APBN New Site

APBN Developing Site

The Tricky COVID-19 Contagion – Deaths Comparison: Worldwide and Italy Analysis

A deeper look into Italy’s contagion data with Dr. Galileo Violini.

A few days ago, several newspapers registered a declaration of one of the most respected Italian epidemiologists suggesting that in Italy the number of infected people is underestimated. The suggestion followed from the assumption that in general there should be about one death every 500–1,000 infected.1 Actually, this estimate that he made also some months ago2 is valid only in a few cases in Bhutan, Faroe Islands, Laos, Singapore, and approximately in some countries of Arabic Gulf, like Bahrain, the United Arab Emirates, and Qatar and, in Europe, Iceland.3

The world data since the beginning of the pandemic indicate a lower value for that ratio, only 1 to 50.3 Since this data covers periods of larger lethality, and the availability of vaccines had the indirect effect of reducing lethality, one can expect a lower ratio, but not as small as 0.001 or 0.002 except under special conditions.

In principle, deaths and contagions are among the potentially most reliable indicators of the spread of the infection, and perhaps the death number is a very useful indicator. However, their estimate is relatively uncertain for many reasons that may affect the comparison in either direction.

From the very beginning of the pandemic, it has been pointed out that deaths have been largely underestimated4 because of those not attributed to the specific cause or because of inefficient registration (in particular in some developing countries). Therefore, probably an estimate of the contagions based on the deaths is higher than the real value by 25–40 per cent. On the other hand, this effect is partially compensated by the contagions that escape identification because of the protocol used to test potentially infected people and of missed asymptomatic cases.

To some extent, because of this compensation, an analysis based on the use of the officially registered data may provide reasonable qualitative results.

At the beginning of the outbreak, when the problem of identifying asymptomatic cases was the object of much interest, it was suggested to compare what happens in several countries, under the simplifying assumption that one could expect a limited variability when several countries are considered.5

This criterion is reasonably solid when applied to countries with similar characteristics of health security systems and mitigation approaches, to which one can now add similar vaccination policies. Thus, if severe discrepancies from the average emerge, they may be interpreted as an indication of data pitfalls. In particular, a larger than the average ratio between deaths and contagions could suggest that contagions are underestimated.

This was basically the point made by Professor Crisanti,1,2 when he suggested that the about 40 deaths per day registered in Italy were not compatible with the number of detected cases.

In order to verify whether Italian contagions data are really underestimated, we first checked the validity of the suggested ratio of 1 to 1,000. For this, we looked at all the countries having more than 1,000,000 accumulated contagions and the Latin American ones with above 300,000 accumulated contagions, and compared the number of deaths between October 6th and October 13th with that of new contagions between September 23rd and September 30th3. Whereas the choice of the period for the deaths does not require a special justification, that chosen for the contagion comparison may, to some extent, look arbitrary. However, it is very reasonable, in view of the typical times of incubation and development of a sickness leading to a fatal conclusion.

Since we are specifically interested in the Italian case, we also carried out the same comparison at a local level.6 We remind that the administrative division of Italy is in twenty regions, one of which, Trentino-Alto Adige, is, for this purpose, subdivided into the two provinces of Bolzano and Trento.

Our analysis shows that the variability of the ratio R obtained dividing deaths by the contagions of the indicated periods, multiplying the result by 1000, is relatively large, but, in any case, a frequent value R ≈ 0.001 is out of the question. In order to identify geographical and social regularity patterns, this variability makes it convenient to present the results in different tables, based on the R-value.

We considered it useful to show as well the value of R calculated from the accumulated data throughout the pandemic. As we shall see, usually the accumulated values are consistently larger than the October values. This is probably a consequence of the vaccination policies of the different countries. To check this hypothesis, we also display, for each group of countries, the average value RD of the ratio between the total number of doses administered in a country (irrespective of being first, second or booster dose) and its population. Even if there is a certain variability between different countries, even in the same group, our hypothesis seems to be qualitatively confirmed.

The average value of the October R for the countries of these four groups is 8.8. The accumulated average value is about twice larger (17.7) and the difference seems to be related to the country’s vaccination campaigns, being larger for the countries where it started earlier or was more effective.

For what concerns Italy, a detailed regional analysis will be made below, but already this country comparison does not seem to support a large underestimate of the contagions.

In the remaining eleven countries we analysed, the value of R is larger and, for the countries listed in Table 5, this might be just an indication of a possible underestimate of the contagions whereas for the remaining eight countries of Table 6, contagions seem to be really underestimated.

The results of Tables 5 and 6 mainly point to countries where the problem was already known, but one should not overlook the case of South Africa and some specific comment is deserved.

South Africa is the only African country considered in this analysis, because of our criterion of selecting countries above a relatively high contagion threshold. However, this criterion may be totally inadequate for Africa. Last week, the WHO regional director for Africa, Matshidiso Moeti, warned that it can be estimated that in Africa only one case out of seven is detected.7 If this estimate is correct, the total number of infections in Africa would be in the order of 59 million, which would have implied that several African countries should have appeared in our list.

There is no need to add much comment about the low level of vaccination in that continent. It would be tedious to recall the rich countries statements that no country would be left behind, like those of about one year ago in Abu Dhabi,8 but certainly the further reduction of vaccinations for African countries commented a few days ago by Ms. Moeti9 should worry the world, because of the possible appearance of new variants of the virus.

Returning to the case of Italy, the following table presents the same data for each region and for the provinces of Bolzano and Trento. The list is made according to increasing October R values. For each region, we also display the percentage, D/P, of administered doses with respect to the population and the ranking of the region with respect to D/P.6

The improvement of the ratio is general. There is only one region, Sardinia, where the current value of R is larger than the accumulated one and this might justify more testing as suggested in ref. 1. However, this effect cannot be ascribed to vaccination, because Sardinia’s vaccination rate is above the average national rate, and one of the regions with a higher vaccination rate. Indeed, in Italy, there does not appear to be a connection between R and vaccination, possibly for two reasons, the relatively high vaccination rate and its uniformity over the national territory. This makes unlikely a consistent and uniform underestimate of the contagions. [APBN]


  1. La Gazzetta del Sud, October 11, 2021
  2. Quifinanza.it, June 19, 2021
  3. https://coronavirus.politologue.com/ accessed October 17, 2021
  4. G. Violini, B. Pirouz, B. Pirouz, G. Medrano Asensio, Las muertes en exceso son debidas al COVID-19, acento.com, December 16, 2020
  5. B. Pirouz, G. Violini, A novel approach to evaluate the COVID-19 contagions. Asia-Pacific Biotech News, April 30, 2020
  6. https://lab.gedidigital.it/gedi-visual/2020/coronavirus-i-contagi-in-italia/, accessed October 17, 2021
  7. https://www.reforma.com/aplicacioneslibre/preacceso/articulo/default.aspx?__rval=1&urlredirect=/detectan-solo-1-de-cada-7-casos-de-covid-en-africa/ar2277909
  8. G. Violini, Nadie está a salvo de la pandemia hasta que todos estén a salvo, El Espectador, January 31, 2021
  9. C. Anna, WHO says Africa’s already thin vaccine supply to drop by 25%, AP News, September 9, 2021

About the Author

Galileo Violini

Former professor of Theoretical Physics (University of Rome La Sapienza and University of Calabria); Author of 250 publications (science policy, high-energy phenomenology, physics teaching, epidemiology) and, with N. Queen, of “Dispersion theory in high-energy physics”; Co-founder and Emeritus Director, Centro Internacional de Fisica, Bogota, Colombia; John Wheatley Award of the American Physical Society (APS); Spirit of Abdus Salam Award of ICTP and Salam-Family; Former UNESCO’s Representative in Iran; Former Director of an EU program at Universidad de El Salvador; Advisor of several Latin American countries; APS Fellow and Former Member-at-large of the APS Forum of International Physics; Doctorate honoris causa, Ricardo Palma University, Lima.