e. protect against three or four identified strains. In 2017/8, these are:

  • A/Michigan/45/2015 (H1N1)
  • A/Hong Kong/4801/2014 (H3N2)
  • B/Brisbane/60/2008 (B/Victoria lineage)
  • For quadrivalent vaccines: B/Phuket/3073/2013 (B/Yamagata lineage)

In other words, the vaccine will protect, assuming it’s effective, against all flu viruses that are antigenically identical to these viruses, even if they are genetically different strains.

A digression: *-like viruses and how vaccines are made

If you have read the vaccine leaflet or the CDC website on flu vaccination, you might have noticed that the strain specifications actually say “*-like virus”, e.g. “A/Michigan/45/2015 (H1N1)-like virus”. What’s with that?

There are different ways of manufacturing the flu vaccine. This year, some of the vaccines manufactured in cells (Flucelvax), specifically in a cell line based on cocker spaniel kidney cells (MDCK), most of it is manufactured in eggs (Fluzone, Afluria, Fluarix, FluLaval, Fluvirin and the adjuvanted Fluad vaccine for the elderly) and this year, a recombinant vaccine called Flublok is available. This year, there is no live attenuated vaccine, but in previous years, a live attenuated vaccine nasal spray (FluMist) was available. But all of these viruses are based on the same strains, regardless of how they were cultivated or whether they’re inactivated flu vaccines (IIV) or recombinant flu vaccines (RIV).

The annual vaccine starts life as a set of candidate vaccine viruses (CVVs), grown in eggs by the CDC or one of the other major laboratories (called Collaborating Centers in WHO lingo) in the WHO’s GISRS (Global Influenza Surveillance and Response Systems). A CVV is a virus that is antigenically identical to the corresponding flu virus strain, so e.g. the A/Michigan/45/2015 (H1N1)-like virus is antigenically identical to the actual A/Michigan/45/2015 (H1N1) strain of influenzavirus. The reason that antigenically identical CVVs rather than the actual virus are used in vaccine production has largely to do with the fact that not all viruses grow well in eggs and/or MDCK. This is in particular the case with subtypes of influenzavirus that are pathogenic to birds, as these tend to destroy eggs. For this reason, an attenuated form of the virus is created that causes less severe disease but is still antigenically identical, i.e. it will elicit an adequate immune response that will stimulate the creation of antibodies against the real deal virus.[1] It is worth noting that while the CDC is the US’s contribution to the WHO’s GISRS, decisions on what vaccine viruses to use in America are made by the FDA.

The Gamble

It takes approximately six months from CVV release to sufficient stocks for mass vaccination. For this reason, the GISRS needs to make a decision on what strains to include in the vaccine by February or thereabouts. There are several factors that are considered in selecting the strains:

  1. Is there a suitable CVV available for the candidate strain? If there are no vaccine viruses to manufacture viral antigens for the strain, then it won’t be possible to include them. While relatively infrequent, this does happen from time to time.
  2. Which viruses are going to be the most likely to circulate in the next flu season? Using a wide network of surveillance centres, studies and other data sources, the WHO has quite a lot of information on emerging viral trends and predictions.
  3. What strains are likely to cover the most ground? It is always a pleasant surprise when the strains chosen cover some other strains as well due to identical antigenic typing.

To the flu virus, six months is a virtual eternity. No wonder, then, that sometimes by the time the vaccines are rolled out, a new and possibly antigenically different variant is on its way to give everyone the sniffles. Indeed, most of the time, when the strains identified don’t turn out to be the dominant pathogenic strain, the dominant pathogenic strain ends up being one for which there either was no vaccine virus at the time or was not even known at the time. Nature can outpace even the best prediction.

OK, but 10%? Seriously? That’s crappy.

What is vaccine effectiveness? There are two metrics that must be differentiated here: effectiveness vs efficacy.

Vaccine efficacy is the protective effect of a vaccine in a randomized study, i.e. one where part of the population gets a vaccine while others receive a placebo. Unfortunately for epidemiologists but fortunately for everyone else, randomized vaccine efficacy studies are rarely performed, for the simple reason that it’s unethical to withhold treatment known to be better than placebo when there is a risk of illness and possibly serious sequelae.

Vaccine effectiveness, on the other hand, derives from observational studies, and looks not at a randomized sample but at a real population, and compares the results with those who voluntarily (and knowingly) did not vaccinate. Usually, vaccine effectiveness is given as a single percentage number, with a confidence interval. An effectiveness of 55% (95% CI: 45-65%) means that the flu vaccination is estimated to reduce the chance of laboratory-confirmed flu by 55%, with a 95% confidence interval of 45-65% (meaning that if the study would be performed a hundred times, 95 times the results would be between 45% and 65%). It’s important to note that the effectiveness of a polyvalent vaccine may vary against each of the target strains – typically, flu vaccines are less effective against the faster mutating H3N2 strains than others, for example. A good comparison of the effect of vaccination on the prevalence of influenza-like illness (ILI) with fever in the UK is evident in the charts below.[2]

What about that 10% figure? Actually, it’s a figure from Australia and pertains only to efficacy against a H3N2 strain. In the 2016/17 flu season, the overall VE was a little shy of 40%, while the partial VE for H3N2 was 32%. The H3N2 component did not change from 2016/17, and the CDC estimates that most H3N2 flu viruses circulating in America are still sufficiently similar to the A/Hong Kong/4801/2014-like CVV strains that the flu vaccine will provide adequate protection against them. So perhaps 32% is much more realistic than the 10% estimate.

But what about the risks?

To get the biggie out of the way: no, you can’t get the flu from a flu shot. At all. Ever. You can’t give it to anyone else, either (the thing anti-vaccine ‘advocates’ refer to as ‘shedding’). In the overwhelming majority of cases, the only side effects of the flu vaccine are limited to soreness at the injection site, a mild fever and some general malaise and ‘flu-like symptoms’. The latter have nothing to do with the flu – they are caused by the immune system’s activity rather than anything specific to the flu. The only groups of people who should talk to their doctor before getting a flu vaccine are

  • people who are currently sick (usually, you would have to wait to recover),
  • people with a history of Guillain-Barre Syndrome after vaccination,
  • people on immunosuppressants (who might therefore not be able to mount the immune response necessary for successful immunization and might therefore need to take a break from their immunosuppressants), and
  • people who have anaphylactic allergies to eggs (who should therefore receive the recombinant vaccine).

The vaccine is safe and effective for everyone else. Anti-flu advocates generally trot out the same tired memes, so here are the facts:

  • Guillain-Barre Syndrome: the question of whether vaccines are correlated with GBS, a very serious autoimmune illness of the nervous system, is somewhat controversial. The data show that the added risk by the flu vaccination is about 1.4 to 1.6 cases per million vaccines administered. This figure changes slightly every year, but remains roughly in this range. Importantly, the risk of very serious complications from the flu is much, much higher, while GBS has over the years become eminently treatable when recognized in time.
  • Autism and mercury: absolutely no evidence at all as to the association of autism with thimerosal/thiomersal, a vaccine preservative that contains ethyl mercury. Thiomersal is only contained in multi-dose flu vaccines, specifically Afluria multi-dose vials, FluLaval multi-dose vials, Fluzone multi-dose vials and Flucelvax multi-dose vials. As such, you can get single-dose pre-filled syringes or single-dose vials that contain absolutely no preservatives of all flu shots. Even so, the association between thiomersal and autism has never been conclusively proven.
  • Products of conception and cell lines derived from it: none of the flu vaccines contain, or are manufactured in cell lines based on, human tissue.
  • Foreign DNA: one of the more ridiculous arguments against vaccines is that they contain human or animal DNA because they are grown in cell lines derived from human or animal cells. Some have even claimed transgender individuals and gender dysphoria are caused by female DNA in childhood vaccinations, which is of course completely insane. Cell-grown flu vaccines (Flucelvax) are grown in MDCK cells, which is a cell line derived from renal tubules of a cocker spaniel. Before you worry about starting to bark and growing big, sappy eyes after your flu vaccine, let me reassure you: not only are no cells or DNA in the vaccine, even if they were, they cannot simply start rewriting your DNA. If that were a case, eating meat or even plants would risk ferrying foreign DNA into your cells. Fortunately, that’s not how cells work, and a vaccine will not rewrite your DNA (if it would, biotechnology would be so much simpler!).

But if vaccines work, why should I get vaccinated?

This argument is one that perhaps deserves to be considered separately. “Surely, if vaccines work,” so anti-vaxxers argue, “you will be protected whether I get vaccinated or not. What business of yours, then, is it whether I get vaccinated?” Wrong.

One, this assumes vaccines work perfectly. They don’t – nothing does. A person may for whatever reason fail to mount the sufficient immune response when vaccinated, lack the immune capacity without knowing about it or the vaccine batch may have been damaged (very rarely, this happens) and has become ineffective.

Two, it assumes everyone can be vaccinated. Some people, for instance people on short- or long-term immunosuppression, cannot effectively be immunised with vaccines as their immune system may be too suppressed to create a strong enough response. The consequence is that the weakest among us are exposed to pathogens without the chance of protection. Their only chance to avoid something that in their case may be much more severe than a mild cold is by relying on herd immunity. The same goes for children under the age of 6 months, who lack the immune maturity for vaccines to take. For the flu, with an R0 of about 1.5, we would need at least a third of the population vaccinated for herd immunity to be effective, assuming a 100% effective vaccine – knowing that flu vaccines’ VE is less than that, we’re looking at closer to 60% for adequate herd immunity. That means you, too. It is vitally important that everybody does their part, even if they themselves don’t expect, or care about, getting the flu.


In the end, what is a mere six months to us is a long, long time for a virus. Most viruses exploit this by trying to outpace our OODA loops, posing a double challenge: not only do we have to predict successfully what the major circulating flu viruses will be, but do such predictions with the risk of a new strain emerging that the CVVs based on original strain predictions will not protect against.

None of this is a good reason not to have the flu vaccine. The flu vaccine is incredibly safe (people with anaphylactic egg allergies can get the recombinant vaccine), the risk of side effects is very low and even a reduction of 30% in flu-like illness can mean a great deal. And to those for whom herd immunity is the only way to avoid serious illness, it can mean everything.

References   [ + ]

1. Often enough, this happens by extracting the genes responsible for the neuraminidase and haemagglutinin type, removing the part – called the polybasic cleavage site – that is responsible for destroying chicken eggs – and combining the genes with the genetic material of a human flu virus that is known to grow well in eggs. The resultant DNA is introduced into Vero cells (monkey kidney cell line) and cultured. The resultant viral material is then injected into eggs for culturing.
2. From Wenham and Edmunds (2015), How effective is this year’s flu vaccine? In: thebmjopinion.

3 thoughts on “Is the flu vaccine really only 10% effective? A sojourn into immunization epidemiology

Join the conversation