Decoding COVID-19 Risk: What’s Wrong With Our Perceptions?
Cassie Guo, Ph.D, Senior Data Scientist — Epidemiology
Abhineet Gupta, Ph.D, Director of Resilience Research
Back in March 2020, it was unclear as to whether people in the US should wear face masks in public. With little knowledge of infectious diseases, people were left with confusing and contradictory information, particularly when health experts and officials in the US made earlier assertions that mask wearing was not recommended for most people. Against this view, officials in several countries in Asia made face masks mandatory in public places. By the time New York state mandated their mask wearing policy on public transport on April 15, the number of daily new cases in the US exceeded 30,000.
What influences our perception of pandemic risk?
While there are many speculated reasons for differences in initial response to the pandemic — ranging from political to cultural to ideological — one particular reason is indisputable. At the psychological level, it is difficult for people to relate their own experience to a disaster that they have never experienced, and even more difficult to assess how frequently it will happen when they don’t have a mental image for it.
The COVID-19 pandemic is not the first pandemic in American history, nor is it the most deadly in history, but it is the first pandemic of such great magnitude in the modern era. In the US, the most recent pandemic that people may recall is the 2009 H1N1 influenza pandemic. Though there was significant media coverage of the outbreak, with less than 5,000 confirmed deaths, H1N1 did not elicit the type of response and reaction as the current pandemic. By contrast, several countries in Asia have had a more severe and extensive experience with pandemics, including SARS and MERS in recent years. This difference in exposure shaped individual experience and memory for these kinds of events, thus influencing not only government preparedness, but also people’s perception and response amidst the current pandemic.
In Dan Kahneman’s Thinking Fast and Slow, he explains the concept of the availability heuristic: people estimate the frequency of events based on “the ease with which instances come to mind.” As a result, people think striking, emotionally resonant events are more likely to happen than unseen and abstract happenings. In the face of an unprecedented pandemic, people may instinctively search in their mental images for something that is readily available and come up short. What’s worse is that the effect of memory dims over time — and the last pandemic in the US was 11 years ago.
Kahneman goes into more detail on this effect in his book:
Victims and near victims are very concerned after disasters. After each significant earthquake, Californians are for a while diligent in purchasing insurance and adopting measures of protection and mitigation…However, the memories of the disasters dim over time, and so do worry and diligence.
How can we help people understand pandemic risk?
To improve perceptions of risk, we need to make these “unseen and abstract happenings” more concrete by visualizing tangible examples related to our own experiences. At the organizational level, this could mean conceptualizing risk in the context of a particular workforce and facility. We incorporate this framework in the COVID Calculator, which is a tool designed to help leaders visualize an outbreak of COVID-19 in their workplace.
With COVID Calculator, decision makers can easily envision scenarios specific to their own workplace by generating a 10-day forecast of total new cases in their facilities, given the event of an initial infected case. To tailor this estimate to the specifics of a particular workplace, the model uses their facility parameters, including the number of employees, office layout, physical address and COVID protective measures. For example, imagine that the Chief Risk Officer of a company in Menlo Park, CA has to make a plan for reopening an office of 500 employees while allowing some employees to work from home. They can input their workplace information directly into the tool to generate a forecast that will help inform these decisions.
With the awareness of risks specific to their own inputs, decision makers need to take actions for risk mitigation and they need to understand how different actions can increase or decrease risk. By testing and building different scenarios based on different COVID protective measures implemented in their facilities (such as mask wearing and social distancing), decision makers are able to see how their decisions may impact risk in the workplace. For business leaders, this type of comparison and visualization on simulated data helps to contextualize the abstract benefits of safety measures, and can give them a better understanding of how these measures directly impact the risk of COVID spread at their workplace.
Additionally, this tool has incorporated the current county level case rate, which impacts the potential number of newly-introduced infected employees during the course of 10 days. Unlike many other tools which forecast on the country or the state level, COVID Calculator gives a localized assessment that shows how the pandemic is spread within communities and facilities, which may make it more easily relatable to what people are actually observing in their communities.
Now let’s shift back to the decision on mask wearing. How do we help people understand the impact of mask wearing on reducing pandemic risk? While it is difficult to make decisions on conflicting information, the uniqueness of pandemic risk makes it increasingly important to realize that individual risk contributes to the risk of the community; hence, the key to fighting an infectious plague is to emphasize individual responsibility.
To hold individuals accountable, we need to understand the nature of disease transmission.
Long before the current pandemic, in 2005 Lloyd-Smith et al investigated the contact tracing data from eight different infectious diseases, including SARS and influenza. They observed that the distribution of secondary cases is heavily skewed towards the right tail. These results demonstrate that individual infectiousness is highly variable, and only a small fraction of individuals contribute to the majority of cases. In other words: it only takes a few superspreaders to sufficiently cause a large epidemic.
“These results suggest the 20/80 “rule” (whereby 20% of individuals are responsible for 80% of transmission) — which was previously thought to apply to STIs — may also apply to directly transmitted infectious diseases.” — Modeling Infectious Diseases in Humans and Animals, Keeling & Rohani
This heterogeneity and uncertainty in disease transmission can make it increasingly difficult to properly conceptualize risk. We wanted to help people understand these nuances, which is why they are reflected in the outputs of COVID Calculator. Let’s consider an open floor plan office with 100 employees with the lowest possible case rate as an example.
As shown in the figure above, when less than 50% of people are wearing masks in the office and there is one infected employee on day 1, for an average scenario, we expect between 2 to 5 cases after 10 days. However, under severe scenarios, as many as 15 or more people could be infected after 10 days, meaning a small portion of the people spreading COVID could result in large consequences. Moreover, it is challenging to predict who these highly-infectious people might be ahead of time. But understanding these nuances is important: we can’t effectively achieve high individual compliance if we fail to understand the underlying nature of disease transmission.
“The epidemiological implications of these observations are interesting… this documented heterogeneity implies an increased disease extinction risk, …though outbreaks are more severe when they do occur.” — Modeling Infectious Diseases in Humans and Animals, Keeling & Rohani
We cannot easily visualize that the consequence of one infected individual being careless or irresponsible results in increased risk for the entire population. Nor can we identify superspreaders beforehand, who can cause large outbreaks. But our individual difficulty to understand and visualize these risks does not change reality: pandemic risk is not limited to one’s physical health, and extends to affect others’ lives, even to the extent of causing a global pandemic.
To form the right perceptions about pandemic risk, one needs to understand that pandemic risk is different from the risk of being in an earthquake, or a car accident — for most people, the cost of the risk-seeking behavior in a pandemic is unseen. Unlike a car accident where the risk is limited to those involved, pandemic risk slowly expands to encompass more and more people.
With COVID Calculator, our goal is to help communicate the complexities of pandemic risk in a way that is visual, tangible, and relevant to one’s own experience. It is our hope that this tool can serve as an effective medium to help leaders understand risk and to help companies build resilience in the face of the disruption.
