Revisiting Risk Judgments in the COVID-19 Pandemic

In the words of a Danish proverb (often attributed to Yogi Berra), "it's hard to make predictions, especially about the future." I felt that way about a blog post that I wrote last year where I argued that COVID-19 infections were coming in waves that corresponded to people's loosening and tightening of public health precautions. By the end of the year my assessment seemed tragically short-sighted, as COVID-19 infections in Colorado surged to a pinnacle that made all previous peaks and valleys look like nothing more than chance variations. 

OK, I thought, I was probably wrong: the behavioral battle against SARS-CoV-2 was lost, and only vaccines would be able to bring us back from the precipice. By March or April of this year that's still how things looked, and numbers were finally on the decline. But in May another surge showed up, and by August the Delta variant had fueled another. Here is Colorado's graph:

The new cycle, after vaccinations blunted the very high numbers last winter, actually looks quite a bit like the old cycle - albeit with peaks more than twice as high as the previous ones. Patterns of oscillation in COVID-19 case numbers have been noted by  researchers and in the media. 

You could look at this graph and tell a story that's purely about viral spread and immunity: in phase 1 people mostly stayed home which kept risk low; in phase 2 people went back to normal activities last fall and suffered the consequences; and in phase 3 vaccines held down the rate of spread until a new variant with increased transmission potential and some level of vaccine resistance caused the fall 2021 surge. That explanation doesn't explain the smaller peaks along the way, but it does lead to a strong prediction: As people go back to work and school this fall, Delta will spread unchecked and we will see a continued surge just like we did in fall 2020. This time the people affected will include those who haven't received a vaccine or those like children under 12 who are not eligible for a vaccine. Some early data from Children's Hospital Colorado suggest that case rates are indeed increasing particularly fast among children. This interpretation would again argue for biomedical prevention: Give everyone another dose of vaccine, and hurry up with approval for the pediatric vaccine as well.

But behavior is also part of the equation, and in fact a recent article argued that the interaction between biology and behavior is what has made the COVID-19 pandemic so difficult to predict or control -- in other words, my original point from June 2020. In my previous post I wrote about how people mental models for risk tend to be all-or-nothing: If a risk is below about 10 percent our Intuitive minds classify it as "not possible," and if it's over 90% we treat it as though it were "certain." For all intermediate levels of risk, our Intuitive mental models tend to assume a 50/50 chance -- about the same odds as a coin flip. If the recent surge in cases pushes enough people's mental risk into the "over 90%" category, they are likely to change their behavior -- to limit trips to the store, to keep their distance from others, to cancel trips or skip eating out, to finally get a vaccine, or to become more religious about wearing masks in all circumstances. Of course, recent policy changes are intended to push people in those directions as well, but school mask mandates or Federal vaccination mandates for employers are just one ingredient in people's overall decision-making process. The more those behavioral changes take effect, the less likely we are to see a huge surge in cases like we did at this time last year.

Perceiving our COVID risk as 90% is, of course, a huge over-reaction, especially for those who are vaccinated. Your actual odds of getting COVID-19 look something like this:

• If you are exposed to someone who has the SARS-CoV-2 virus, there's now about a 1-in-8 chance (12.5%) that you will get the virus yourself. That obviously goes up if you spend a long breathing the same air, if they coughed right in front of you, etc., and down if you just happen to pass them in the hall. (You can calculate the risk level under different assumptions here: https://indoor-covid-safety.herokuapp.com/apps/advanced) But this base rate, due to the Delta variant, is up substantially from last year's 1-in-30 rate (about 3%), which mostly occurred through large-group "superspreader" events. If you are outdoors when exposed to someone with SARS-CoV-2, incidentally, your risk is probably 1% or less even with the Delta variant, especially if there's a breeze blowing. 
• The larger the group the more likely it is that someone in the room will have COVID, so your risk goes up considerably in large events, although still not as much as you might think because this depends on community prevalence rates. If 95% of people in your community are virus-free at any given time, and there are 10 people in a room, the chance that all of them will be healthy is 95% to the 10th power, or still about 60%. Still, avoiding large groups remains a good risk-reduction strategy, especially in areas with higher prevalence rates. And it's particularly helpful if you avoid settings that are known to create higher risk for COVID transmission, like restaurants, gyms, and indoor church services. Some of the increased risk associated with these settings is probably due to the size of the venue and to specific activities that might increase the chance of breathing in viral particles, like eating, singing, or working out.
• If you are vaccinated, your chance of contracting COVID is reduced by more than half -- original estimates were over 90% efficacy for preventing symptomatic infection, but even with the Delta variant and potentially declining vaccine efficacy over time we can conservatively say that you are 60-70% less likely to be infected in the first place. This brings your total risk down to 5% or less and is still the single most effective step you can take other than social distancing.
• If you wear a mask, your risk is reduced by another 40%; and if both you and the person you are talking with are wearing masks, your risk goes down by 80%. Masks are the second-most effective prevention method after vaccination, which is the reason CDC recently recommended a return to universal masking. Cloth masks work like filters to block larger particles and filter out some of the smaller aerosols; they also provide protection through "source control" by blocking the exit of SARS-CoV-2 from an infected person's nose or mouth. Building on the previous steps -- assuming you both are vaccinated and also are in a 100% masked environment -- your risk of acquiring SARS-CoV-2 is now down to 1% or less.
• COVID testing is being used as a backstop in many environments, like air travel or workplaces where not everyone is vaccinated. Rapid tests are good at detecting SARS-CoV-2 infection even in asymptomatic individuals (as opposed to symptom screening alone, which is only about 50% accurate). The reason that testing has not been highly effective so far as a prevention strategy, though, is that we are generally not testing everyone in a population on a regular basis. Current strategies try to balance this by testing only the highest-risk individuals in the population -- e.g., those who have not yet been vaccinated. In combination with other strategies, it's likely to help reduce the risk that you will ever be in the same room with a person who has SARS-CoV-2.
• You can also slightly reduce your risk with additional steps like regular hand-washing, cleaning surfaces, keeping more physical distance between you and other people, using an air purifier, being in a room with a high-quality HEPA air filter, or opening the windows to increase airflow. These are all additional helpful steps although nowhere near as beneficial as the top three strategies of vaccination, avoiding large groups, and universal masking. All of the steps taken together might further reduce your risk to half a percentage point or less. A "layered strategy" for prevention is much more effective than any single step you can take.

So, assuming that you are a person who has taken all possible steps to prevent COVID-19 infection, you are actually still pretty safe this fall even given the higher transmissibility of the Delta variant and the potential drop in vaccine efficacy against this new strain of SARS-CoV-2. I say this as someone who unfortunately has had 4 fully vaccinated family members get breakthrough infections in the past month -- it's certainly not inevitable that anyone is going to get the virus. Even if you do get infected the vaccine provides an additional level of protection: Your infection is less likely to result in hospitalization or death, and there is likely a shorter window of time during which you can transmit the virus to family members or others. The jury is still out on "long COVID" symptoms, but some results suggest a lower level of risk or a lower severity of ongoing symptoms for people who've had a vaccine. If you can't be vaccinated for whatever reason, your risk for both COVID infection and serious illness does increase, and the need for behavioral risk reduction is much stronger. In particular that means being very conscientious about masking and only being around other masked people, but also limiting your overall interpersonal interactions to only those that are truly essential. 

One of the most challenging problems associated with COVID-19 is that risk behaviors tend to be correlated with one another. If someone who is unvaccinated also avoids large gatherings and wears a mask, that helps. If someone who doesn't want to wear a mask does get vaccinated, that helps. Regular testing for unvaccinated people may help to reduce overall community risk. But for many people, COVID-19 prevention is an all-or-nothing exercise: They want neither masks, nor vaccines, nor testing, because their mental model says the risk is close to zero. (Or in some cases, perhaps their mental model says the risk is close to 100% and they have become fatalistic). Other people who are concerned about COVID-19 may take all of the preventive steps, pushing their actual risk close to zero, but they remain convinced that their risk is very high. The irony of behavioral prevention is that when personal risk perception is high, some of us will withdraw from social contact to a greater degree than may be warranted; and when perceived risk is low, others will take no precautions at all. And worst of all, our Intuitive mental apparatus for perceiving risk tends to push us in one of these two directions.

More than a year after my original post on risk perception, the interaction between people's risk perceptions and the pace of the COVID-19 pandemic remains fraught with peril. A realistic view suggests that most people's personal risk is actually rather low. But such a mathematical calculus might backfire if people equate low risk with no risk according to our usual bad-at-statistics mentality. In that case, people will fail to take adequate protective steps and end up with higher levels of risk that fuel the next wave of infection. A more mathematically informed decision-making strategy (courtesy of the Narrative mind) might lead us to take sensible precautions in order to reduce our actual risk level, but not to completely shut down society in the meantime. If we do this, perhaps we will be able to limit the current wave and not have to see the same cycle of up-and-down COVID numbers again by the time we reach fall 2022.