Planning for medical resources, population density should be taken into account to prevent the mortality of future pandemics
I read both NYT articles, this and this, regarding the Covid-19 death tolls in Turkey and elsewhere. I was making similar observations and charting the excess death toll for Istanbul since March 1.
I also read my dear friend Nihat Ö. Ayhan’s analysis about this issue: Excess Deaths: Are the Governments cheating on us about the Covid-19 related death toll?
I disagreed with his conclusion and explained my rationale here in detail:
While down that rabbit hole, I started researching an interpretation of Metcalfe’s Law for pandemics or epidemiology in general. Surprisingly, there weren’t any simple or sophisticated academic theories trying to make use of such a network analogy.
So I started researching scholarly articles for “population density” and “epidemics”. The first result was bulls’ eye: “Effect of population density on epidemics”. (Ruiqi Li, Peter Richmond, Bertrand Roehner. Effect of population density on epidemics. Physica A: Statistical Mechanics and its Applications, Elsevier, 2018, 510, pp.713–724. ffhal-01905736f)
The paper was referring to a KMK Model from 1927:
“KMK (KermackMcKendrick, 1927) model predicts that the size of the epidemic increases strongly — and in a non linear way — with the initial density of susceptibles.”
This, I think, was pretty similar to Metcalfe’s Law adapted to a pandemic context:
“The impact of the epidemic for any geographical network is proportional to the square of the number of connected users, nodes in the system (n²).”
The paper presents a table showing the impact of population density on the death rate in US states from 1915 to 1918.
New York City & Istanbul: Similar Contagiousness
This table teased me to compare the population density and its impact on the relative propagation of the local epidemic in both cities. The population densities for Istanbul & New York City are respectively 2,904 ppl/km² and 10,908 ppl/km². Although 1.8 times less populated , New York City is 3.8 times denser than Istanbul.
According the above table, all other parameters being equal, one can hypothesize that the propagation of Covid-19 in NYC would be (10908^0.44)/(2904^0.44) = 1.79 times worse than Istanbul.
Edit: In the previous population normalization, I’ve been tricked by a hasty Google search result from Macrotrends.net. The population of NYC (all five boroughs) is indeed 8.55M and not 18.8M.
So the proper population normalization and the conclusion is completely different. If NYC had the same population as Istanbul, it would have 262,160 cases instead of 144,190, as of April 22. [So we start by normalizing New York City & Istanbul populations. If NYC had the same population as Istanbul, it would have 119,165 cases instead of 144,190, as of April 22.]
Based upon an official report dating April 12, assuming that 60% of Turkey’s total Covid-19 cases are still in Istanbul, Istanbul should have a total of 59.204 cases, as of April 22.
New York City is 7 days ahead of Istanbul, as their first case was reported on March 3. In 7 days, by April 29, Istanbul is expected to have a total of 66.3K to 71.6K Covid-19 cases.
Normalizing New York City to Istanbul population density with the above multiplier [1.79], Istanbul should have at least 146.4K cases (262,160/1.79) at a synced stage as New York City. By a factor of 2, this is way beyond the expected range of 66.3K to 71.6K cases, on April 29. Hence, this may explain the low case fatality rate at 2.4% (2,259/95,591), contradicting with the excess death toll estimation (4,883) extrapolated from Istanbul Municipality death stats (2,902) instead of the official 2,259 (x2.05). So, projecting from excess deaths we can fairly estimate a CFR of 4.8% for Turkey as well.
So, the fact that NYC failed with a CFR of 10% and Istanbul did not with a real CFR of 4.8% still slightly under the threshold and had the chance to politically mask it even at 2.4%, may be attributed to population density.
Fighting With Pandemics, Population Density Is Important
“To prevent future pandemics to crash the health system, metropolitan cities with higher population densities may need to plan their medical services capacity based on both population and population density, not only based on population alone.”
This is an overly simplified and definitely non-academic hypothesis and we have to test this against other metropolitan cities and countries around the world as well. But to summarize for now, factoring both population and population density into account New York City & Istanbul case occurrence behaviors seem to be quite similar. So this approach kind of validates the official case figures in Turkey, at least as accurate as New York City.
To prevent future pandemics to crash the health system, metropolitan cities with higher population densities may need to plan their medical services capacity based on both population and population density, not only based on population alone, if they’re not already doing so.
Yet global comparative stats for medical staff, hospital beds, ICU’s and other relevant resources are mostly based on per capita figures. This a sign that the population density is not currently taken into account as a criteria.
