Memphis/Shelby County COVID-19 Updates
Back in mid-April, a month into sheltering in place, when it became clear that we were not going to contain the novel coronavirus, I decided to start tracking the local data myself.
At the time, there just wasn't much information, even less thoughtful analysis, and it was difficult to figure out what was going on. So I built a few spreadsheets, created a handful of visualizations to help "see" the data trends, and started writing weekly Twitter threads with my analysis.
Why should you care about my analysis?
That's a good question. I'm no epidemiologist. But I do have a graduate degree in ethics, where I did a lot of work around biomedical ethics. (Doing work around biomedical ethics and disability is actually how I ended up with a career in special education, by the way.) And so I do have some expertise in how to think about, how to weigh pros and cons around, public health.
And I have a passion for data, for "seeing" trends, making connections, and then painting a story of what's there. I find it helpful, for myself, to do this. Over time, after I did data projects on the side at work, it was added to my official duties. It's just how my mind works. I can't even help it. So, on the one hand, I do this for myself.
And so, as my curiosity led me to dive deeply into understanding the coronavirus pandemic, it just made sense for me to start building spreadsheets and visualizations and to start writing my analysis. This is just what I do - whether at work, at church, at home, wherever.
But others have found my analysis helpful. Many others, in fact. The first update had over 2,000 impressions and nearly 600 engagements, already about 200 times higher than normal tweets. And those numbers grew to 15,000 and 3,000 for the last update.
I mention that just to say that it seems like people appreciate my analysis. And so I offer it as a public service.
The public service aspect is the real reason I put my analysis out each week. And the reason I think it's needed - the real reason you should care about what I have to say - is because the public is not getting the full story from our local officials.
Analyzing the data, I've noticed the ways our elected officials have not always been honest with us and have not always made public health a priority. And so it is my hope that my analysis can inform the public discussion so that we can hold our elected officials accountable.
On this note, check out the op-ed I wrote for the Daily Memphian. (Read it here.) My goal here was to move beyond analysis to advocacy, defining a clear public policy goal and trying to shape the conversation in our community.
I was also interviewed by the Memphis Flyer a few weeks earlier. (Read the interview here.)
I'll note here that I made my first public projection - not to be confused with prediction - when the Flyer interviewed me on July 2. The day before, total cases had just reached 10,000. Looking at trends and growth rates at the time, I projected that we would reach 20,000 total cases by August 1. We hit 20k on July 29, three days early. I think that confirms my analysis.
So that's who I am and what I'm trying to accomplish. The other question you might want to know is where my analysis comes from and what my methodology is.
My data comes straight from the Shelby County Health Department. Each day at 10:00am, they update case counts and reported tests. And so each day, I pull those two numbers and plug them into my spreadsheet. I just pull the total number of cases and the total number of tests, just the most basic raw data.
Once I've got the data, the spreadsheets do a lot of the work. I've written formulas to look calculate the number of new cases per day, a 7-day rolling average of new cases per day, new cases per week, average new cases per day per week, and then all of those for testing data too.
I didn't have access to testing data at first, but once I got that I used case counts and testing data to write formulas to calculate the positivity rate overall, per day, on a five-day rolling average, and per week. And then I created visualizations for each of these.
After a while, I started playing with different time parameters to look at the data since reopening in Phase 1, since moving to Phase 2, since July 4 weekend, and more. I've also started looking at daily growth rate, as well as 7-day and 14-day growth rates. I've added a visualization comparing 14-day growth rates for cases versus tests.
Then I started using growth rate trends to project out - not predict - what cases might look like in the future.
And then I created visualizations around the metrics used by Harvard's Global Health Institute and Center for Ethics. The first is looking at positivity rates and testing to show our current status, as well as how many daily tests we'd need to be doing to get below 10%, 5%, and 3% positive. And then the other is around case rates, using the Harvard benchmarks of 1, 10, and 25 daily cases per 100,000 people.
Every Saturday, I spend a couple hours analyzing the data and writing my analysis in a series of 280-character tweets. I've also added a mid-week report as well. These are usually about 25 tweets long. But each day, I've started posting daily updates.
I wanted to archive my analysis here in one place, both for myself and also for transparency. Anyone who wants to go back to check my work can do so - and can do so easily. I think my analysis is pretty solid, but I know that I'm missing some context because I don't have all the data the health department has. And I welcome feedback. (I am grateful to the Twittersphere for offering some ideas early on that helped shape the way I thought about and presented the data thereafter.) If you have any thoughts or suggestions or context to add, I welcome feedback. Just shoot my a message via Twitter.
Every Saturday, I spend a couple hours analyzing the data and writing my analysis in a series of 280-character tweets. I've also added a mid-week report as well. These are usually about 25 tweets long. But each day, I've started posting daily updates.
I wanted to archive my analysis here in one place, both for myself and also for transparency. Anyone who wants to go back to check my work can do so - and can do so easily. I think my analysis is pretty solid, but I know that I'm missing some context because I don't have all the data the health department has. And I welcome feedback. (I am grateful to the Twittersphere for offering some ideas early on that helped shape the way I thought about and presented the data thereafter.) If you have any thoughts or suggestions or context to add, I welcome feedback. Just shoot my a message via Twitter.