Why I Built a Decision Support Tool for Covid-19 Risk
We need better ways to calculate the cost of Covid in everyday activities.
Hi all,
Greetings from Santa Monica!
I spent the weekend building something cool: A data-driven decision support tool that weighs coronavirus risks against the benefits of certain activities. The idea is to help people assess the cost-benefit trade-off of regular activities (going to a restaurant, getting on a plane).
Call it the cost of Covid.
The calculation varies for each individual and situation:
For a 40-year old living in Los Angeles, riding a subway or a bus has a Covid cost of $38.
For a 25-year old living in Montana, going to work in a shared office has a Covid cost of less than $1.
How to Think About Risk
A bit of background. I was a Wall Street quant in a past life. I set a record at Goldman Sachs for most nested IF statements in one spreadsheet (insert quant joke).
For the past few months I’ve advised hospitals and healthcare systems on Covid-19. I’ve seen how healthcare leaders are using data to drive decisions on testing, safety protocols and when to reopen.
As states continue to open up, many of us are asking ourselves: What is the safe and responsible thing for me to do?
Responsible people take risks every day. And we take steps to mitigate those risks. We drive, but wear seat belts; we bike to work, but wear helmets.
The risk-reward calculus for some activities is different now.
I wondered: How can we use economic risk modeling tools to guide our decisions on which activities are worth the risk?
Activity Risk Scores
There are dozens of decisions to make each day. Should I get a haircut? Have a picnic? Fly?
Are you going to risk serious illness for a trip to the movie theater?
We know the basic risk factors: Outdoors is safer than indoors. Being at least 6 feet away helps. Masks are safer than no masks.
CDC last week released guidance for many activities.
News sites have published risk scores for various activities based on epidemiologist estimates (see The New York Times and MLive). These risk scores should be taken as general guides, not a precise scoring system. They don’t necessarily represent the consensus of medical professionals.
I wanted to take the analysis a step further. Every person and situation is different. For some people, getting on a plane is no problem; for others, going to eat at an outdoor restaurant isn’t worth the risk.
The solution I created was inspired by Dr. Bob Wachter, Chair of UCSF Department of Medicine (and one of the most active healthcare leaders on Twitter). According to his back-of-the-envelope estimate, the risk of catching Covid-19 from one cross-country flight is 1 in 750.
I followed Dr. Wachter’s methodology in building my model. I used the New York Times epidemiologist poll data to create risk scores, with each activity rated on a scale of 1 (low-risk) to 5 (high-risk).
The model takes into account both economic costs (lost income, medical expenses) and non-economic costs (how much would you pay to avoid being sick for one day?)
I listened to a podcast with Balaji Srinivasan, a biotech executive and investor. He says there’s now a health-related cost or health tax associated with everyday activities, based on the risk of infection.
Balaji posed the question: Even if getting the virus is not as bad as dying, is it worth getting coffee at your local cafe for a chance you'll get a virus that knocks you out for 10 weeks?
Let's say the virus does $10k of economic damage to you. You're making $50k a year, and the virus puts you out of work for 10 weeks (for a severe case). If there’s a 1% chance you’ll get infected from going to the café, and a 10% chance you’ll develop a severe illness, that activity has a Covid cost of $10k * 1% * 10% = $10.
Calculating the Cost of Illness
For many of us, the pain and misery of being sick far outweighs any lost income.
How do you quantify the cost of being sick – pain and suffering, missing out on future events, potentially losing years of lifespan?
Some say you can’t put a value on a human life. Turns out, our government leaders have been doing it for years to evaluate consumer safety and environmental programs. US federal agencies estimate the value of a statistical life (VSL) at $7-9 million. This works out to $300 of value per day for a person living to be 70 years of age.
So a disease that makes you seriously ill for 10 weeks might cause 10 * 7 * $300 = $21,000 of damage.
I can already hear people saying: How can you apply dollars to an illness that can mean life-or-death? My response: We need a good yardstick to help us make difficult decisions. Time and money are the best measuring sticks we’ve got.
Shutting ourselves inside all day is not a realistic option for most of us. For those who are willing to re-enter public spaces, we’ll need to find a way to weigh coronavirus risks against the benefits of certain activities - including emotional well-being.
Using Data for Better Decisions
This model is designed to help people assess the health economic cost of everyday activities. I built the model so anyone can enter their own estimates. You can adjust the model for geographic region, individual risk level, activity type, income and cost of a sick day.
I hope this model helps you think about risks & make reasonable choices. And let's do all we can to lower the risks.
Important Disclaimers
This model should not be used to justify risky behavior. It does not take into account the knock-on effects of actions that may put your friends and neighbors at risk.
The model is a marginal analysis at the level of each individual, but from what we know about disease, the collective costs are greater than the costs that individuals bear on the margin. If everyone made decisions based on this analysis, there is a risk that they would all be worse off than expected because they could drive up the spread.
The model is directional at best. It’s not meant to give you any definitive yes or no answers as to whether you should engage in any activity.
The model is a work in progress. The assumptions are not perfect. I started building it on Saturday and have worked on it nonstop through Monday afternoon. I’ve gotten a ton of helpful feedback already. I’m sure many of you will have feedback, too - I welcome any and all comments.
My hope is that this tool will help launch a conversation. I hope it will stimulate more rational, data-driven decision making when it comes to the coronavirus.
You can view the model here. Feel free to play around with it. Share it with others who may be interested. And please let me know what you think!
Warmest wishes and much love,
Daniel Zahler —> zahler@gmail.com
Enjoy this?
There are a few things to do:
Hit the “like” button (so I know to do more posts like this)
Follow me on Twitter.
Become a subscriber!