Expect Next Week’s Covid Death Tolls to be Sensationally Exaggerated
Covid data has been underreported every holiday so far. Thanksgiving week’s underreporting will make for sensational (and misleading) headlines next week.
--
The chart below shows the CovidComplete forecasts submitted to the CDC’s Ensemble model for the two week period ending today. The forecasts were tracking perfectly through Wednesday, then strongly diverged on Thursday through Saturday (11/26 to 11/28).
If CovidComplete had really overestimated the death count for those days, that would be good news. But the reality is that the shortfall is just the latest example of underreporting Covid data around holidays. The deaths weren’t lower. The reporting of the deaths was lower — by about 3000 deaths.
Those 3000 deaths will be added to next week’s reports on Monday, Tuesday, and maybe a little on Wednesday. That deflates the death reports this week, and it will inflate the death reports next week.
According to CovidComplete’s forecasts for next week (shown in the graph below), Tuesday’s deaths were already going to be about 2500. If we add 1500 of the underreported 3000 deaths on Tuesday, about 4000 deaths will be reported that day. The previous one-day high was about 2750 on May 7, so expect some sensational headlines about the highest daily death toll ever. But realize much of that is really a timing issue around reporting.
That said, Wednesday, Thursday, and Friday next week were already looking like they could slightly surpass the previous all-time daily high, and next week’s total was already likely to surpass the prior all-time weekly high. We will hit a new all-time weekly high next week, but the high that’s reported will be 3000 more than it really was.
More Details on the Covid-19 Information Website
I lead the team that contributes the CovidComplete forecasts into the CDC’s Ensemble model. For updates to these graphs, more graphs, forecasts at the US and state-level, and forecast evaluations, check out my Covid-19 Information website.
My Background
For the past 20 years, I have focused on understanding the data analytics of software development, including quality, productivity, and estimation. The techniques I’ve learned from working with noisy data, bad data, uncertainty, and forecasting all apply to COVID-19.