Let’s start with a simple explanation of what it means that this year’s flu shot is “23% effective”. What happens is that the CDC monitors people coming to doctors’ offices with acute respiratory infections and checks them for influenza by laboratory testing. It then checks the proportions who were vaccinated or unvaccinated. Then, they calculate effectiveness as 100% x (1 – (odds of being vaccinated among those with influenza /odds of being vaccinated among those without influenza )). This is what they found:
Of the 2,321 children and adults with ARI [acute respiratory infections] enrolled at the five study sites, 950 (41%) tested positive for influenza virus by rRT-PCR; 916 (96%) of these viruses were influenza A, and 35 (4%) were influenza B (Table 1). The proportion of patients with influenza differed by study site, age, race/ethnicity, and interval from onset to enrollment (Table 1). The proportion vaccinated ranged from 46% to 66% across sites and also differed by age, sex, race/ethnicity, and self-rated health status.
The proportion vaccinated with 2014–15 seasonal influenza vaccine was 49% among patients with influenza compared with 56% among influenza-negative controls (Table 2). After adjusting for study site, age, sex, race/ethnicity, self-rated health, and days from illness onset to enrollment, VE against medically attended ARI attributable to influenza A and B virus infections was 23% (CI = 8%–36%).
Of those with influenza, 48.9% were vaccinated. Of those without influenza, 56.2% were vaccinated. Converting those to odds gets you 0.957 for influenza positive people and 1.283 for influenza negative people. So effectiveness = 100 x (1 – (.957/1.283)) = about 25%. But that’s unadjusted. So the CDC did some adjusting, and it’s 23%.
But let’s remember the reality here. First of all, the difference in the percentages here were 49% of those with flu were vaccinated versus 56% of those without flu were vaccinated. That’s a bad year. In a good year – say 2013, the numbers were 32% and 56%. That yields an effectiveness of 62%. But even in a good year, one third of people who were influenza positive were vaccinated!
This isn’t a randomized controlled trial, and I’m not going to calculate an NNT. It’s also not a perfect methodology to establish how the influenza vaccine reduces the absolute probability of your getting the flu. But the differences between this year and others don’t seem to be as huge as many are saying they are. And it’s just a shot. It’s hard to see how the benefits don’t still massively outweigh the harms.