Testing for COVID-19, local and worldwide practices

E : Here's an update on the COVID-19 testing data.


Here we include the worldwide data, and results for the 24 countries that we have been following daily.

In addition, some comparisons with testing locally here in Belgium.


Belgium is of course special in that it is divided into three regions, with linguistic but also and perhaps especially, poilitical and economic separations.


So are testing practices for COVID-19 in Belgium in line with the mean value of such testing around the world?


Are there differences in each of the three regions of Belgium?


Let's start with the global view to find the quantity of tests done by country.


That looks like this :




One can regroup this, to compare with the prevalence of COVID-19, confirmed in each country.


And that looks like this : Confirmed cases along the bottom (X) and Total tests for COVID-19, done (Y):




Next, the data from our sample of 24 selected countries, which we've been following :

Tests performed per million inhabitants - 28 March, 2020

And with Belgium excluded, about the same look, and a very slightly different mean :


Tests performed (excluding Belgium) per million inhabitants - 28 March, 2020



In Belgium, total reported tests had a value on March 18 of 18,360 total tests done , or 1584.2 tests done per million population.


How does that compare with the mean from those 24 countries?

Now sampled without Belgium since we'll be comparing ...


You've probably guessed the result:


Mean and for my 24 country sample :  1,348.7±1,465.1
23 countries, excluding Belgium : 1,338.0±1,497.2 
Result for Belgium : So 1584.2 tests done per million population.

Comparing that with the mean : (don't worry about how these tests are done) :

Z = 0.16444 

The P-Value is .434708.
The result is not significant at p < .05. (means, no difference from the average number of tests done).


Conclusion :

Belgium is very much in step with average testing performance of these 23 other countries.



How about by region in Belgium?

We don't know that, because total tests, and tests done by region were reported for a few days, but no longer.


We have to explore for potential differences in testing practice, by comparing "positive cases."


A "case" can be taken as synonymous with a "positive test."


The distribution of laboratories in the three regions is not symmetrical as one can readily see from the map below.


So results look like this:

  Laboratories  Flemish region  Brussels region  Walloon region
  22 16 2 4
  % of labs : 72,73% 9,09% 18,18%
# of COVID-19 Positive tests   3,880 709 1,490
  % of +'ive tests 68,826% 11,663% 24,511%
  % of deaths 36,4% 23,2% 40,5%


Are there really fewer cases of COVID-19 in Walloonia? Or simply that fewer tests are being done?



The labs doing the tests are located on this map : those with a yellow "L" (for 'lab') in Flanders, blue "L" for labs serving Brussels, and the green "L" for labs in the Walloon region. The map certainly fits with the above numbers. Clicking on an icon, gives the name and address of the lab.



If you don't test for the disease, how are you going to find it?


Does this discrepancy or laboratory maldistribution lead to an otherwise inexplicable difference in reported cases by region? It would seem so. 


As outcomes become more apparent (those who recovered; those who died) in each region, this may prove to be a significant cause of underdiagnosis and treatment, or delay of treatment.


For now let's take all the deaths reported so far, reported by region, and see if the number of tests done per million population was different in the three regions.


Those results look like this :

Tests (ou cas) par Région en Belgique - 28 mars 2020

Tests in Walloonia are happening at about 20% below the other two regions. 

I prefer my maldistribution of laboratories explanation, rather than evoking that it is somehow tied to the usual and frequent state of non-cooperation that exists between these three regions. 


Actually, to optimally meet the COVID-19 challenge that increases daily, both problems need to be effectively addressed at "full speed ahead." Is Belgium ready to do so?


Conclusions :

  1. Testing rates vary greatly from one country to the next around the world.
  2. Almost invariably, the more confirmed cases, the more tests done. The two would seem inextricably linked, since a "positive test" = a "confirmed case.". 
  3. On average 1,350 tests per million population have been done per country in the sample of 24 countries studied. The large standard deviation for this sample, ± 1,465 captures the large variability in testing practice. 
  4. A country (like Belgium) can be in step with what other countries are doing on average with regards to testing.
  5. Significant discrepancies can exist within the country itself (like Belgium), and these may prove to be determinants of variations in outcome from one region or locality to the next.



Of course, testing and equity in availability to testing, are just a small part of the required solution.


Here's a very nice article to help pull in all the required parts.


Governments in quiet denial, just won't help any longer.




<<<<<<<  Home



0 Poster un commentaire

Inscrivez-vous au blog

Soyez prévenu par email des prochaines mises à jour

Rejoignez les 5 autres membres