New cases. Cumulative cases. Hospitalizations. 7-day average death rates. Breakthrough cases. R0. Positive test rates. 2020 introduced much of the world to casual epidemiology, and while some of us hid our heads in the sand, many of us paid close attention to new case hotspots, superspreader events, ICU capacities, and how close we might be to herd immunity. The Seattle Times has a basic primer on how to read infectious disease numbers. On this page, I am going to try to filter through the mountains of information and perhaps present numbers in ways not seen elsewhere.
Cases are easy to count, and thus case numbers are readily available, often down to the city or neighborhood level. (A prior version of this page used city-specific data compiled by a concerned citizen during the first 18 months of the pandemic.) Early on, case counts were important: What areas might I want to avoid because they are experiencing outbreaks? But three important factors make case counts less informative as we reach the end of 2021:
So, a positive case no longer means the same thing in December 2021 as it did the prior December. 10 new positive cases in a day says nothing about symptoms or actual risk of hospitalization, long term consequences, or death for those 10 people. A new variant might spike case counts, but need not lead to panic. Really. That said, the more newly positive cases, the more virus circulating, and the greater the chance that anyone, vaccinated or not, might be conveying COVID to anyone else, and thus the greater the rate of bad COVID outcomes.
Deaths are the final arbiter of risk, but deaths capture multiple risks and there are many uncertainties in death investigations and certifying of causes of deaths. Someone dies with symptoms of cardiovascular depression, pneumonia, high blood pressure, and diabetes, with evidence of opioid use, a positive viral test for SARS-CoV-2, and a history of obesity: Different medical examiners might come to different conclusions regarding the single underlying cause of death versus other contributing causes. The current emphasis is clearly on the SARS-CoV-2, the virus that causes COVID-19. For those interested in learning more about death certification in the US, this New York Times Magazine article addresses some of the myriad issues through the lens of the opioid crisis, and this Guardian piece addresses the politics involved, particularly in counties with coroners as opposed to medical examiners. Closer to home, KNKX has a three-part series on challenges inherent in the death investigation system in Washington.
In between cases and deaths, often literally, are hospitalizations. Hospitalization rates are less widely available, but I do present hospitalization rates by vaccine status below.
The curves we wish to reverse: Flattening of cumulative curves means zeroing out curves representing added counts. Newly confirmed cases and deaths are presented, as rates per 100,000 residents in 7-day trailing averages. This smooths out daily noise as well as missing reporting days (e.g., weekends and holidays), although much variability remains. These data are from the New York Times compilation of COVID data, graciously made available to the public on GitHub. Each chart starts with all 39 counties and statewide series hidden so you can select counties of interest. I mark the first month of variants of concern, but note that variants do not become prominent for a few weeks after they emerge, depending on infectiousness. At this writing, omicron is just spreading. (And is no fun, trust me.)
If these curves are going up in your county, please consider lowering your transmission risk level. And remember, the death curve will trail the cases curve by 2-4 weeks. So if the cases curve is going up now in your county, the death curve will likely go up, if it is not already doing so, next month.
With vaccines becoming available early in 2021, the above curves are lower than they could have been. With ongoing vaccine hesitancy, they are not low enough.
The state Department of Health began publishing (in a PDF report, at first biweekly and now apparently weekly) case, hospitalization, and death rates by vaccination status. These are given as the count of events in the prior 4 weeks per 100,000 people in the age group with the vaccination status. (4 weeks makes events divided into age-by-vaccine-status groups a bit more stable.) They do not publish death rates for the young and middle aged groups because deaths among the vaccinated under 65 are so rare that rates would be highly unstable.