Thursday, May 07, 2020

Novel Coronavirus outbreak, forecasting the end of the curve

The Netherlands Retrospect: RIVM terugblik "In the winter of (2017/2018), mortality was increased for 15 weeks (weeks 51 2017 to 14 2018). The increased mortality rate coincided with the flu epidemic that occurred from week 50 2017 through week 15 2018. The mortality rate during the 18-week flu epidemic was estimated at 9,444 (8,885 for the entire winter period from weeks 40 to 20). This was the highest mortality rate ever measured since the start of the monitoring in 2009. The mortality mainly affected people aged 75 years and older, but there were also several weeks in which the mortality was increased in people between 55-64 years and 65- 74 years.
A brief resurgence of the flu epidemic in weeks 10 and 11 2020 overlapped with the first 2 weeks of the COVID-19 epidemic. The mortality rate in weeks 10 to 19 of the COVID-19 epidemic was 9,768, of which 213 were in the first 2 weeks.

Anders Tegnell Sweden "What is the world basing their decisions on?" "there is the report (model) from Imperial ...that isn't even peer-reviewed." 

Last update: Thursday May 7th. Today is #rokjesdag The first wave is now recent history. It was fascinating, but the daily dose of death is depressing. Makes me think of a story I heard that a woman writing about genocide (in China?) killed herself. 

Conclusion May 7th 2020: My prediction was just a bit too low, but much more accurate than most, certainly compared to a certain college. The predictability of some things never ceases to amaze me. The laws of large numbers, etc. My respect for both Donald Trump and Elon Musk has grown and it was wonderful to compare notes with via Twitter. The memeshere like any technology is a double-edged sword.

Holland is on a "routekaart" based on the European Union "roadmap" which is not connected to reality and is stuck in a bureaucratic quagmire ;) A pain, but not the end of the world.

Guesstimate on March 19th 6 000 TOTAL* deaths in NL from the virus
Netherlands: Decline in April? 
Estimate of total weekly deaths, pencilled in. 

FORECAST March 19th, 2020
The forecast above, measurement below.

RIVM summary: "In the week from 23 to 29 April 2020, total mortality in the Netherlands was strongly increased (mortality reported within 2 weeks - around 97% reported). A total of 3,580 deaths have been reported, usually we expect between 2,609 and 2,913 deaths at this time of year. That's between 667 and 971 more deaths than we would expect. This is around 1.5 times the daily reported mortality from laboratory-confirmed COVID-19 in the same week (533). Mortality was slightly increased in the 35-44 and 55-64 age groups, increased in the 45-54 and 65-74 age groups, and greatly increased in the 75 and older age group. Mortality was increased in the combined regions of Zeeland / North Brabant / Limburg, Overijssel / Flevoland / Gelderland, and Utrecht / North Holland / South Holland. Mortality was not increased in the Groningen / Friesland / Drenthe region."

On Tuesday you can expect a higher number because on Monday data from the weekend is collected and sent to the

DateMeasuredGuesstimateGrowth RateGrowth
13-Mar10Declining by100%
14-Mar121% per day20%
1 May489356362.0%

because the 6 000 is spread over 4 weeks the peak shouldn't be much higher than in the influenza season of 2017-208 below: Maybe 3000 (normal line) + 2800 COVID19 = 5800? + 2200 = 5200? (adjusted April 2nd). (April 3rd 6 000 per week is more probable). No longer an epidemic in the Netherlands.

Italy Deaths 

DATE : DEATHS (cumul.): GROWTH %

March 19th Guess 3,5k 18% Measured 3,4k
March 20th Guess   4k 17%  Measured 4,0k
March 21st Guess 4,8k 16%  Measured 4,8k (as expected)
March 22nd Guess 5,6k 15% Measured 5,5k 
March 23rd Guess 6,5k 14%. Measured 6,1k
March 24th Guess 7,4k 13%  Measured 6,8k 12%
March 25th Guess 8,3k 12%  Measured 7,5k 10%

Adjusted Guess downwards on 25/3

March 26th Guess 8,4k 11%  Measured 8,2k   9%
March 27th Guess 9,3k 10%  Measured 9,1k   11%
March 28th Guess 10,3k 9%  Measured 10,0k 10%
March 29th Guess 11,2k 8%  Measured 10,8k   8%
March 30th Guess 12k 7%     Measured 11,6k   7%
March 31st Guess 12,9k 6%.  Measured 12,4k   7%
April 1st     Guess  13,7k 5%  Measured 13,2k   6%
April 2nd   Guess 14k 4%      Measured 13,9k  5% 
April 3rd    Guess 14,9k 3%   Measured 14,7k  6%
April 4th    Guess 15,4k 2%   Measured 15,4k  5%
April 5th    Guess 15,7k 1%   Measured 15,9  3%
April 6th    Guess 16k  0%     Measured 16,5  4%

Adjusted upwards again 6/4

April 7th    Guess 17k 3%      Measured 17,1k 3,7% 
April 8th    Guess 17,5k 3%   Measured 17,7k 3,2%
April 9th    Guess 17,9k  2%  Measured 18,3k 3,4%
April 10th  Guess 18,2k 2%   Measured 18,8k 3,1%
April 11th  Guess 18,4k 1%   Measured 19,5k 3,3%
April 12th  Guess 18,6k 1%   Measured 19,9k 2,2%  
April 13th  Guess 19k             Measured 20,5k 3%
April 14th  Guess 19k             Measured 21k 2,9%
April 15th  Guess 19k             Measured 21,6k 2,9%
April 16                                   Measured 22,2  2,7%
April 18th                                Measured 23,2k 2%
May 6th                                   Measured 30k    1%

Track total deaths here: 

Guess March 23 Less than 16k deaths? Guesstimate March 19th less than 20k deaths in Italy.

Update April 29: Total measured COVID deaths in Italy (and the UK) are higher than my guesstimate. In the Netherlands, the guesstimate has been remarkably accurate. It falls between the daily measurement and weekly measurement (RIVM Sterftecijfers). The decline of deaths is slower than expected. The government seems to be backtracking on earlier statements of stopping the lockdown because the Outbreak Management Tea, no longer "believes" in Groepsimmuniteit. The acid test is what happens when IC bed use for COVID 19 in NL falls below 700 around Bevrijdingsdag next week... 

Update April 7th. People are canceling activities in July and August as well as the end of May here in the Netherlands. Few people are infected in the north of Holland it seems, but there doesn't seem to be much "serological" or antibody testing taking place. It is not even clear to me whether it is possible. Good news: flattening the curve works. Bad news: little evidence yet of herd immunity anywhere on Earth?? After Easter Austria and some other countries are easing lockdown. Lockdown of countries like Pakistan where very few people seem to be sick seems ridiculous because they will probably need a second wave lockdown again. Boris Johnson currently in intensive care... 

Update April 3rd. "Dunkelziffer" = officially unknown number of certain [negative effects] events. 
The official number of deaths caused by COVID19 is lower than my forecast, but the actual number of people who have died in The Netherlands is higher than my guesstimate. 

Update March 19th: tracking curves and forecast for Italy and The Netherlands

Worldwide Panic: Euro 2020 has been canceled and pretty much everything else. Also Berkshire Hathaway meeting in May has been cancelled. People buying guns in the US. And unexpectedly this has become the TOILET PAPER PANIC. Heaven knows why... 

The good news is that it seems the spread in China was stopped by immunity as much as drastic government measures. So Italy (and then the rest of the world) should do fine even without tracing every case. 

Italy showing the same trend as Wuhan / China earlier: "The 1% drop in the rate of growth of total registered cases per day: 63%, 37%, 27%, 50%, 20%, 23%, 23%, 25%, 24%, 11%, 23%, 21%, 17%, 20%, 17%, 13%, 13%

Update March 13th. Now classes as a global pandemic. Containment no longer a goal, except in China where cases have dropped from 60k to 12k today. Worldwide reported number of people under treatment has jumped up to 59k from a previous dip to 39k. 

Public life has stopped in much of Europe until  March 30th Flights to US stopped for Europeans and Chinese. A palpable sense of panic also in stock markets.

Update March 9th it no longer looks like this Novel Coronavirus will be contained. Measures taken in the rest of the world have been much less vigorous than in China.

I believe the estimates I made on February 4th were wrong.

 Although active cases in China have dropped from 60k in China to less than 20k in that country today, worldwide cases are increasing again. At the beginning of March they dropped to 39k and are now up to 44k with the majority outside China. My estimate of 4k deaths (between 2k and 8k) and 200k sick now seems (much) too low.
Notes "herd immunity" is something that seems to be underestimated. "WorldWar Z it is not" Elon Musk, but the  Novel Coronavirus and reactions to it is a shock to the system.

Update February 27th active cases worldwide dropped to 46 551 yesterday.

February 26th number of sick worldwide yesterday dropped to 48 186. (81 004 - 32 818) This drop is not being widely reported or registered as far as I can tell. The numbers being reported does not include the number of people who have recovered.

24 februari number of sick worldwide decreasing quickly: Feb 23rd 79 554 minus 2 628 deaths minus 25 235 recovered is 51 691 sick.

Update Feb 19th. Spread a bit slower than I thought:

February 4th 2020:the percentage of growth is slowing. I guess the total infected will peak around 100k and not higher than 200k. Deaths between 2k and 8k. Peak in less than 4 weeks. (China)


Manuel said...

Thanks for the detail info!

Martin Spanjaard said...

Hi, do you have information about the incidence rate per age cohort, or something like that? And about the health state of patients in the intensive care before they ware hospitalized?


Martin Spanjaard

Ansgar John Brenninkmeijer said...

No I don't.

SYSCheck said...

These numbers might give additional statistics: