It is great to read Mayor Fuller’s COVID-19 update today where she reports “the number of new cases reported in Newton has been smaller in the last few days.” I appreciate her consistent reporting of the case count for each day, but it is hard to digest these raw numbers and really see by how much the famous “flattening of the curve” is progressing.
I compiled the numbers from her updates going back to 23-March (www.newtonma.gov/gov/mayor/newsletters.asp ) and plotted both the cumulative “Cases Testing Positive” and “New Cases” as you can see in these graphs. I hope others find this as encouraging as I do.
This does indeed look very encouraging.
The only caveat about these statistics and nearly all others Covid19 stats is they are intrinsically tied to how much testing is going on over that whole period and what part of the population is being tested at different points. That said, I would guess that there’s more testing going on per day now than there was a few weeks back. If that’s true, the the curve is even better now than it appears on the graph,
It sure looks like good news.
What’s up with the logarithmic plot? Isn’t that going to distort the curve and make it look flatter than it is?
https://covid19.jackprior.org/2020/04/14/newton-4-14-update/
Day to day things are very noisy, but we have bent in Newton, MA, and the US (although RI not in great shape). In 3 weeks we may be seeing >10 on some days and 0 for others. Single case reports could persist well into June. Nursing homes (land-based cruise ships) are the wildcard in terms of how well they have been mitigated and will influence where we head.
To Jerry’s comment, my rough guess at this point is that actual cases are ~15x reported. Basically we know the case fatality rate should be <0.6% and limited US testing that makes ours close to 6%, and then you have maybe 50% of deaths that haven't been properly attributed. Whatever the multiplier, the trend is headed downward slowly in terms of new infections.
Excellent article here, discussing state level trends: https://thebulwark.com/newsletter-issue/one-chart-shows-which-states-are-in-trouble-with-covid-19/
Quote: “In short, while there is reason to think that we have passed the worst of the danger (for now) in much of the Northeast, there are lots of places where people are only at the start of their trial. We should be ringing alarm bells in Ohio, North Dakota, Iowa, and Nebraska. These numbers look very not good. Minnesota, North Carolina, West Virginia, Kentucky, Arizona, Utah, Colorado, and Indiana are all in the process of ramping up.”
Jennifer,
With a logarithmic scale a constant percentage change is seen as a constant vertical distance on the Y-axis so a constant growth rate is seen as a straight line. If the line is curving up then the virus infection rate is increasing, while if the line is curing down then the rate is decelerating.
A linear graph always goes up as new cases are reported, but 10 new cases on a base of 20 in the early days is a 50% growth while 10 new cases on a base of 400+ now is a 2.5% increase, so its harder to visualize the change in rate.
As you can see on the Johns Hopkins coronavirus tracking site (https://coronavirus.jhu.edu/data/cumulative-cases ) they offer both linear and logarithmic views, so this is standard practice. As they say on this page, “The logarithmic scale helps visualize early exponential growth.” I could post the linear graph too but I was trying to keep things simple.