Relative risk is more informative than effectiveness.

The past few weeks have sent a tremendous amount of disappointing news about the Delta variant and the possible waning effectiveness of vaccines. An outbreak among partygoers in Provincetown found a high number of infected individuals who had already been fully vaccinated. News stories and anecdotes about “breakthrough infections” abound. And the CDC has confusingly made an about face on its recommendations about face coverings for the vaccinated. This has fed a growing anxiety that the vaccines have not delivered what was promised.

The Pfizer and Moderna vaccine, heralded as “95% effective”, led many to hope that they could eradicate this coronavirus. However, and unfortunately, from the beginning, there has been a lack of clarity among not only the media but among health professionals themselves around what “effectiveness” means. Indeed, misunderstandings of vaccine effectiveness persist in media reporting on the pandemic.

Pfizer ran a randomized control trial where they took a diverse pool of sixty thousand people and randomly gave half of the people the vaccine and half a placebo injection with no effect at all. Effectiveness measures the percent reduction of infection in the vaccinated group when compared against the control group. A vaccine effectiveness of 95% means that Pfizer observed 95% fewer infections amongst the vaccinated than the unvaccinated.

An identical way of stating the result of the trial is that 20 times as many people who received the placebo developed symptomatic COVID-19 as those that received the vaccine. In the language of risk, we would say that the vaccinated group had 20 times less risk of symptomatic COVID-19 infection than the control group.

While “95% effective” is numerically the same as “20-fold risk reduction,” it leads to more confusion. It is far too common for “95% effective” to be incorrectly interpreted as a 5% chance of getting infected at all, which is not the case. If the message had been that vaccines reduce risk by a factor of 20, then we would be equipped to both better understand the power of these vaccines and better plan for moving into an open, but increasingly vaccinated world.

20-fold risk reduction allows one to consider their everyday experiences and make judgement calls. Hanging out with friends outside was already very low risk. If everyone is vaccinated, the risk after vaccination becomes effectively zero. On the other hand, going to a packed, sweaty dance club and sharing drinks with strangers was very high risk before vaccination, so vaccination makes it 20 times less risky but still risky.

The Delta variant itself is more infectious than previous strains and can partially evade immunity. This further reduces vaccine effectiveness. But, like with the clubbing example, behavioral changes also change effectiveness. The virus is not the same today as it was in the Fall of 2020, but we are also not the same people.

Clinical studies are done in the best of circumstances, and most medical professionals expect the effectiveness to decrease in the general population. Scientists do their best to ensure that the population of individuals in a study are representative of all of the people in the world. They perform complex matching and outreach to ensure characteristic diversity in the pool of subjects. But one of the hardest factors to control for is the psychological and behavioral changes in the broader population over time.

In particular, people are taking more risks now than they did in 2020, and the risk of infection has increased for all, vaccinated or not. Just imagine you get vaccinated and then increase your risk by a factor of four. For example, you stop staying at home on Zoom all day, and go back to the office, bars, and restaurants. You start meeting with all sorts of people you don’t know very well. Then your risk of infection is now a fifth, not a twentieth of what it was. And further imagine that a new variant comes along that is twice as infectious as the old one. Now your risk reduction is down to 2.5, which would amount to a total “effectiveness”—combining the vaccine, your behavior, and the variant—of only 60%. But 60% is still better than most seasonal flu shots.

Perhaps this is the hardest part about where we are in the pandemic. Everyone wants to return to normal and never think about coronaviruses again. But a preponderance of evidence indicates that SARS-CoV-2 will become endemic and will circulate like other common viruses. It is likely that all of us will become immune to COVID-19 in one way or another, though, because of the vaccines, the disease will cause much less death and suffering than it did before. But even with such dramatic reductions in mortality, any COVID-19-associated deaths in the United States feel like too many for a society that has been terrorized and torn apart by the pandemic.

And given that children do not yet have a vaccine available, parents worry that their children remain at risk. But we accept much more deadly risks in our lives. It is uncomfortable to accept that a 3 year old has a similar risk of severe COVID-19 as a vaccinated 40 year old. Doing a cost benefit analysis about your child is emotionally impossible. I understand how any risk, no matter how small, can feel intolerable. Unfortunately, that same 3 year old is more vulnerable to death or serious injury by driving, swimming, or even eating. The risks entailed with COVID-19 do not seem to be much different than those of just growing up. Since we want nothing more than for our kids to be safe, we delude ourselves by never conceptualizing the risks of their ordinary activities. We live our lives as though these risks are zero. Though it seems impossible to imagine this now, in time, we’ll accept the remaining risk from COVID-19 as well.

Many thanks to Sarah Dean, Jordan Ellenberg, Eric Jonas, Lauren Kroiz, Deb Raji, Lawrence Recht, Chris Re, and Isaac Sparks for reading drafts of this post and offering insightful comments and suggestions.