Covid-19 Estimating delays between confirmed cases and deaths

Table of Contents

Introduction

This is an attempt to look at the delay between patterns (e.g. peaks, valleys, slopes) in Covid-19 confirmed cases and the resultant patterns in deaths. 

Methodology

We are using the Johns Hopkins University (JHU)  data compilations of the cumulative confirmed cases and deaths for countries of the world  (see https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv and https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_global.csv).  From these data we extract the data by country and create chats of the daily 7 day smoothed daily confirmed cases and deaths by country organized by region. The charts are created using the Perl Excel package (see https://metacpan.org/pod/Excel::Writer::XLSX::Chart)

The patterns include peaks, valleys (for example Indonesia), and slopes (see for example Australia or Latvia), with peaks being the most common. Problems with identifying corresponding patterns include:

• Peaks that are very broad or not sharp enough (see for example Sweden).
• Peaks which consist of multiple sub-peaks where the sub-peaks for say confirmed cases do not match those for deaths (see for example Hong Kong or Mexico).
• Peaks in deaths that appear to precede the confirmed cases (see for example Tajikistan)

Other matching problems include:

• Countries with sudden step increases in the confirmed cases or deaths or both, presumably to catch up on missed data entry for the country (see for example Ecuador and Kyrgzstan)
• Countries with low statistics, especially for deaths (see for example Taiwan).
• Countries with no deaths reported (see for example Cambodia and Laos). 

If one takes the statistics of the Diffs they are seen below:

A map of the world countries colored by the median diffs (countries with no Diffs are the same shade as Antarctica) is shown below:

Also the individual Diffs are given in the table below.

The images below show the actual time series of confirmed cases (solid line) and deaths (dashed line) for representative countries and regions of the world.  We choose these countries since they have large numbers of confirmed cases, and are expected to have a reasonably good accounting of the cases. The Diff is the approximate number of days prominent matching features in the time series (e.g. peaks, valleys and slopes) of the two cases. In many, but not all cases we identify the matching structures and the Diff in days. You can click on an image to blow up its size and see details. 

Northern, Western, Southern Europe, Iran, US and South Africa

Belgium Diff= ?, 15 days	Denmark Diff=32 days	Finland Diff= 14 days, 9 days	Germany Diff =19 days, 21 days
Greece Diff = 14 days	France Diff= -7 days, 14 days	Italy Diff = 6 days, 14 days	Iran Diff=8days, 12 days
Netherland Diff = 10 days, 19 days	Latvia Diff 13 days	Lithuania Diff 11 days	South Africa Diff=9 days, 5 days
Spain Diff= 4 days, to come	Sweden Diff=peaks not sharp enough	US Diff= 12days, 14 days, 12 days, 6 days	UK Diff=14 days, 14 days

Central Europe

Austria Diff = 9 days, 19 days	Bulgaria Diff = 17 days	Czech Republic Diff = 11 days, 6 days	Hungary Diff = 19 days
Poland Diff =15 days	Romania Diff = 20 days	Slovakia Diff hard to clearly identify	Switzerland Diff = 14 days, 12 days

South Asia

East Asia and Australasia

Australia Diff = 10 days, 21 days	China Diff = 5 days, 5 days	Hong Kong Diff = 10 days, 10 days,  20 days	Japan Diff = 13 days, 11 days
South Korea Diff = 11 days	Taiwan Diff = ?

Commonwealth of Independent States (CIS)

Armenia Diff 8 days	Azerbaijan Diff 10 days	Belarus Diff ?	Kazakhstan Diff $
Kyrgystan Diff ?	Russia Diff 20 days, 0 days	Tajikistan Diff -ve days	Ukraine Diff 16 days
Uzbekistan Diff ?

South America and Mexico

South East Asia

Cambodia Diff ?, 0 deaths reported	Indonesia Diff 7 days	Laos Diff ?, 0 deaths reported	Malaysia Diff 2 days
Phillipines Diff ? difficult to match peaks	Thailand Diff ? difficult to match peaks	Vietnam Diff ? only 35 deaths