Effect of medical infrastructure and testing on COVID-19 death toll

What's next?


Your contact at diconium

Geethasri Ramkumar
intermediate consultant data | analytics

The COVID-19 outbreak has shaken the base of even the biggest economic powerhouses in the world and brought the whole economic machine to a complete standstill. Every country around the globe is dealing with this and taking measures to stop the spread in their own way. One of the most harrowing aspects of this outbreak is the number of deaths reported on a day to day basis worldwide and it deserves to be dissected and analyzed. The country with the highest number of confirmed cases doesn’t really mean that it has the highest number of deaths and vice versa.

The number of deaths in a country is influenced by many factors like medical infrastructure, the number of tests conducted to identify infections early, the age of the population, etc. It would be interesting to look at the top six countries affected by the virus at the moment namely the USA, Spain, Italy, France, Germany, UK and understand the influence of the number of new cases every day on the number of deaths every day in each of these countries.

The dataset taken from Kaggle can be used to analyze this factor and it contains data till 2nd May 2020.

Effect of medical infrastructure and testing on COVID-19 death toll

The demographic of the country with respect to the average age of the population could increase or decrease the fatalities as the virus has been known to be extremely dangerous to the elderly above the age of 50. Beyond this, the existing medical infrastructure along with the testing capacity of a country for early detection could prove to be the deciding factor in determining the fatalities for that country. What makes this particularly important is the fact that some countries may have a similar surge in confirmed cases over time and yet differ drastically in the number of deaths overall. This is mostly due to the above-mentioned factors which vary from country to country and certainly not something that can be built overnight.

A country always needs to have the resources (financial as well as medical) ready, with or without a pandemic outbreak.

A country’s infrastructure to handle this kind of outbreak could be stretched to its limit, causing it to rupture, resulting in a surge in the number of deaths. This is because the hospitals are overloaded and overwhelmed, lifesaving equipment is limited in availability and the number of tests conducted to identify potential infections is not enough to catch up with the rate of spread. Even when the rate of increase in daily new cases is similar for two countries, the influence of the number of new cases on the number of deaths could still significantly vary. For each country, this influence could be quantified to an extent using regression techniques to determine an influence factor – A number which represents the influence the daily addition of new cases has on the number of eventual deaths.

Table (1) shows the influence factor of the number of daily new cases for each country on the number of new deaths.
Table 1: Characteristics of each attribution model category Table 1: Influence factor of the number of new cases on the number of deaths

The higher the influence factor, the higher the impact on the daily new deaths. The table clearly shows that Germany with a factor of 0.039 has managed to keep the influence of new daily cases on new deaths comparatively low. According to Real estimates of mortality following COVID-19 infection, it takes an average of two weeks from the onset of symptoms to death. This two-week period is considered in the calculation of the influence factors and the assumption that, cases registered today would have an impact on the number of deaths two weeks later is to be noted. Countries like Germany that have a lower influence factor despite the higher number of confirmed cases, have the necessary or better wherewithal to tackle the surge in infections.

Whereas in countries like Italy, Spain, France, etc. where the factor is comparatively higher, the daily new cases have a higher impact on the deaths, as the health infrastructure and capacity available were breached much quicker in these countries.
Table 2: Attribution models implementation specifications Figure 1: Portrayal of the number of deaths in the USA if it had German infrastructure
Figure (1) shows what the scenarios could have been with respect to the number of daily deaths in the USA, if it had a similar infrastructure to Germany.

The figure shows that the number of daily deaths (red line) would have been a lot less than it is. The numbers indicated in the plot cannot be used as it is for any kind of analysis, but it helps to understand the importance of infrastructure readiness.

So far, the impact of new cases on the death toll has been discussed for different countries with different levels of preparedness for a pandemic of this scale. Besides this, another important aspect of the pandemic which varies drastically between countries is the mortality rate. Comprehensive testing not only enables spotting cases ahead of time and thus reducing death toll, it also aids in building an accurate picture of the mortality rate in a country. There are two types of mortality rate – case mortality rate and infection mortality rate. Case mortality rate is the ratio of number of deaths to the number of confirmed or registered cases whereas infection mortality rate is the ratio of number of deaths to the number of actual infected cases. The infection mortality rate is not easy to calculate as the number of actual infected cases is not readily available in any country. In most of the countries, people displaying only mild symptoms are asked to self-quarantine themselves and a lot of these cases are not registered. The decision to register such cases, is the prerogative of the local city officials/doctors and this makes it even more difficult to arrive upon a value for the actual number of infected.

This brings into picture the other major effect of comprehensive testing during this pandemic. Countries which test a lot of people irrespective of the seriousness of the case, can keep the number of registered cases as close as possible to the actual number of cases and tend to show lower case mortality rate. Countries which register only the hospitalized cases tend to show higher case mortality rate and this can be misleading especially when directly comparing case mortality rates between countries. Figure (2) shows the evolution of case mortality rates among the top six countries. Germany has the lowest case mortality rate among the top six and shows a stark difference from the other affected countries. This could be due to the massive amounts of tests conducted even on cases with mild symptoms. It is also worth noting that, countries like UK which show very high case mortality rates, mostly register cases that are hospitalized, and the real mortality rate could be lesser than what is recorded.

Table 2: Attribution models implementation specifications Figure 2: Evolution of mortality rates over time
Does testing capacity always have an influence in lowering the death toll?

Utilizing the data containing total tests conducted per thousand people in each country over time, the table (2) shows the most recent number of tests per thousand and corresponding case mortality rate. Germany which has the second highest tests per thousand has the lowest case mortality rate and countries like France and UK, which have lower tests per thousand have higher case mortality rates. But Italy which has the highest tests per thousand currently, also has considerably higher case mortality rate. On the other hand, USA which is fourth in the number of tests per thousand has a low case mortality rate.

This shows that, although increased testing can be used to present a clearer picture of the real mortality rate, it alone cannot be used as an indicator for the death toll and hence other infrastructure elements like hospital beds, staff, equipment, etc. also determine the death toll. Nevertheless, testing would assist in isolating infected cases quicker, thus bringing down the rate of spread.

Table 2: Attribution models implementation specifications Table 2: Current tests per thousand people and corresponding case mortality rate

Countries that lack the necessary levels of infrastructure and readiness cannot concentrate on curing the problem. It is paramount that in these countries, prevention of this spread is prioritized to reduce the number of daily new cases which in turn reduces the number of deaths. Moreover, thorough testing and registering all the cases, mild or serious, facilitates a better cognizance of the true effects of the virus. Having said that, there is considerable amount of effect of the number of new daily cases on daily new deaths irrespective of testing capacity in some countries due to the state of its existing medical infrastructure, shown above through the influence factors.

Without any doubt, the target should be to minimize the pressure on the existing health infrastructure, and this is indispensable in countries where the infrastructure is comparatively brittle, especially in developing countries.

diconium is a specialist in the fields of data and artificial intelligence. Whether in the areas of search, social and content, personalization and analytics or data science – our expertise helps you to collect the right data at the right time, to forecast services and offers and thus to make data-driven decisions. We look forward to discuss with you about your specific use case. 

Learn more about our Data & AI portfolio and our approach.

What's next?


Your contact at diconium

Geethasri Ramkumar
intermediate consultant data | analytics

That's us

Since 1995, we have supported industry leaders such as VW, Mercedes-Benz, Bosch, Allianz, Kodak Moments and Sick in exploiting the potential of digital transformation and of-fering millions of customers extraordinary experiences. As a strong partner, we accompany companies through the entire digitization process: from Innovation & Strategy, UX, Data & AI, Commerce and Technology to Building of Digital Units.

One for all, all for one! More than 1.000 colleagues serve our clients from 11 locations worldwide: Stuttgart, Berlin, Hamburg and Karlsruhe as well as Bangalore, Detroit, Beijing, Lisbon, London, San José and UAE. What unites us? The courage to break new ground, mutual attentiveness, teamwork and the motivation to make a difference and create sustainable impact. diconium is a 100 percent subsidiary of the Volkswagen Group.