Posted on 03/23/2020 3:01:09 PM PDT by stremba
Please be nice; this is my first vanity. I did this just out of curiosity because it’s not something I have seen elsewhere. All data is from https://www.worldometers.info/coronavirus/country/us/
I was interested in the growth rate of new Covid 19 cases and have been monitoring the link above. I noticed a large spike in cases yesterday along with a relatively small number the previous day. It occurred to me that daily growth rates are too noisy to really matter, but a longer term average might be a better way to look at it.
I started my data analysis with the 3/7 data, which was the first day with 100 new cases. I calculated for each day a three day average growth rate as follows. First take the ratio of cumulative cases for the current day to the cumulative cases 3 days prior. Take the cube root of that ratio, Subtract 1 from that cube root. Multiply that result by 100 to convert to a percentage growth rate.
The results are below. Not sure how it will format on here so the format is
Date——new cases——cumulative cases——3 day growth rate. Note that the growth rate is not calculated until 3/10 since there I only included data starting with 3/7.
3/7——116——435——na
3/8——106——541——na
3/9——163——704——na
3/10——290——994——31.7
3/11——307——1301——34.0
3/12——396——1697——34.1
3/13——550——2247——31.2
3/14——696——2943——31.3
3/15——737——3680——29.4
3/16——983——4663——27.6
3/17——1748——6411——29.6
3/18——2848——9259——36.0
3/19——4530——13789——43.5
3/20——5594——19383——44.6
3/21——4824——24207——37.8
3/22——9339——33546——34.5
I hope so because I think another 30 days of 30 percent increases would be a very big number.
I strongly suspect that the “new” cases are actually old cases that were not detected but now with more testing kits available, the numbers are “growing”.
“Maybe we’ve reached the peak of growth?”
I would take no comfort in that just yet. A 30%+ daily growth rate represents a virus out of control — perhaps not everywhere in the U.S. but certainly in NYC.
Too soon to make an assessment of a trend.
What is more important to me would be the percentage of those dying verses total tests conducted during this period of testing for those having symptoms or contacts with those infected.
Like predicting the number of ants from ants spotted. I suspect its more widespread but that the great majority don’t get symptoms.
I think it's just hard to tell because no single population has attacked it in exactly the same way.
03/19/20 55%
03/20/20 44%
03/21/20 39%
03/22/20 32%
While that looks like it's dropping quite a bit, I'd expect a small spike when the weekend and post weekend numbers work through, but it's definitely a downward trend.
I don’t trust any US data before today. We are detecting milder cases and detecting earlier as we roll out test kits, so the new numbers are inflated relative to our older numbers. The true percentage growth rate is certainly smaller than what we see. How much smaller? Insufficient data. I’m going to start my serious US analysis with today’s data, and I may even throw out this week’s data too.
In general: China, South Korea, Italy (finally!), Germany (just starting), and Spain (just starting) have controlled this or are at least on track to control it. Iran’s numbers are complete fiction. Our numbers are untrustworthy. Even the countries that have controlled it could still have major outbreaks return.
Number of “cases” means very little actually.
It is only meaningful when the “number of cases” is broken down by attributes like “no symptoms”, “mild”, “severe”, hospitalized, ect. Why? If the numbers are running true to experience, most “new cases” are in the mild or no symptoms categories, with a smaller number registered as severe, and a much smaller number ending in “deceased”.
Whole numbers say less about the spread than when the number of cases testing positive is weighed against the number of cases tested, which gives you an idea of the spread - as a % - in the population. THAT number had been declining - as the number of tests keeps increasing the tested positive numbers reflect smaller and smaller %’s of those tested.
Secondly, the positive rate on the tests of 10% is good news, that means that only a small part of those at risk or showing symptoms are actually sick. The majority are not sick, and that's in the worst populations.
Also, the number can't increase at a high rate forever, sooner or later you run out of people to get the disease. For example, an increase of 30% per day, you'd reach the entire population of the United States in about 34 days. At some point it has to start leveling off, regardless of the infection rate.
Good stuff. Thanks.
Bounces but relatively flat.
Problem with case analysis is testing is variable across regions.
Waiting for specific criteria to test selects for a steady rate of new cases.
Deaths is a better parameter in my opinion.
New York and Washington look bad in that regard. Most of the country looks well off, except maybe Louisiana.
The NYC region needs to be contained.
Maybe, but number of confirmed cases is the data that’s actually available. I wanted to post based on hard numbers, rather than speculation. Certainly these numbers could be misleading for the reason you point out, but the available numbers are the best data I had to work with.
With Italy potentially fudging the numbers up, Germany and others fudging them down to prevent a panic, and the lack of testing its hard to say much at this point.
I also suspect they’re testing the hell out of the healthcare workers and that could distort the data.
As yet to be defined.
Everything is a guess ~2.
Thanks for the additoional info. I had worked through the 2 day averages as well, but they seemed more noisy, especially earlier in my data set, so I opted for the 3 day averages. It’s always a bit of a judgement call with data analyses such as this.
The recovery rate should begin coming close to the new case rate +14 days or so. The current number of recoveries is about 545. Two weeks ago (March 8) there were only 541total cases.
unless you test people without symptoms and not reporting, we will never know how high/low the death rate is.....I suspect many many people have had it or have in now, without reportable symptoms....
Certainly true, but unless this outbreak behaves differently than it has in other countries and differently from all other diseases, exponential growth at a constant rate for a sustained time won’t occur. Again, epidemiology is not my area of expertise so I won’t say it couldn’t happen, but it seems unlikely.
My understanding is that outbreaks follow logistic rather than exponential growth. This type of growth is characterized by a slow growth rate initially, followed by a rapid increase in growth rate, then a slowing of growth to a linear rate and eventually a near zero rate. A few countries already have reached the linear and/or near zero growth stages (S Korea, China). It took about 4-8 weeks from initial outbreak to reach that point. The US is about 3 weeks in, so I wouldn’t expect rapid growth to be done, but the slowing of the rate may be an indicator that we’re not too far from the linear growth stage.
True, but that isn’t really likely unless this pandemic behaves completely differently from all others. There’s already evidence from other countries that this disease will not behave differently. The growth rate will decrease; it’s a matter of when and how fast.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.