Population density - does it relate to prevalence of COVID-19 disease?

I'm sure you've been asking yourself that question for at least a month now.


Here is population density in the European Union


Population density in the EU by country (Jan 2017 estimate)


Here is the prevalence (again that's number of cases of an illness in a country, divided by the country's population, in millions) for the 24 countries we've kind of randomly decided to follow. They're values from yesterday.


Prevalences, ranked - 27 March, 2020

Now I'm going to tell you a very short story, and finish by presenting a hypothesis.



You get onto an elevator. You are alone and press (with your car key) the button for floor number 12.

You notice that you are slightly relieved to be alone in the elevator. You even mutter to yourself as you notice that sensation: "... never made much difference before."


On the way up, a stop at floor 6. The door opens and an attractive person gets in, says hello with a nod of the head, and turns like you to face the door. Around floor 8, she sneezes 3 loud, rapid, consecutive sneezes. Neither of you is wearing a mask. Your thinking, "Oh Jesus! Save me Jesus!" just as the other person says, just below a very wet nose, "... sorry 'bout zat." 


Next a stop at 11. Just one floor to go. 


Doors open. A guy in front of the elevator holds the door open and waves. 

Six people file in, and he's the seventh. 

Two are blowing noses as they get on. Two are coughing.


One is doing a sneezando duet with the second occupant from floor 6.


Quick ride up to the stop at 12. You have been holding your breath since the doors closed at 11.

From the back corner you mutter between puckered lips "'... scuze me." You put your hands together held belly button high, fingers pointing forwards and palms together, kind of like praying, but to indicate that you'd like a safe pathway out of this Petri dish.


You're out. You're saved. You hear the laughs, sneezes and coughs as the door closes.


You regroup and think : "If I not sick and at death's door in about five days, it's a miracle."


That of course, is a hypothesis

It would seem to make sense of course that, close to sick people, we get sick too.

Sort of a universal truth.

But what about at a more global level?


Do countries where people live all jammed together, have more of this current viral nuissance?


How would one go about answering that ?

Locally, we could email the guy who got off the elevator at 12, in about a week or 10 days, and see how he's feeling.


Globally, how about this?


Let's gather population densities in the 24 countries that I've been following. Those data can be accrued without travelling. Will use units of "people/ km²." One could do that in sq. miles, but using the metric system is a sign of true intellectual prowess, and an unflinching devotion to logic.


Let's put those data in a table, and sort it. 


Population density sorted - 28 March


So a country like Belgium, with a population density of 383 people per square kilometer, is pretty densely packed. 

But does it have more positive COVID-19 cases than Russia, for instance, where only 9 people live out on each km² of tundra?

Let's see: Belgium - 682.5 cases per million population. Russia? 7.1 cases per km².

Great. We're onto something.



I've really scared myself with that, and I'll be right back after a trip to the john.



Better now.

But that isolated comparison may not hold if we look for a correlation between these two variables for all of these 24 countries. Let's try that.


Let's start by sorting the countries with the highest prevalence of this disease at the top, and bring along the  population density for each. Like this. The values for prevalence (cases per 1 million population), are today's values, 28 March.


Sorted by prevalence this day - 28 March


So clearly, a big sneeze in Belgium, ... (let's pick a different country)


So clearly, a big sneeze in the Netherlands is going to create more cases of infectious disease.

It's just got to.

Correlation with confirmed cases - 28 March

Then again, maybe not. There is just a horrible linear correlation (or non-correlation) coefficeint of R² = 0.014456. Fourteen chances in 1,000, that if you give me a country's population density, I'll be able to tell you it's prevalence of disease. Well that's pretty poor. And the data points are not neatly aligned in a line. They're all over the place. Or, it means that a population squeezed together doesn't have more cases because of that squeeze, even if they have lots of cases.


Hard to believe but true.

Maybe we're looking at the wrong variable.

Maybe by including all the cases, we're not really finding the truth.



Let's look at death rates. Those people were really quite sick with this viral thing. Death's must be higher if people who are living in close quarters.


That (below) doesn't look very good either. Data pairs all over the place, nothing that looks linear, and a near total lack of correlation (R² = 0.014456). 

Correlation with deaths - 28 March


Oh well.

Wait a minute. Shouldn't that be a death rate instead of actual deaths? Sure. Deaths divided by confirmed cases. What percent died by country? I'll bet that that has got to be higher in places cramped for space. Instead of deaths, death rate.


Correlation with death rate - 28 March

Nope. That's not it either. Data pair points all over the place, and an essentially non-existent coefficient of correlation.


So far, variations in population density,

don't seem to play a role in common parameters associated with viral illnesses.


Last try. What about those who have recovered? Instead of viewing it as a risk or impediment, perhaps having more people around you, lends some restorative support. You get better more quickly. For psychosocial reasons. Perhaps.

Correlation with Recovered - 28 March

I really thought we had something going there. But it just didn't pan out. No correlation, is the right correlation or conclusion here.


The hypothesis that countries with high population densities must propagate this viral disease more than those with wide open spaces, is not sustained. And yet : social distancing.


Based on these comparisons that idea, or hypothesis, has to be rejected. 

Why did we get into this ? To explore that :


    1. Intuition is at times wrong.
    2. What we were taught to take as gospel may not always be true.
    3. Science involves sometimes looking around, and falling on a solution by chance. But as Louis Pasteur said: "Chance favors the prepared mind." Not to compare myself with Louis, but I once said in response to a question about direction in my research: "If we knew what we were doing, it wouldn't be research." I think that a lot of what is vital to understanding the propagation of this coronavirus, falls into this category. We just don't know. Not yet. Trial and error is being applied in many countries. It is not a bad method, but takes time. Taking notes, sharing observations, sharing mini-conclusions like our little study here of the effect of population density, are all important.
    4. Such a little free-for-all of "what-if" data comparisons is healthy.
    5. One must not forget what is already known about this virus and its behavior at a basic science level. Today, basic principles will be increasingly translated through amazing tools of genetics, virology, and molecular biology. These findings will then need to be scaled up, as with vaccine production.  But thoughts of the "what-if" type and powers of careful observation should never be too far distant.




<<<<<<<  Home



0 Poster un commentaire

Inscrivez-vous au blog

Soyez prévenu par email des prochaines mises à jour

Rejoignez les 5 autres membres