COVID-19 - fatality rate the-state-of-the-art

2020-08-04

(polska wersja)

  • COVID-19 fatality rates are a non-linear function of age

  • based on reported deaths and diagnosed cases, only case fatality rate (CFR) and population fatality rate (PFR) can be determined

  • the estimated infection fatality rate (IFR) differs from case fatality rate (CFR) at least 10 times

  • the fatality rates difference for different age groups is four orders of magnitude

The estimation of SARS-CoV-2 Infection Fatality Rate (IFR) remains one of the most important questions of the COVID-19 epidemic. Commonly available data on the number of cases and deaths do not allow for the determination of the IFR - the confirmed cases are only a small percentage of those actually infected. On the basis of confirmed cases we can only determine the Case Fatality Rate (CFR), which determines the number of deaths in relation to the diagnosed and confirmed patients and the Population Fatality Rate (PFR), which, in turn. determines the number of deaths in relation to population size.

In order to determine the Infection Fatality Rate, it is necessary to estimate the number of people who have been infected with SARS-CoV-2 as accurately as possible. It is worth noting at this point that it is practically impossible to directly determine this number, but it can be done, for example, based on statistical analyses of mortality or, above all, based on serological tests, which allow to detect the presence of SARS-CoV-2 specific antibodies. Such tests make it possible to determine if a given person has been in contact with the pathogen, in this case the virus. More precisely it checks if that person's immune system has produced antibodies characteristic of that virus. The presence of such antibodies confirms the infection, which does not necessarily mean that such a person has developed a fully-blown infection or disease, in this case COVID-19.

Since the beginning of the epidemic serological tests have been carried out in many countries. In recent months, the results of two extensive population tests for SARS-CoV-2 antibodies from Spain and Switzerland were published in The Lancet. The study in Spain was carried out between 27 April and 11 May, and 61 075 people were tested, the result being ca. 5% of them tested positive for SARS-CoV-2-specific antibodies. The study was very extensive. It covered the whole of Spain and the cohort (group of people) tested was demographically representative of this country. A similar study was carried out in Geneva, a canton in Switzerland. This canton is the most affected region by the COVID-19 epidemic in Switzerland. In this study the cohort of 2766 people were tested in 12 weekly series from 6 April to 9 May, and SARS-CoV-2-specific antibodies were detected in about 11% of people.

It is worth noting here that the absence of antibodies is not, in the case of SARS-CoV-2, a clear indication of the absence of infection. Researchers from the University Hospital in Strasbourg and the Karolinska Institute in Stockholm indicate that a significant proportion of people suffering from COVID-19 may not be seroconverted and thus may not produce antibodies characteristic for SARS-CoV-2. For such cases the immune response is based on T-cell. Many other studies indicate the key role of T-cell in the immune response against SARS-CoV-2, including the cross-reaction of cells characteristic of other coronaviruses. Further studies, relevant in the context of fatality, indicate that the number of antibodies specific for SARS-CoV-2 is clearly decreasing during the first few months in individuals with relatively benign COVID-19. It can therefore be presumed that many people who have contracted a SARS-CoV-2 infection have not produced any or only the amount of SARS-CoV-2-specific antibodies that would allow them to be detected in the tests, especially if they were performed many weeks after the infection. The mechanisms mentioned above may explain the relatively low degree of seroprevalence (number of people in the population with a positive test result for SARS-CoV-2-specific antibodies) in studies in Spain. Considering the number of deaths in this country in relation to the population size, the epidemic seems to be at a more advanced stage there than these studies have shown.

All fatality measures for SARS-CoV-2 are strongly age-dependent. In the plot below (based on data for Lombardy, the whole of Italy, Spain and Sweden) they take the form of exponential functions. The CFR and PFR rates are clearly in the upper and lower limits of the range in which IFR must be contained (red-dashed area) respectively:

The plot above shows CFR values for medics in Italy and Spain as well. They are significantly lower than CFR for the general population. Assuming that people which have existing health conditions do not work with COVID-19 cases, we can narrow the range of IFR for younger groups of population, especially for those who are not demonstrably affected by associated conditions.

The average IFR values determined on the basis of population tests carried out in Stockholm, Milan, Bergamo and Krakow are shown in the plot as crosses. The IFR calculated based on the studies in Spain and Switzerland are marked in bold line with crosses. The grey dashed line with crosses shows the IFR calculated based on statistical analyses of mortality in Italy. As can be seen, virtually all estimated IFR values fall within the hatched area between CFR and PFR.

This plot shows how much COVID-19 fatality is age dependent. Giving it as a single number expressed as percentage makes no other sense than presenting pure statistical. It is also clear how much influence on fatality is exerted by coexisting diseases and other burdens. In age groups from 20 to 69, the fatality for non-risk groups is 10 times lower. For the 70-79 group it increases noticeably, although it is still 2.5 times lower than for this age group against the whole population. The results in the oldest group confirm how serious risk factor is age alone. Interestingly, the CFR fatality rate for people under 60 years of age not belonging to risk groups is at a similar level as the estimated IFR fatality rate for this age group.

The above data clearly show that the fatality of SARS-CoV-2 infected persons

  • is exponential function of age,

  • is strongly dependent on coexisting diseases,

  • differs by up to four orders of magnitude for different age groups and risks.

These conclusions are optimistic for healthy and young people only. The opposite is true for older people. The most optimistic fatality predictions for people over 60 years old show that between 1 and 10 people per 1000 infected people die and for people over 80 years of age between as many as 1 and 2 people per 100 infected people die.