If you follow @tjmule on twitter, you know he's long been an advocate spiking case loads are followed by spiking hospitalizations followed by spiking deaths. At this stage of the pandemic, they're not likely to happen in parallel. But we're just data folks so the only thing we have to back this up is...well...data. Then we saw @epiellie's tweet about Lead Time Bias. Click the tweet thread below to read more but it's an excellent and accessible demonstration of what we've been advocating -- we have to be more patient with the data right now before we assume everything is rosy and we can return to normal. We hope it's clear we're not "rooting" for more deaths to prove out what we're seeing. We're just hoping that we can use data to understand what's happening.
We decided to dive into the data a bit more to look at states that have to started to surge in the last 3-4 weeks to see if the Lead Time Bias holds up. We look Arizona, Texas, Florida and end the article by looking at Georgia with a few predictions on where we think death counts will go.
(note: we refer to the June spikes as wave 1a or wave 2. we've seen both used so we're trying not to take a position on whether this is part of the first wave or a whole new one)
(note2: our time window is from mid june (ish) to July 3rd. data during and after the July 4th holiday is a little unpredictable at this point)
(Data available here: https://www.covidmappingproject.com/states/texas)
Texas started to see its case surge on or around June 16 with 4,098 new cases and a trailing 7 day average of 2,279.
Deaths "peaked" on July 3rd -- about 17 days after the case spike started.
(Data available here: https://www.covidmappingproject.com/states/arizona)
Arizona started to see its case surge on or around June 16 with 2,392 new cases and a trailing 7 day average of 1,543.
Deaths "peaked" on July 2nd, 16 days after the case surge started (it's likely that AZ was off on July 3rd for holiday and had limited reporting on that date)
(Data available here: https://www.covidmappingproject.com/states/florida)
Florida started to see its case surge on or around June 13 with 1,542 new cases and a trailing 7 day average of 2,581.
Deaths "peak" on July 3rd, 20 days after the case surge began
(Data available here: https://www.covidmappingproject.com/states/georgia)
Georgia's second wave (or wave 1a or however you want to view it) started later than the other states. It started on or around June 20th with 1,800 new cases and a trailing average of 1,001.
However, we still have not seen a corresponding uptick in deaths. Clearly we hope this is a trend that continues, but will it?
What we think will happen in Georgia
To be clear, this is not an advanced model nor is it very "risky". We base our assumptions on what has happened in AZ, FL, and TX:
- Deaths start to go back up anywhere from 17-20 days after the first surge in cases
- We use a *very* simple run rate of new cases per day until the end of July (~2,600 new cases per day)
- We create a lower and upper bound of expected deaths (1.1% lower and 2.4% upper) based on cases
- We recognize that there are more cases than are reported but for the purposes of this model, we're just using reported data
In conclusion, we think a few things will happen:
- We expect deaths to start to rise again in Georgia on July 10th
- The upper bound (for the rest of the month) will hit around July 22nd with 83 reported deaths
- The lower bound will hit around July 21st with 38 reported deaths