Skip to content

Instantly share code, notes, and snippets.

@mike529
Last active July 28, 2021 20:48
Show Gist options
  • Save mike529/0ce687824f70ef01e674f21392f3b061 to your computer and use it in GitHub Desktop.
Save mike529/0ce687824f70ef01e674f21392f3b061 to your computer and use it in GitHub Desktop.
COVID Risk and Safety

Introduction

Throughout the pandemic many people have been frustrated with public health guidance as either too strict, too loose, or simply incoherent. In this article I will argue that the public health guidance makes sense when viewed from the perspective or their actual goals, which are based on making people feel safe, even if it seems nonsensical when viewed from the perspective of risk mitigation.

Risk Mindset:

Many people, especially mathematically inclined ones, rely on a risk based framework for evaluating the proper COVID mitigation decisions. In this framework each mitigation has a cost associated with it (loss of quality of life, economic damages, etc.) as well as a corresponding benefit in terms of reduction in COVID deaths.

Mitigation Cost Benefit
Wash Hands Low Low
Social Distancing High High
Ban Walks in Parks Medium Low

Obviously we don't have precise numbers for either the costs or benefits of these actions. However, by making educated guesses based on the available data we can choose the best set of mitigations.

Safety Mindset:

Public health agencies often make decisions which seem hard to justify from within this risk framework. This is not because they are malevolent or incompetent. Instead it is because they are focused on a fundamentally different, and sometimes contradictory goal. Their chief objective is to allow people to feel "safe". If you follow the recommended actions, then you are "safe" and don't have to worry about catching COVID. Even more importantly, safety provides a framework for assigning blame for catching COVID. If you are "safe" then you don't need to worry about catching COVID and if you do catch it, it is not your fault.

Safety Requirements:

As part of their mission to support safety, public health agencies are significantly constrained in what sorts of recommendations they can make.

Universal Safety:

It must be possible for everybody to be safe. If somebody is not being safe then they are to blame for any bad consequences. If it is impossible for them to act safely then that is unacceptably unfair.

Binary Safety:

The main point of safety is to not need to worry about the underlying danger. Constantly making granular assessments as to whether you are being safe enough is in direct conflict with that goal. Therefore it is important that mitigation measures have clear and unambiguous dividing lines between safe and unsafe behavior.

This is also important because in the risk based approach there is a danger of constantly spiralling to ever greater precautions. How long should you wash your hands? How high quality a mask should you wear? How far away from other people should you stand? All of these questions explicitly deciding what an acceptable level of danger is, which is emotionally and mentally draining.

Risk Vs Safety - Driving:

The difference between these two approaches can be seen more easily with a non-COVID example. There is approximately 1 death per 100,000,000 miles driven. From a risk mitigation perspective there are several actions which can decrease these odds:

  • Wearing a seatbelt - ~50% reduction in deaths
  • Using an airbag in addition to the seatbelt - ~20% reduction in deaths.
  • Driving during the day rather than the night -~20% reduction in deaths.
  • Driving during non-rainy weather - ~20% reduction in deaths.
  • Driving a car rather than a motorcycle - ~2500% reduction in deaths.

From a safety perspective however, only the first two actions matter. If you are being a "safe" driver, then the risk of dying in a car crash can be ignored.

This example also illustrates why safety rather than risk is the focus of our public health agencies. Under ordinary circumstances thinking about the risks can be paralyzing. Somebody who needs to drive home from work late at night in the rain is not helped by thinking about the riskiness of an action they can't easily avoid.

Risk vs. Safety - Case Studies:

In this section we will examine several decisions which have been frustrating or baffling from a risk based standpoint, but make perfect sense when examined from the safety based perspective.

Masks:

At the start of March the Surgeon General famously tweeted that masks were not effective for the general public but that those hoarding them were preventing medical personnel from having a sufficient supply. Under significant pressure, this point of view changed and the dominant viewpoint became that masks were one of the most important mitigations that ordinary people could use.

Examined from the perspective of safety rather than risk, the change in messaging becomes much clearer.

At the start of the pandemic the standard for safety was for medical personnel wearing properly fitted surgical masks. Surgical masks were not available in sufficient quantities for universal masking and therefore they could not be necessary for COVID safety.

At some point the consensus was reached that loosely fit cloth masks were "effective" enough. Since they were widely available, it was possible for these masks to become part of the "safety" standard. Once this happened, cloth masks were "safe" and there was no need to think about the relative effectiveness of cloth masks vs surgical masks or N95s.

Vaccines - Early Distribution:

At the start of 2021 vaccines faced a similar problem. There were not enough vaccines for everybody, so they could not be necessary for safety, but it was important that people get the vaccines when possible. In order to square this circle, the messaging was that vaccines did not impact transmission, but only made you less likely to get sick. Because vaccines could not make you "safe", everybody still had to wear masks and social distance. Since masks and social distancing made you "safe" there was no urgency to try and jump the line to get your vaccine.

Vaccines - Natural Immunity:

Since the start of the pandemic it has been apparent that previous infection provided powerful protection against future infection. There were a small number of breakthrough cases detected, but as data from Israel confirmed, prior infection provided ~90% effectiveness against infection. This protectiveness is slightly worse than a full course of Pfizer of Moderna (~95% effectiveness), but better than a single dose of Johnson and Johnson (~75% effectiveness).

To judge from the public health messaging however the situation is completely different. Prior infection is treated as giving little to no protection (or at best transient protection) while any distinction between the effectiveness of Pfizer vs J&J is minimized or ignored.

Again this makes perfect sense when viewed from the perspective of safety rather than risk. Prior infection is not available to everybody (or at least we don't want to apply it to everybody), which means that it cannot provide safety. On the other hand a full course of an approved vaccine is "safe" which means that a single dose of J&J provides equal safety to two doses of Pfizer.

Vaccines - Universal Distribution:

In May, the government announced that vaccines would be available to anybody who wanted one. Immediately this cleared the last barrier for vaccines to become a necessary part of safety. At this point a schism emerged in the public health messaging. Vaccines were vastly more effective than any of the previous social distancing measures. Anybody who was vaccinated and following no social distancing precautions, was at objectively lower risk than somebody unvaccinated and taking all of the previously recommended precautions.

In light of this fact some experts recommended that vaccination be considered a necessary and sufficient condition for safety. Once vaccinated you were done with the pandemic and no longer had to worry about infection.

Other experts were unwilling to consider vaccinations sufficient, yes vaccinations reduced your chance of infection significantly. But the risk was still there, as such it wasn't safe to eliminate other precautions while COVID was still circulating.

It is important to note that this schism is not inherently scientific in nature. Everybody agreed on the basic data, but the question was on priorities and how to focus attention. Those who favored increased precaution might focus on the number of breakthrough infections and hospitalizations, while those who favored a return to normalcy would focus on the chances of dying once vaccinated (less chance than being struck by lightning!).

For a while it seemed like the first group was ascendant, cases were falling everywhere and cases among the vaccinated were very low. The arrival of the delta variant along with rising case levels (primarily among the unvaccinated) meant that the second group became ascendant which culminated in the newly announced mask recommendations.

Conclusion:

If I was writing this essay a few months ago, this is where I'd begin a tirade about how we need to focus on risk and stop wasting our efforts on chasing illusory safety. However, I now believe that a foucs on safety is inevitable. People do not respond to danger with risk analysis and attempting to craft public policy on these sorts of analyses is almost certainly doomed to failure.

Early in the pandemic the failure of the existing safety narrative gave early prominence to people who operated in a risk-based framework. They (myself included) urged swift decisive action, to be followed up by more careful examination of the evidence. In a world governed by risk this makes sense, the costs of mistakes are low, as they can be fixed quickly. In a world governed by safety, the costs of taking an early mistaken action are enormous. Would waiting longer for universal masking have been worth it, if that meant higher quality masks could have been used? Would it have been worth waiting on school closures until better estimates of learning loss could be gathered?

On the other hand it is important to work on integrating the risk perspective into a more safety conscious field. This is not as simple as doing a cost-benefit analysis and then simply implementing the recommendations. It needs to be interwoven into the fabric of the organization and appropriate channels need to be created to translate the risk based recommendations into the more safety minded language that these organizations speak.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment