No, the CDC did not just apologize and admit that this year’s flu vaccine doesn’t work, part 2


It looks like this year’s going to be a bad flu season.

Hard as it is for me to believe, it was only five weeks ago when I discussed an announcement by the CDC that this year’s flu vaccine would likely be less effective because it isn’t a good match for the influenza strains in circulation this year. Those familiar with how the flu vaccine is developed every year know that the composition of the vaccine depends on the WHO’s choice of the three or four strains that its experts deem most likely to cause significant human suffering and death in the coming flu season. Basically, the WHO has to make an educated guess many months in advance about what strains will be circulating during next year’s flu season. When it guesses well, the flu vaccine is pretty effective. When it guesses poorly, not so much. Every year, even with the most careful deliberations going into making the choice of strains of flu virus to include in the vaccine, it’s always a bit of a crapshoot.

The reason is antigenic drift. This year, it just so happens that a strain of H3N2 is the predominant strain circulating in the population. Unfortunately, H3N2 has undergone drift, such that the strains in circulation now are different than the H3N2 strain chosen to be part of the flu vaccine and thus a poor match. The current H3N2 strains didn’t make an appearance until it was too late to alter the formulation of the flu vaccine, and they had increased greatly by September. Basically, only 48% of the circulating H3N2 strains causing most of the disease are a good match.

Unfortunately, yesterday we learned that this year it’s even worse than expected. When I said that the CDC didn’t just admit that this year’s flu vaccine doesn’t work, it was true. However, there’s no denying that this year’s flu vaccine has been quite disappointing as measured by a the CDC’s most recent study of vaccine effectiveness (VE), which the CDC has been carrying out each year since 2004. According to a press release from the CDC, year’s VE study revealed:

A report published in the January 16 Morbidity and Mortality Weekly Report (MMWR) estimates that getting a flu vaccine this season reduced a person’s risk of having to go to the doctor because of flu by 23 percent among people of all ages.

Since CDC began conducting annual flu vaccine effectiveness (VE) studies in 2004-2005, overall estimates for each season have ranged from 10 percent to 60 percent effectiveness in preventing medical visits associated with seasonal influenza illness. The MMWR report says this season’s vaccine offers reduced protection and this underscores the need for additional prevention and treatment efforts this season, including the appropriate use of influenza antiviral medications for treatment.

“Physicians should be aware that all hospitalized patients and all outpatients at high risk for serious complications should be treated as soon as possible with one of three available influenza antiviral medications if influenza is suspected, regardless of a patient’s vaccination status and without waiting for confirmatory testing,” says Joe Bresee, branch chief in CDC’s Influenza Division. “Health care providers should advise patients at high risk to call promptly if they get symptoms of influenza.”

Note that, in this case, effectiveness means “real world” effectiveness, as compared to efficacy, which refers to how well a treatment does in a randomized clinical trial. Naturally, this news has produced headlines such as:

Well, you get the idea.

So how did the CDC come up with this estimate? This MMWR study tells the tale.:

Patients aged ≥6 months were enrolled when seeking outpatient medical care for an ARI with cough at study sites in Michigan, Pennsylvania, Texas, Washington, and Wisconsin.* Study enrollment began once laboratory-confirmed cases of influenza were identified through local surveillance. Trained study staff members reviewed appointment schedules and chief complaints to identify patients with ARI. Patients were eligible for enrollment if they 1) were aged ≥6 months on September 1, 2014, and thus eligible for vaccination; 2) reported an ARI with cough and onset ≤7 days earlier; and 3) had not yet been treated with influenza antiviral medication (e.g., oseltamivir) during this illness. Consenting participants completed an enrollment interview. Nasal and oropharyngeal swabs were collected from each patient and placed together in a single cryovial with viral transport medium. Only nasal swabs were collected for patients aged <2 years. Specimens were tested at Flu VE laboratories using CDC’s real-time reverse transcription–polymerase chain reaction (rRT-PCR) protocol for detection and identification of influenza viruses.

Participants were considered vaccinated if they received ≥1 dose of any seasonal influenza vaccine ≥14 days before illness onset, according to medical records and registries (at the Wisconsin site) or medical records and self-report (at the Michigan, Pennsylvania, Texas, and Washington sites). Vaccine effectiveness was estimated as 100% x (1 – odds ratio [ratio of odds of being vaccinated among outpatients with influenza-positive test results to the odds of being vaccinated among outpatients with influenza-negative test results]); odds ratios were estimated using logistic regression. Estimates were adjusted for study site, age, sex, race/ethnicity, self-rated health, and days from illness onset to enrollment. These early interim VE estimates for the 2014–15 season were based on patients enrolled through January 2, 2015.

To boil it down, the CDC enrolled patients who sought care for an acute respiratory illness with cough and determined what percentage of those who tested positive for flu and those who did not had been vaccinated. In essence, this is a prospective test-negative case-control design. Basically, controls were patients with a medically attended ARI who tested negative for the flu virus, while cases were patients who tested positive. The primary exposure of interest was, of course, the flu vaccine. The idea of this study is to determine whether a higher proportion of people who tested negative for the flu were vaccinated than those who tested positive.

One thing that should be pointed out. This study design does not answer the question that most people are interested in: If I get the flu vaccine, what by how much does it reduce my chances of getting the flu? Remember, this patient population consists of people who have an ARI with cough severe enough that they sought medical help for it. Of course, these people almost certainly don’t have the flu, given that by definition they have no symptoms. On the other hand this study says nothing about people with the flu or ARIs who don’t seek medical care, nor does it say anything about the ability of this vaccine to prevent severe disease, hospitalization, and death. Still, as a quick measure of effectiveness after the flu season is only halfway over, it’s not bad. Think of it this way. To determine the absolute efficacy of the flu vaccine would require following a huge number of people, vaccinated and unvaccinated, given that in an average flu season around 2% of the population will get the flu, and seeing who does and doesn’t get the flu that year. To reliably detect, say, a 50% reduction in flu would require thousands of people. Under the CDC’s study design, of people showing up at a clinic during flu season, many more will actually have the flu; so a determination of flu vaccine effectiveness can be made with a lot fewer people.

In this case, there were 2,321 children and adults with AR enrolled in the various study sites, and 950 (41%) tested positive for the influenza virus by PCR. Among patients who tested positive for the flu, 49% had been vaccinated, while 56% of who tested negative were vaccinated. After the CDC adjusted for study site, age, sex, race/ethnicity, self-rated health, and days from illness onset to enrollment, the VE against medically attended ARI attributable to influenza A and B virus infections was 23% (CI = 8%–36%). Depending on age range, effectiveness ranged from 12% to 26%:

Another factor that influences how well the flu vaccine works is the age and health of the person being vaccinated. In general, the flu vaccine works best in young, healthy people and is less effective in people 65 and older. This pattern is reflected in the current season early estimates for VE against H3N2 viruses. VE against H3N2 viruses was highest — 26 percent — for children age 6 months through 17 years. While not statistically significant, VE estimates against H3N2 viruses for other age groups were 12 percent for ages 18 to 49 years and 14 percent for people age 50 years and older.

CDC recommends that people get a flu vaccine even during season’s when drifted viruses are circulating because vaccination can still prevent some infections and can reduce severe disease that can lead to hospitalization and death. Also, the flu vaccine is designed to protect against three or four influenza viruses and some of these other viruses may circulate later in the season.

The bottom line, obviously, is that so far this season the flu vaccine isn’t the greatest. Still, it offers some protection and, given how safe the vaccine is, there’s no compelling reason not to get it.

Clearly, what is needed is a much better flu vaccine. No one has ever denied that. However, the fact that some years the vaccine is not very effective because the strains used to make the vaccine don’t match the circulating strains very well because the WHO guessed wrong last February doesn’t mean that vaccines don’t work. All it means is that this vaccine is not s good as many other vaccines because of the inherent difficulties involved in making an effective flu vaccine that are a consequence of its biology, specifically its ability to mutate that results in drift, such as what occurred this year.

It’s only about a month until the WHO has to pick the strains to include in next year’s flu vaccine. Certainly, the committee tasked with choosing the strains will include the drifted H3N2 strain, but who knows if that will be the dominant strain next year. One thing’s for sure: I don’t envy the people who have to come up with these recommendations. In years when the vaccine is good, there’s no glory and the antivaccine movement attacks their vaccine, but in years when it’s bad they can expect nothing but opprobrium directed at the vaccine without even the personal satisfaction of having produced a highly effective vaccine.

Certain viruses are just incredibly difficult to design effective vaccines against. Unfortunately, influenza is one of them. Just because its efficacy varies from year to year and this year appears to be significantly lower than average is not evidence that vaccines are harmful or don’t work. It’s just that this vaccine isn’t the greatest, something that’s always been acknowledged. Next year, it’s likely to be much better. Until a vaccine is developed that targets a relatively invariant part of the virus and is thus universal, this is likely to be the case. Fortunately, scientists appear to be getting closer and closer to a universal flu vaccine.