COVID-19: How Many Lives Saved In April?

How do we re-ignite the economy?

View Your State Methodology

Analysis


Headline: Celebrate the fact that #StayHome has saved tens of thousands of lives.

It takes about two weeks for an intervention to have a measurable impact on slowing the spread of the coronavirus. This is because it take one to two weeks for an infection to turn into symptoms that are serious enough to seek medical treatment.

I am standing up this website to include all the data. I am not a web developer, but I wanted to better appreciate the work of web developers, and I'm always looking to learn new things. Thanks to an MIT Human Computer Interfaces, TMCC Beginning Web Development and some Stack Skills online courses, I am giving it the “college try.”

Method: This analysis examines the confirmed case growth rate prior to social distancing versus after. The orange line is based on miles traveled per day, and is included as a behavioral change related to #StayHome. The data comes from mobile location (thanks to Geopath and Intermx for proving aggregate county-level location data). The yellow line shows the confirmed cases two weeks after social distancing was implemented. R2 is included for each trend projection, which are fit to the time series confirmed cases data sourced from John Hopkins github.

The data and analysis is straight forward, as I think it is far better to have easy to understand models wherever possible. Here are the fundamentals:

  1. The virus has a two week period from infection to serious enough symptoms to seek medical care. Therefore, the hypothesis we are testing is, “Does social distancing have an impact on slowing the rate of growth, after factoring in the two week incubation period of the virus?”
  2. The confirmed cases are tracked, and a regression curve fit to the data from the weeks prior to social distancing having a measurable impact. Given the 2 week incubation period, the trend is based on March 15 to March 31 (gray line in the chart). This is compared to the period of time when social distancing should flatten the curve (April 1 to April 15 – yellow line).
  3. Social Distancing behavior per mobile phone is included, with the same two week lag for the incubation period) as a visual reference.
  4. The number of cases averted is the difference between where the trend from March 15 to March 31 was taking us vs. the trend after social distancing.
  5. The model fits a polynomial to the case time series, first at the US overall level and then for 50 individual state models. A sigmoidal distribution is most likely over time, with three phases, starting with a slow build phase, followed by exponential growth, and then leveling off phase. The R-squared for each model averages 99%. The model was validated by first checking the US Overall projection five days forward. The model averaged within 1% of actual cases for these five days. As a final check on the model, I compared the overall model to the 50 individual state models and they converged at between 23,000 lives saved and 25,000 lives saved.
  6. Lives saved is based on a 1.5% IFR (as observed in Diamond Princess Cruise), and only includes lives saved from the cases averted in April. The lives saved is likely greater than that if we were to project into May and beyond. However, since it is uncertain whether social distancing will hold, I did not project beyond April.
  7. The “what if” is an optimistic view if we continue to lower the growth rate and flatten the curve. I think it is unlikely that US will be as successful at lowering our rate of growth to the levels seen in S. Korea, China, Norway, Germany, but I am hopeful that sharing more data analysis will inform behavior.
  8. The COVID-19 Deaths vs. other Common Causes is borrowed from Business Insider, but since I first did a similar analysis for NYC back in February, I thought it would be helpful to use their chart and overlay each states current deaths per 1,000,000 and show the growth rate over the past week. I find this helps people understand that this is far more serious than causes of death like auto accidents that are not growing at 50% per week.

Findings: In the US, over 25,000 lives have been saved, and 1.5 million additional cases have been averted.

US COVID-19 Chart


The Risk Score is a measure of probability that COVID-19 growth will return (2nd Wave). The R0 is translated to a weekly growth rate. If new infections grow at less than 20% per week, the virus is dying out. If the weekly growth rate is higher, the virus is expanding. The weekly growth rate has been trending down for the US, but as of April 15, new cases are growing at 48%. This equates to a "high" risk score.


In terms of different states, only Washington is below 20% weekly growth. States that are trending down, and are below 40% current weekly growth are:

  • Alaska (30%)
  • Arizona (31%)
  • Idaho (22%)
  • Louisiana (29%)
  • Michigan (38%)
  • Montana (22%)
  • Tennessee (34%)
  • Utah (37%)
  • Vermont (25%)
  • Washington (18%)
  • Wisconsin (37%)
  • Wyoming (30%)

Of these dozen states, Louisiana saw the largest decrease over the past week. None of the states are in single digit weekly growth just yet — so we all have to be disciplined about #StayHome.

That sates that have the most work to do in flattening the curve are Kentucky — it is the only state to have increased, South Dakota and Rhode Island (they saw the smallest reduction in growth rate over the past week).

US COVID-19 Weekly Growth Chart vs. Italy and South Korea

For comparison, the chart shows the trend for South Korea (now at 2%), and Italy (now at 20%).

Analysis of each state includes a risk score based on growth rate, deaths and cases per capita. You can see each state by clicking on the map below. Each county is shown in the state view. At a county level, there is a wide range in growth rate and cases per 100,000. Click on your state to see how your county is doing

We haven't complete our first challenge of flattening the curve, but it isn't too soon to think about the next challenge — how do you increase commerce to keep people employed in such a way that weekly growth doesn't shoot above 20% causing a second wave that is worst that the first?

We can find some ideas from South Korea, which has been trending in single digit weekly growth rate for a couple of weeks now. Re-opening includes extensive testing, location app tracing, mask requirements and other social distancing practices. We need to be smart about how we adapt, else, we could crash our economy far worse than China or others that are able re-open and test, trace and contain the virus to under a 20% per week growth rate. For more on this topic, see the discussion on Rebooting The Economy.

What If We Get below 20% This Month?

What if we don't?

Given that COVID-19 was the #1 Killer in the US last week, working together to stop the growth has moral, social and economic implications. If we reduce the growth rate to 15% by April 22, and get to 5% by the end of the month, the virus essentially will burn out as each person infected passes it on to less than one additional person. The chart shows that lowering our growth rate to 15% then 5% over the next two weeks would save thousands of more lives just from actions in the last two weeks of April. It would avert another 250,000 cases. It would give businesses a shot at rebuilding without worrying that they will get their workers and customers sick.

US COVID-19 Chart Lives Saved

Can we stay below the 20% weekly growth level? South Korea did it. They are down to 2% weekly growth. We've been doing it for a month. I know we can do it for a few more weeks. However, if we can't (or won't) social distance and maintain our #StayHome behaviors, we will experience a second wave that this fall. As shown in my previous analysis, if we have lost the will to work together as a society to social distance and save lives, COVID-19 will return to being a top killer in the US. It would be 10x worse than what we just experienced. It will move well beyond New York and other big cities to infect every county — 85% of US counties have at least one confirmed case right now. Such a spread will make repairing the economy far more difficult.

US COVID-19 Chart Second Wave Forecast

If you encounter a skeptic, encourage them to run the numbers themselves. Take the current weekly confirmed case or number of deaths in your state, and simply multiply by 150% (our current growth rate), then multiply that number again by 150%, and again and again for the next 36 weeks. It is easier in a spreadsheet - here is a Google Sheet I set-up so you can copy it and give it a go.

A more sophisticated SIR model will show that there are two ways to slow the virus. SIR stands for Susceptible, Infected, Recovered. People are only infectious to others for a few weeks, so if the weekly growth rate slows, there are fewer people in any given week to get new people (those susceptible) infected. The other way to slow it down is for more people to recover and have immunity. Immunity is not a guarantee. Clinical research in South Korea are showing potential re-infections among some people that were confirmed to have "recovered." It may be that not everyone that recovers becomes immune. Even if Herd Immunity is possible with the Coronavirus, it requires somewhere around 70% of the total population to have been infected. As you can see from the 1.5% Infection Fatality Rate (IFR) in the spread sheet, we would kill millions before we approach herd immunity. Keep in mind that the small bump you see in April, in the chart above, was enough for COVID-19 to become the #1 cause of death in the US last week. The sharp increase in the fall is a consequence of letting the coronavirus return to spreading the way it was before #StayHome.

US COVID-19 Deaths vs Other Leading Causes of Death

Knowing the math of how much worse a second wave could get, why not embrace social distancing, mask wearing, hand washing, and yes re-building the economy based on social distancing? It is not an either / or situation between lives and the economy. We are the most creative and innovate country on the planet. We can afford to invest in sustaining businesses via the SBA so they keep people employed and have a window to re-orient for a post COVID-19 world. We can afford a GI Bill for Coronavirus Education & Training for the millions of newly unemployed that can take the time to re-skill for higher paying opportunities. We can accelerate high-tech manufacturing in the US, telemedicine to increase access and quality of care, modern infrastructure, AI for all kinds of economic efficiencies. Innovation in energy, and, my personal favorite, better data collection and analysis for government and businesses. There is so much we can do to pivot our economy to the future.

Update May 1, 2020: The forecast produced in this analysis proved to be highly accurate. The model forecast we'd hit 1,036,543 cases at the end of the month. We crossed that level on the last day of the month, ending slightly higher than forecast. Ultimately, weekly growth got below 20%, but not by much — meaning the US has not been as successful as Italy, South Korea or many other countries is aggressively slowing the spread of the virus.

State By State Analysis

Re-Igniting The Economy

Government data on the economy lags by a month to a quarter — sometimes a year. Decision makers need to know where the consumer economy is going (not just where it has been).

Get Below 20% Per Week

We've got to #StayHome and social distance until we are below 20% weekly growth per week. If we don't we are increasing risk of spread and thereby contributing to the loss of lives and livelihoods. To quote a Republican leader, Senator Lindsay Graham, "Try running an economy with major hospitals overflowing, doctors and nurses forced to stop treating some because they can’t help all, and every moment of gut-wrenching medical chaos being played out in our living rooms, on TV, on social media, and shown all around the world."

Modernize "Test, Trace, Contain"

When a country can't trace how someone got the virus, and it spreads so fast it becomes one of the top killers in a matter of weeks, the only option is to go medieval on the virus and shut things down and get below 20% per week growth rate. We have amazing technology. Let's use it as we emerge from lock-down to ensure we keep the virus at bay! Increase testing. Use your smartphone to opt into privacy secure tracing. Wear a mask. Avoid keep social distancing if you can. Help contain it by staying home if tracing shows you might have been exposed.

Coronavirus GI Bill & Help Those In Need

In January and February, the virus moved faster than we did. We lost our inability to test, trace and contain the virus. To regain the upper-hand required an unprecedented global behavior change that wrecked the economy. Tens of millions are out of work, and many service jobs will not come back for years. We need a Coronavirus GI Bill. Let anyone unemployed get on-line education or training for the jobs of the future. Pay people to upgrade their skill sets. If you can, take a family in need under your wing. Support your local food bank.

Forecasts to Aid Reinvestment

Governments and Businesses have never seen this dramatic of a change in consumer behaviors. Government data lags by weeks, sometimes months making it difficult to make sound business decisions. This paralyzes businesses, and slows economic recovery. My friends in Market Research are in a unique position to help. Join me in working with ESOMAR, The ARF and leading research firms to pool our survey and sales data to make consensus sector forecasts for where consumer spending is heading. It will help businesses and policy makers to make better decisions faster.

More COVID-19 Analysis

...

Recent Analysis

  • Forecasts (Jan to April) Vs. Actual :
    Summary

Alert Me When New Analysis Is Released