The Secrets of the Pyramid - Covid and Demographics

“Le Secret de la Pyramide” : This is the French title of movie “Young Sherlock Holmes”, and it is true that even after 9 months of data crunching on Covid public data analytics which helps understand the dynamics of the epidemic, I feel there are a lot of secrets we still need to decipher about Covid. Here are some insights all based on public data analytics.

Age and existing conditions are, we all know, key factors driving the severity of Covid-19 symptoms.

However, we have very little cross-country analysis on these risk factors provided in the popular apps reporting Covid metrics. In my research work with the help of volunteers For Solidarité Covid – Expats in China, we have been striving to benchmark these metrics across countries,

For the comorbidities, we have a few benchmarks, however there are many caveats : the definitions and list of conditions, the availability or reliability of cases by condition, the scope of patients for whom we have the data (ICU cases in France until June, census hospitalized for Brazil with many not reported, only deaths in Italy)… we made the best we could with these type of outputs which already tells a lot…

The age factor is subject to less definition related caveats and critical in the analysis, provided they are properly reported for all cases and deaths, which is another challenge for many countries. Therefore, back in April we started collecting data on ages on various country web sites with the support of native friends from each country. Some countries only report the ages of the deaths, but we are interested also in the ages of cases as it has a direct impact on the nb of deaths which is the Key Performance Indicator most reliable across countries which have / had different guidelines in terms of testing / Hospital admission / ICU reporting.

Over the summer we wanted to extend our ages of cases / deaths benchmark to more countries and we ran (with the help of Valérie Novi-Santoro) into COVerAGE-DB, a project led by Max Planke Demography Institute in Germany and with many contributors around the world.

COVerAGE-DB is a database of COVID-19 confirmed cases and deaths as reported by statistical agencies, standardized and in harmonized age groups, referenced by INED, covering 90 countries, exactly what we needed ! The data process was not easy to get to the ouput on CovidFlow as we filtered to total country data from the massive Open Source file, filtered again to the latest data point for each month, complemented with some data entry / process we already had implemented (Brazil Hospital Open source, France data including the Elderly care facilities breakdown of Hong Kong between imported and locally acquired). I was delighted to exchange with Tim Riffe, one of the project leaders to confirm that CovidFlow was the first dashboard regularly updated produced with this data set, and also the first to decumulate the data by month so we can have the distribution on monthly cases and deaths by age group. The pyramids shape up here, the Maya ones from Mexico on this example, unfortunately no demographics data for Egypt…

This data set is providing us a lot of insights :

1) In Asia, the share of elderly in cases is lower than its weight in population : Elderly are better protected in Asia, mainly because they live with their children and Elderly care facilities are not so common. They also have less Cardiovascular / Obesity prevalence than in Western countries

Visa Versa, In several Western countries, we get a higher proportion of cases among elderly vs Population : an impact of the Elderly care clusters

2) over lethality of males due to larger prevalence of cardiovascular comorbidities inc obesity which are particularly vulnerable to Covid. Another factor is the large scale testing in elderly care facilities which generates a lot of positive asymptomatic or pauci-symptomatic cases among women (as there are more women in those facilities with other conditions than the Cardiovascular / Obesity diseases most vulnerable to Covid).

3) We can also see that some countries in developing countries sporadically present a high lethality rate (above 10%) on the 50-70 age group (Mexico, Ecuador, Brazil and some countries in Africa, and only Romania in Europe).

These benchmarks also have their bias, of course the testing capacity, the sensitivity of the test (number of rounds of amplification), and the countries who report those demographics or not. See below how Africa is low in testing vs the nb of cases.

The lethality in hospital will be also dependent on the level of saturation of the ICU. Unfortunately, we don’t have any longitudinal patient data (except for Brazil) so we can only compute a cumulated lethality ratio which we measure at each month-end time point. In France It has dropped month by month after the first episode, however for those areas where the ICU has been close to saturation, we can observe a rise in the cumulated lethality, specifically in the 70Y+ age group, the age group in lower priority of higher risk to go to ICU.

The black box of elderly care facilities

Ageing population, more comorbidities and clusters around Elderly care facilities are the key drivers of the heavy toll the Western countries are paying to the pandemic.

Starting with the metrics on Elderly care facilities, with this visual, outlining painfully how much these institutions, key players for old age and dependance in Western countries, have paid a heavy toll to the pandemic.

But there is more to tell when we go into the demographics. In France, the official data by age group doesn’t not include early cases (March to May) and any deaths in the Elderly care Facilities and the hospital data doesn’t cross age and gender so we’ve had to make some assumptions (based on demographics of residents source www.lesmaisonsderetraite.fr + www.doctossimo.fr) to come to these Lethality ratios Hospital only and including cases and deaths in Elderly care facilities.

On all age groups, we can observe a big drop between with or without Long Term care for women. That is because there is now regular massive testing in the Elderly care facilities, a larger proportion of women in these facilities, better anticipation on how to treat the patients, hence a big drop of mortality, specifically in the 80-89Y group. A lot of positive cases but a large proportion can be asymptomatic as an Italian report shows up to 50% of elderly can be asymptomatic.

Now looking at the gap between men and women mortality on the elderly across Western countries, we find the gap in all countries, even though we have evidence that Elderly care Facilities deaths in the spring may not be included in the official figures for Spain and Italy. In these countries those spring deaths were estimated through a nationwide survey submitted to the Facilities so as to report the Covid suspicious deaths. 

We have looked into these surveys and made some estimates of the extra spring deaths, which we reported in CovidFlow, field Extra deaths : 8175 until end April in Italy, 15815 until end May in Spain. So a next step would be to confirm these extra deaths, whether they were ever included in the Covid statistics, and apply some demographics assumptions the same way we did for France. For France, when the tide receeds, upon request of the University of Strasbourg associated in the COVerAGE-DB project, we hope to get some input of the actual demographics from ARS Grand Est (Agence Régionale de santé) and use these as a proxy for all France. Next time !

This concludes the series of articles about the exclusive analytics we have developed as volunteers into the “Solidarité Covid – Expats in China” project. It’s about 6 hours of daily work to update the data sets and produce the analytics just on my end, plus the precious contribution of volunteers Matthieu Bouquet, Laetitia Bernard Granger, Meiping Xue, Valérie Novi Santoro, Ma?l Deschamps, and the input of COVerAGE-DG Tim Riffe. Many thanks to all of them ! There are certainly plenty of other Tutankhamon treasures in Open Source to explore, but not sure we can absorb more workload without more volunteers (there is no multi handed Shiva in the Egyptian Pantheon !) or without the support of a proper sponsor eager to help our innovative research project on the dynamics of the epidemic. Please reach out !

Please check the other articles on the other two Exclusive Analytics of Solidarity Covid - Expats in China project :

https://www.dhirubhai.net/pulse/china-covid-post-wuhan-tiny-numbers-mayhem-metrics-sources-gabay/

https://www.dhirubhai.net/pulse/covid-19-brazil-icu-france-age-group-china-latest-analytical-gabay-1c/

Please follow our work on this public free access (and low cost!) website :

https://deeperin19coviddata.wordpress.com/

If you have wechat and wish to receive the daily analysis for China + bonus, please add me ID carolegabay