Observations about control of the numbers in Belgium

The COVID-19 situation in Belgium, must surely benefit from placing it in the larger context of the COVID-19 situation in our world.


The unfailing help of Johns Hopkins CSSE will be turned to. This graphic summarizes the case data from 188 countries with infection as of yesterday June 27 ...


World - Daily Cases - June 27, 2020


Conclusion :

The burden of daily cases in the world is increasing markedly. If current trends in Belgium are comforting, and they are, the relationship to this curve is nevertheless unmistakable. On the left, when Belgium announced its first cases in February and March, where did these cases originate? From elsewhere in the world. SARS-CoV-2 got into Belgium from elsewhere.


And on the right, the world has now many more cases of COVID-19 to report and to "share" with Belgium than in February. An increase of thousands of percent. 


And yet, the world continues to "reopen." Belgium continues to reopen. Not just planning to do so, but already well underway, including international travel. At the beach, at café terraces, in grocery and other stores: a rare mask or social distancing "fanatic" still in evidence since the sun has come out.



While the above continued increase is impressive and concerning, this illness concerns less than 1% of the world's population. 


Less than 0


So, are we getting too excited abut a disease that involves such a small percentage of our world?


That typically happens with events that kill and maim people. Human nature magnifies the problem by asking: "What if it were me?"  Unlike in centuries past, bad news travels much faster today at electron speed.


COVID-19 kills about 5% of those who contract it worldwide. That means that the body count continues to be added to each day, just as non-lethal cases increase. But with time, when compared with February 27th, the death rate is slowing.

Trends 4 - Change in Deaths and Death Rate compared with Feb 27


Specifically, the death rate from COVID-19 on February 27 was 3.41%. (n°1 below).

On 27 June, 5.04% (n°2). That is of course a higher rate, and a 47.9% change (n°3). 


Change in Death Rate explained - June 27, 2020


If one backs up far enough to look at the last 90 days ...


In dark purple below are the cummulated deaths. This is a body count, and will never come down, but simply round and flatten at the top. It will stay at that maximum forever after the last death is added. An odd epidemiologic monument hidden away in an archive.


World 90 days - deaths and death rates - June 27


The light purple or gray curve is the change in death rate. It is now coming down.

Again, this is a comparison each day with the death rate on February 27.


In the last 8 days, the numbers present like this. In each case the number above the column is the percent above the death rate value identified on Feb. 27. It's still above that date's value. It's coming down. 


Last 8 days of change in death rate - June 27

Why would death rates be diminishing? One explanation selected out of many, is that medical therapy is learning, as the pandemic proceeds, how to better apply medical therapies that save lives. This, without a vaccine. Another of course is that a significant number of people learned that if you stay indoors and away from sources, you don't die. More and more came to believe in that. It took a while for that to take effect. No one knows where that is headed today.



A final word of hopeful news at the global level


Doing body counts correctly is not the goal, or shouldn't be.


The goal is to avoid transmission, and get those who are actively ill to a recovered state.

How are we doing with that goal of saving lives and healing the sick where possbile?


There is a gradual improvement. This is expressed below as a ratio: the number of Recovered person(s) for each Actively ill reported case. 

The numbers are large: 4 945 557 recovered. 4 381 834 actively ill in the world today.

That generates a ratio of 1.134 :  1. So just over 1 person recovered for each person still sick. 

This relationship has only turned positive (ratio above 1) in the last 2 weeks.


It is hopeful and comforting.

But it does not come on this day with any guarantees of permanence.

World R to A ratio - June 27, 2020




We are now at a point in the pandemic in Belgium, where confirmed cases and deaths have diminished significantly on a daily basis. This has been noticed at a global level, and resembles the trajectory of most other countries. The virus arrives on a certain date. It does its thing for 4 or 5 months. Then it leaves. That would be nice. But some effort is required to get it out the door and on its way...

New Deaths declining for these countries - June 24, 2020



These numbers have diminished to a point where those doing the counting in Belgium have changed their method for reporting the COID-19 case results to the Belgian public. When there's nothing to count, the most important task may become deciding where to have lunch.


In Belgium, values are now averaged over 7 days, but two days a week, no daily results are reported. Or if you prefer, 0 is the unreported result. That looks like this, and certainly makes for a less attractive and comprehensible graph :


Nouveaux cas - 27 juin

Key question on this day

If the need arises, and cases rise again, as happening in other places in our world, will Belgium respond again at its slow pace to an obvious need to resume vigilance?


In what follows, all data are taken from SPF Santé publique, through their agency named Sciensano.

The data discussed here, their data, can be downloaded as Excel spreadsheets at this link.




What have we learned from carefully following those data presented over time?

  1. Data accrual has had its problems
  2. Even when data are lacking, conclusions are still drawn and presented
  3. A defensive attitude has arisen which seems to have separated experts, government, and probably the public, and which the Press of course loves. "If it bleeds it leads." Bring on the pain.
  4. The data are presented to the public surrounded by an aura of certainty, while much variability in the numbers actually exists. What follows tries to discover, present and underscore that variability.




Case data are now presented as 7 day averages, like this:

Tendences clées - 27 juin


OK. That seems like a quick read. 


If one reads the fine print, (see below) one learns that for their 7 day averages, at least the 3 last days reported have data that "are not yet consolidated." This means the numbers will most certainly change. Or, 3/7th of the numbers are bad numbers. "Garbage in. Garbage out."


This can pose a degree of aggravation for anyone trying to follow along day by day. That's less important. They would probably agree.


Of course the goal is to get to "no cases to report." Forever.

Just counting whatever comes in, (and complaining about what didn't come in from Flanders once again), that is not the goal.


So do current methods of reporting at Sciensano contribute to that goal of completely ending the infestation in Belgium or not? For that should be the goal. 


La merde c\\\'est de la merde


We have grown used to this footnote. Perhaps too accepting.

We have lived with results that were, as it says again today, "pas encore consolidés" (incomplete) when following reported deaths in Belgium.


A very touchy subject in Belgium which still has a death rate around 16% when the world's mortality figure is closer to 5%. 


Understanding outcomes that lead to death of a patient is an important part of medical therapeutic success (and failure). Does an illness affect men or women more? In what age groups? Are certain areas in the country more affected? Answering these and other questions about death events comes with the turf of trying to avoid death of a patient when one is charged with selecting and doing the treatments. Basic and essential natural history of a disease is what it's called. 


As already presented here in other articles, a trend of incomplete records has been in evidence for months, with only gradual and minimal improvement. Writing to the government officials involved got no response.


Age and gender data are chronically missing. In Belgium, it remains at just under 30% of reported deaths.

In fairness, the number of deaths where the location of the death was unknown has dropped from a value in the 30's to 19. It took over 3 months to make that headway.


When 9 732 people have died in Belgium, maybe such omissions can be defended.

"À la guerre, comme à la guerre."


In our little corner of Belgium, we find them, with 2 728 records incomplete, totally inexcusable.


As Benjamin Franklin wrote: "Three can keep a secret, if two are dead."


In Belgium, two who did the job right live in Brussels and Wallonia, but the third is in Flanders and he's not talking. That person had and has many other things to do besides look under the bedsheets to see if the decedent was a woman or a man. It may also have required checking a box on a form: ___F  ___M.

Yet another imposition, when the patient is dead anyway. "Dead Men Tell No Tales" said the sign.


Mortality Summary with missing data values - 27 June, 2020


We've even found the culprits. For yesterday's incomplete records count of 2 728 deaths, one notes that coming out of the Flemish region, only 46% of deaths have complete data. We're not talking about subtleties like ABO Blood type or BSA of the decedent (which would be very good to know for certain living COVID-19 patients), but whether the decedent was a man or a woman, and about how old. Also missing, where they were located physically when they died. Real, basic, stuff.


Panneau sommaire - 27 juin


Next, as those doing the counting work with the mortality figures they have actually received proceed with their task, another special phenomenon has been observed.


Sometimes deaths just disappear from the roster.


We have called these "Les ressuscités" (those resurrected). On 27 June the summary panel looked like this. Five more deaths removed from the books yesterday, for a total of 100 such ressurections since 9 May. Up in thin smoke. Should definitely help the Case Fatality Ratio if it continues.


Les ressuscités - 27 juin



Nothing to get excited about.

It's really just epidemiologic book keeping. Happens all the time. 


There may even be a Belgian method to their madness. 


But for anyone interested in following the numbers, one should know that the shakiness is not only for terminal cases as noted above, but also with the incoming confirmed cases who are still alive. 


And as the numbers get smaller, and the Belgians begin to "open up" at the beach and local cafés, one often hears comparisons of yesterday's numbers with those just announced today at noon. They are a population dedicated to their government and the numbers it feeds them. (Government? Did I say government? What government? The one we voted for last elections? Ha!)


The Belgian population depends on it's government for direction (96% in one poll, follow government's directives daily in all this pandemic stuff). Not many revolutionaries, anarchists or activists on the horizon here. So deciding on that upcoming vacation to Croatia or Corsica or for an educational visit with the kids to a porc-packing plant in Germany next month, depends on careful consideration of "les chiffres clés" (the key numbers).


So just how firm are those numbers?

How many numbers are missing as cases get reported? How variable are these numbers or are they set in stone once published?

Here's a closer look ... On June 26:


(click to enlarge)

Case missing data - June 27


93 cases where the date of a positive test is unknown, 1394 where the region of the country is unknown, 195 cases with age missing. We've learned to see and accept such gaps from the reported deaths. 

Notice that the 61,106 cases, occupy only 14,388 records in the Excel file. Or, 4.25 cases per record.


That fact might look like this:

Good luck finding your personal record if you had a positive test on March 26th, as one example. The highlighted colum are the case numbers reported on a single line of the record.


Multicase records


Summarizing by region of Belgium, case numbers look like this

Let me help you with this: On the 26th of June, differences as noted between the total cases of COVID-19 presented to the public by region of Belgium, and the data in the Excel spreadsheet.

It's certainly up to them to control how the results emerge, which results emerge, and to whom, and where to place the emphasis. Maybe "SPF Santé publique" should be renamed "SPF Santé presse," or "SPF Santé média."


But perhaps more important for any subsequent study, and going beyond just serving government intentions, 3 to 5 cases presented on each record of the spreadsheet (that's on the left here below as the number of cases per record (c./ r.) ).



Cases by region and number of reports


Any subsequent analysis of results may have a real hard time taking these cases apart. That's at least one point. Other reasons to not use this method have been presented, and we'll spare the reader that exercise here.


Here is an opportunity for the conspiracists to chime in: "No surprises here. They are keeping a separate parallel record with all the data, line by line. And of course, you will never see that record."


"Will the conspiracist please sit down now? Thank you."


They did the same thing with death reports we saw above: Lumping more than one death on a line of the record.

It has taken just over 3 months to not get those sorted out. Not even close. Not a pinch of improvement in months. Until perhaps the last week. We'll see. The beat goes on.


But with cases, certainly the number of cases reported to the public, one might assume that they must match up perfectly with what is contained in their Excel spreadsheet.


If not, where are the numbers coming from? Which are the correct numbers?


Let's have a look ...

We'll look at cases as reported, and data from their Excel spreadsheet.


Let's go back to June 20th and the week preceding that date. That's a week or two before today's date of June 27 when this was written. Certainly that amount of time should permit the smoke of numbers "not yet consolidated" to clear. The data extracted from reports and the database look like this :


(click to enlarge image)


Excel vs reports


If one focuses on the two green lines, the row above are the numbers out of their Excel spreadsheet, and below, the numbers reported. They don't match up very well. On the 20th, no data on the spreadsheet for the 19th, since these data are not in yet, one supposes. 


Let's look at a specific day, June 15 :

Not only are the reported cases (71) different from those in the spreadsheet (148), but the 148 new cases occupy only 88 lines of the spreadsheet (1.68 c./ r.). So still, they are doing with the case records, just as they did with the death records. Making them, well ... inseparable forever. Just wonderful for subsequent study of what actually transpired in 2020 in Belgium. "Do they still teach History in school? No? Then we should be OK."


Excel vs reports and records


Let's just work for a moment with Total Confirmed Cases. Somehow that should provide an overall view with greater clarity. Less chance of getting lost in the details.


How many new cases are added, to get to the next day's value of Total Cases?

That depends if you are reading the values given in their Reports (and presented by the media), or in the Excel spreadsheet (which they are kind enough to share). In both cases, these are "their" numbers.


"Answer is?" : It doesn't match up perfectly. But should it? Here are the values for the June dates indicated.


Total cases - June 20th and before

If one calculates the % error between reported and Excel total cases, that error is small, since the numbers are large. So they slip by pretty easily. But should each pair of numbers be the same?


Many people who have been following along, might answer: "Yes. The numbers should be the same."


Why aren't they the same?

I don't know. Ask Sciensano. That's a pretty solid response. Ask them. It's a valid question.


But my hunch is that we are in a soup of Belgian numbers. Un potage fait maison.


I also am pretty convinced that the Chef keeps tinkering with the case numbers, just like with "Les Ressuscités" for the deaths that one day, just disappear from the soup, ... I mean, from the count. 


Some will defend that it's just how this job is done. You tinker with the numbers already presented as somehow definitive, and you change them. They're just not definitive yet and need some help getting there. Then you decide what to tell others. Or don't see the need at all.


But clearly, or not so clearly ...


One should not tie up too much attention and cognitive force on memorizing today's numbers when they come out at the next noon. That's a recommendation to our readers. 


Count on it. In the days to come, as they work up the 7-day Averages and compare with the previous 7-day Average of new cases, those numbers are just floating in the soup. Take a taste, but don't drink the whole bowl. And especially, don't ask for the recipe, since it changes every time they make the soup.


If you have your doubts about my suggestion or numbers presented, download their Excel spreadsheet today, and again in 5 days, and see how the case numbers for the same date  have been modified. 


Is it a random thing that now, no data are presented for Sundays and Mondays, getting back ino the swing of things on Tuesday? Is Tuesday a "what the hell" choice?

Probably not. Cases are always lowest on Tuesdays :


Cas totaux par jour de semaine - 27 juin


Any Tuesday is much more likely to be a "Looking Good!" day for all those listening attentively.


Of course, the monthly mortality figures can be glossed over in Tuesday's presentaion, since those numbers risk being less reassuring:


Décès par jour de semaine - 27 juin


"So you mean, it's a conspiracy?"

Almost certainly not, but who knows.


Much more likely, hideous inefficiency, further paralyzed by government preferences directed at a bunch of nice guys who usually just sit quietly at their desk and get their surveillance work done. Without any TV cameras rolling. Now they're being pushed on each day to "get some good numbers." So they do.



"What becomes of all these numbers?"

If one elects to calm one's nerves and avoid obsessing, the real importance lies of course in how the numbers are put into use. What is the message being sent to the public? Here's an example of that reality in Belgium :


Data taken from their daily or weekly report, presenting graphically the age and gender of those who died.


Deaths by age and sex (sans footnote)


One might have a quick look in passing at the above graph, and conclude that those over age 60 or so are mostly involved, and that males and females are almost equally at risk of dying of COVID-19. Depending on one's age, one might feel a lot better, or a lot worse. But wait! A  footnote has been removed from the graph (but added below) ...


That footnote informs us that for 2 724 deaths, the age or gender or both of the decedent is not even known.  Yet they divide up the pile by age and gender to present thir truths.


They're just not going to be stopped it seems.


Deaths by age and sex


So what is the right thing to do with this graph?

Choice #1: Release it to the media who can show it several times a day to the public? Perhaps not but has been happening daily.


Choice #2: A better choice might be to place it in the bottom drawer of one's desk, and spend one's time solving Belgium's problem (almost completely a Flemish problem) with getting simple case data (age, gender) reported correclty and promptly. Sounds like too much work with an unwelcoming audience.


Choice #3: Tinkering 3 months after the fact and still getting nowhere fast. Then publishing the same graph. That's about where we are today.


If the present misuse of data continues, who knows? Perhaps tomorrow, in response to a graphic like this one below, someone will announce on TV :


"We are happy to report that things in the world are now better, or will certainly be tomorrow."

"Time to plan that trip to Mexico after all. Tomorrow that curve should drop to zero or almost zero."


Weathermen of the pandemic.


World - Daily Cases - June 27, 2020


A final opinion :

It might be better for Belgium's epidemiologists (and the government that controls them) to focus on how to get rid of the lingering fat tail on the graphic below, and while there's still a chance to do so.

This pursued through any and all means available and necessary.

Naively put: this remains a very infectious disease. Belgium, work on your fat tail.


Nouveaux cas c tail

Implementing such known actions might be a much better pursuit, than getting the data right so a Minister of This Or That can announce to the people that they can all go on vacation and send kids back to school when they get back with their tans.


Such messages and plans for reopening while the tail is still fat, and the world is still sick and infectious, need a wiser approach.


Yesterday, the process of contact tracing in Belgium (which never actually got it together), was taken apart, leaving now only 40 or so contact tracers. They apparently didn't have enough work and some people refused to talk to them about their contacts. "Some people were rude with us. Just not enough to do..." they said to the Press. The government reacted: "Your fired! And you, and you, and you!"


Forget the word "preparedness." It's not currently in any of the 3 languages spoken in Belgium.


Instead, it's time for Belgium to decide to participate in uniting the world, not just in its little country with its anachronistically still divided 3 regions. Fat tail. Fat chance.


If the "experts" and government are still a little touchy about that 15.9% mortality rate, here is the cause of that statistical result. No matter how the statisticians turn on their heads to find excess deaths and compare with Z-Scores the mean rates in Europe and the world, Belgium's mortality rate is still way up there on the list: Belgium you started everything too late. But over time, you finally began to get it together without being showy. You are still continuing that proud and polite tradition of carefully thinking through all your options, until there is no longer any option left. It seems an entrenched tradition. Then announcing that you have already been doing the right thing since years and years.


While trying to keep Church and State separate here ...


Proverbs 16:18, which says “Pride goes before destruction, a haughty spirit before the fall.”




A word or two for the epidemiologists in the audience

Before you get the numbers just right and leave your offices to join in the festive and folkloric carnivals, get your mind right.


As Seneca wrote centuries ago ...


"Si non sit ignoranti quem portum navigantibus nullus suus ventus est."



Which port



No numbers can defend for very long, bad decisions. That includes a decision for inaction when right action is needed.

Be proud. Speak up. At least one person is listening.




<<<<<<<  Home



0 Poster un commentaire

Inscrivez-vous au blog

Soyez prévenu par email des prochaines mises à jour

Rejoignez les 13 autres membres