Today I wrote the first of my bonus posts as a thank you to those who are kind enough to support the COVID-19 analysis financially. It’s on the Salmonella outbreak that has been linked to several onion varieties lately. Please refer to the images in the FDA recall alert to see if the onions in your pantry are linked to this outbreak. For more of my analysis a preview is shown below. You can view the full article as well as future ones by starting your subscription today.
I have some new resources to share regarding COVID-19 today and will be sure to highlight them as we go.
The first of the new resources shows us how newly reported cases have been added to the case curve, by date of symptom onset. As you may recall, using this graphing method, cases are backdated to when the person first began having symptoms or if they were asymptomatic it is backdated to the date on which their specimen was collected. One of my Twitter followers pointed it out to me (thanks @SpenceDawg118) and it helps to explain visually what I was trying to explain in writing on Wednesday, where I indicated that we may have more of a 21-day window of uncertainty rather than a 14-day window as displayed on the Georgia Department of Public Health website due to testing backlogs.
The tool looks very similar to the GA DPH case curve but has a slider function so that you can scroll through time. You can see the way that cases and the 7-day moving average move over time. But the *key* piece of information is the blue bars at the bottom. Those are showing where the new cases were added each day through backdating the cases to date of symptom onset. I encourage you to go look at this graph in motion because the static images don’t do this justice. Below is the graph for the date of 24April when Georgia reopened and around the time of what was previously our peak for cases.
Notice that there are small additions (blue bars) outside of the 14-day window of uncertainty, but the big changes are happening inside of that window. Fast forward to what we’re experiencing in more recent history. The next graph looks at 28 days ago, when we were in the early stages of our current case surge.
Notice how the bars are creeping up outside of the 14-day window in greater number now. There is no statute of limitations that says that cases from before the 14-day window are not counted - they most certainly are. But this is the beginning of our testing backlog problem - the initial signs that our testing capacity was overwhelmed. Plus, imagine how contact tracing is going to go for those whose cases were just reported but were tested three weeks ago. How are they going to remember all of their close contacts that long ago? Lastly, let’s look at the graph as of today’s daily report from DPH.
We still have significant backdating well outside of the 14-day window, with some of the cases dating back to July 1. So I feel more confident in saying that the decline we see in the case curve (by date of symptom onset) is not real. It is an artifact of our testing backlog. Given more time and laboratory throughput, we should expect that curve to continue to climb as it has been. The unfortunate consequence of this, of course, is that we really do not have a good idea of what is happening in Georgia. Our best data are 3-4 weeks old. That means that the actions we do or don’t take today to influence disease transmission won’t have impacts that we can see for 3-4 weeks. It makes it harder to tell if anything we’re doing is working.
In terms of testing, Georgia reported 28,367 new test results today and 3234 were positive, for a percent positive rate of 11.4%. We have been stably above 10% since early July.
The World Health Organization recommends that we be at or below 5% for at least 14 days before lifting any public health interventions. Georgia has ignored the WHO guidance, the White House Coronavirus Task Force and the Centers for Disease Control and Prevention for its reopening strategy. But for those who want to read how this should have gone, you can find the original WHO guidance here.
It was a big day for cases with 4109. The new statewide total is 209,044. A month ago, on July 7th, our total was 100,470. Every day seems like more of the same news and we tend to get used to it and not see the problem - I want you to know that we have more than doubled the case total in the past month. This is a big acceleration of the pandemic.
Another interesting tool that has become available recently is an event planning tool from Georgia Tech. It provides an estimate for how many people at your gathering or event will be COVID-19 positive based on local disease trends and group size. There’s county-level service across the United States so my out of state readers might enjoy this tool also. Since Paulding county schools have been trending on Twitter and making national headlines recently, let’s take a look the likelihood that at least one person is positive in a group of 25 people.
It’s hard to read, but the pop up says that the risk is 48%. The darker the county, the higher the risk. For example, in Bibb county, the risk of someone being positive in a group of 25 people is 79%.
Georgia Emergency Management Agency has been getting later and later each day with providing their daily report on bed capacity. In fact, it still isn’t up as of 6:00 pm when I publish today’s post. Obviously, they do a lot of critical work in this pandemic. So I don’t write this comment to criticize but to show people that they might be getting overwhelmed too.
In the meantime, there were 280 new hospitalizations reported today and the number of currently hospitalized has decreased to 2981. We haven’t seen a number that low for current hospitalizations since July 18th.
It’s certainly good news. But this may also be impacted by our testing backlogs. Based on what I’ve been told, Georgia Department of Public Health reports hospitalizations based on who was positive at the time of admission, not including those who was diagnosed after admission. For that reason, bed capacity may be a more reliable metric to examine for hospitalizations.
Another new tool provides hospital by hospital data on diversion status. This is a tool used by Emergency Medical Service (EMS) personnel to gauge what is the closest hospital that can serve a patient’s needs when they respond to an emergency. You can look at the whole state, sort by county or sort by EMS region. Unfortunately, the EMS regions are not the same as the GEMA Hospital Regions (A - N). The tabs along the top can be used to sort the data also. For now, that’s the best we are probably going to get at hospital by hospital bed capacity data beyond the GEMA hospital region level.
Unfortunately, it was a big day for deaths with 91. The DPH is reporting 92 but the math doesn’t add up relative to yesterday’s total or by adding up the new deaths per county today. Not sure what’s going on there. The other thing that’s weird about the death reporting is that based on my tracking (grabbing the information every day), our highest date of reported deaths was April 7th with 100. But based on the Date of Death Report graph from DPH, they’re indicating that date only had 79 deaths. By their accounting, today set a single day death record. The previous record (according to them) was 21May with 90 deaths, whereas I have that date down as 67 newly reported deaths. So something is fishy about the death reporting curve. I’d be curious to know if others are seeing this as well.
In either case, the deaths today were 91 (or using DPH math, 4117 - 4026 now equals 92). The new statewide total is 4117. The largest single county type for deaths was nonrural counties with 35 of today’s deaths. Another 24 came from the Atlanta suburbs.
I want to show two interesting ways to look at deaths right now. First is looking at cases and deaths graphed together (cases on the left y-axis, deaths on the right y-axis).
The rate of increase for death is almost parallel now to the rate of increase for cases and that is a frightening idea. We have had a tremendous surge in cases as I noted above. If this rate holds and we double the size of our deaths in the next 4 weeks as we have for cases, then we’re looking at 8000 deaths by Labor Day.
The next graph shows how our state fatality rate (gray, left y-axis) has varied over time in relation to our case total (blue line, right y-axis).
Our case fatality rate (CFR) dropped like a stone once our case surge began. And a lot of COVID-deniers took that as a sign that things weren’t so bad. But now that deaths have surged (as predicted) following the surge in cases, the CFR has abruptly halted its growth. Since we aren’t doing anything at a state or regional level to halt the spread of the illness and hospitalizations are having capacity problems, we should brace for that gray line to start increasing again, especially if our rate of increase for deaths matches our rate of increase for cases that we’ve seen in the past month.
Have a safe, socially distanced, weekend. I’ll be back on Sunday with the Week in Review. Be safe and be well!
This week’s podcast looking back at last week’s trends and a special focus on the scientific method
Also, I spoke last night at the virtual March for Masks on improving communication between scientists and the public.