Epidemiological Philosophy of Pandemics

Introduction

There is strong evidence from various studies on the importance of the dose of the inoculum of a pathogen that can lead to severe infection.¹ Such pathogens include those causing influenza ^2,3 and the measles⁴, as well as Human Immunodeficiency Virus (HIV),⁵ tuberculosis (TB),^{6, 7} Streptococcus pneumonia,⁸ HBV⁹, flavivirus West Nile virus¹⁰, and Coronaviruses.^1,11

The proposed mechanism by which a high viral inoculum leads to more severe disease is via a dysregulated and overwhelmed innate immune response to a higher viral dose, where immunopathology plays a role in viral pathogenesis ¹². This may be the case for COVID-19.¹³

It has been suggested that a minimal viral inoculum may be controlled subclinically by innate defense mechanisms, while massive doses can overwhelm the innate immunity and may cause severe disease and rapid death. ¹²

Unfortunately, this issue has not undergone any challenge trials. Furthermore, epidemiological studies to correlate the association between the clustering of cases with both the mortality rate (MR) and the case fatality ratio (CFR) are lacking.

Several research groups have developed epidemiological models of COVID-19. These models use confirmed cases and deaths, testing rates, and a range of assumptions and epidemiological knowledge to estimate the number of true infections and other important metrics.¹⁴

CFR has gained great importance in the COVID-19 pandemic, because the expected total mortality burden of COVID-19 is directly related to the CFR. According to WHO, countries are making their final CFR estimates as active cases are resolved Unfortunately, current CFR calculations during ongoing epidemics have been criticized due to the wide variation in CFR estimates over the course of an epidemic, making them difficult to compare for several reasons. These models might not accurately track the pandemic, as they apply previously determined infection fatality ratios (IFRs) from local sources or abroad. This makes currently used models using predetermined CFR subject to great bias. According to the World Health Organization (WHO), there has been broad variation in naïve estimations of CFR that may make them misleading. ^15,16,17.

The testing capacity may be limited and restricted to people with severe cases of disease and priority risk groups. ¹⁷This makes continuous massive COVID-19 testing for continuous estimation of (IFR) or CRF a difficult task.

In the context of current variance and difficulties in CFR estimates and to shed light onto the unknown parameters associated with COVID-19 mortality, which are poorly understood, we investigated whether the CFR is associated with the MR expressed as the number of mortalities/million members of the population (M population). Moreover, this study also searched for an association between the MR expressed as deaths/M and the number of cases/M population, with number of cases/M population representing the density of infection.

Methods

Results

Table (1) and fig. (1) show a meaningful nonlinear regression (logarithm model) tested with two-tailed alternative statistical hypotheses. The slope value indicates that with an increase in the “test no./1M” of one unit, there was a decreasing effect on the number of total cases/M corrected to 1M tested, which was estimated as (–40924.93) and recorded as a significant effect (P-value < 0.05). The relationship coefficient was (0.44178), with a meaningful and significant determination coefficient (R² = 19.517%). Another source of variation that was not included in the studied model, i.e., the “intercept” showed a highly significant level (P-value = 0.0062).

• Download figure
• Open in new tab

Figure (1):

Long term trend of the scatter diagram concerning impact of test no. / 1M. on total cases corrected to 1M. tested

Table (2) and fig (2) show a meaningful nonlinear regression (power model) tested with a two-tailed alternative statistical hypothesis. The slope value indicates that with an increase in the tested no./1M. tested by one unit, there was a negative impact on the CFR, which was estimated as (–0.931198). This was shown to be significant (P-value < 0.05). The relationship coefficient was (0.43924) with a meaningful and significant determination coefficient (R-Square = 19.293%). Another source of variation that was not included in the studied model, i.e., the intercept, was not found to be significant (P_-value > 0.05).

• Download figure
• Open in new tab

Figure No. (2):

Shows the long term trend of scatter diagram effectiveness of tested no./1M. tested on corrected CFR.

Table (3) and fig, (3) show a meaningful nonlinear regression (power model) tested with a two-tailed alternative statistical hypothesis. The slope value indicates that with an increase of one unit in the total number cases corrected to the number per 1 M inhabitants tested, there was a positive influence on the total number of deaths/1M, which was estimated as (1.016959). This was recorded to have a very high significant influence (P-value = 0.000), as well as a relationship coefficient of (0.80260) with a meaningful and significant determination coefficient (R² = 64.416%). Another source of variation that was not included in the model, i.e., the intercept, showed no significance at P > 0.05.

• Download figure
• Open in new tab

Figure (3):

Long term trend of the scatter diagram concerning impact of total confirmed cases corrected to :1 M. tested on total deaths / 1Million

Standardization for the equation: total deaths /M=0.006293x(total cases corrected to the number per 1 M inhabitants tested)^1.016959

We considered the Diamond Princess for the standardization of our results. Data can be found at the public github repository: https://github.com/thimotei/cCFRDiamondPrincess. Further references are listed in the discussion section of this paper.

The number of deaths was 7, the total number of passengers was 3711, the total number of cases was 619 (301 symptomatic, 318 asymptomatic), and the CFR was 1.9, as at February 20, 2020.

The expected total number of deaths corrected for 3711 passengers was 5.004227, and the actual number of deaths was 7 (Providing our estimates were based on the confirmed cases equation).

Table (4) shows a meaningful nonlinear regression (Power model) tested with a two-tailed alternative statistical hypothesis. The slope value indicates that with an increase of one unit in the total number of deaths/1 M inhabitants, there was a positive impact on the CFR, which was estimated as (0.366580). This was recorded as a very high significant influence (P-value<0.0001), and the relationship coefficient was (0.61437) with a meaningful and significant determination coefficient (R² = 37. 746%). The intercept showed no significant P_-value.

The curve in fig 4 clearly shows that the CFR increases with decreasing increments.

• Download figure
• Open in new tab

Figure (4):

Long term trend of the scatter diagram concerning total deaths No./1M. population on CFR

Standardization of the equation: CFR=0.093200xtotal death no./M^0.366580:

We considered the Diamond Princess for standardization to calculate the CFRs. If the total number of deaths is 7, the CFR is 1.479.

If the total number of deaths is 14, the CFR is 1.911.

CFR = deaths/cases

1.911=7/? x 100

=368.421052 total cases, which includes all symptomatic cases (301 symptomatic) and some of the asymptomatic cases, because our equation calculates slandered confirmed cases, not all cases.

Discussion

In this study, we proved that we can estimate the CFR and the total number of confirmed cases once the ratio of deaths/M inhabitants is known. By building new models and equations that are suitable for IFR we can calculate the true number of total cases (confirmed and not confirmed) rather than an estimation of the equivalent number of confirmed cases. Further, it is possible to standardize the current equations and predicted models, which can be re-evaluated and readjusted. The Diamond Princess cruise ship possibly does not represent a convenient sample, as there are conflicting data regarding the IFR, according to previous estimations .^18,19,20 It is more convenient to obtain our IMR equation and model rather than using the CFR. However, we considered the CFR and the confirmed number of cases for this urgent situation. Anyway, we considered the initial assessment for our equations and models as being encouraging.

The results prove that the positive influence of the COVID-19 MR on the CFR and number of confirmed cases was very high by non-linear regression (tables 3 and 4). The most important confounder is the testing coverage (tables 1 and 2). We tried to adjust this factor, because decreased coverage can lead to a spurious increase in the ratio of the number of cases/million inhabitants tested and a spurious increase in CFR. In table 1 and table 2, despite adjusting cases per 1 M inhabitants tested for COVID-19, we found significant negative associations between the number of tests and the number of cases/M inhabitants and with the CFR. This limitation led to the underestimation of asymptomatic infections, which were estimated to be 10–70% of the total number of true infections elsewhere.²¹ Furthermore, limited access to testing could result in the undercounting of deaths.

Another limitation in this study (due to limited coverage in part) was that we estimated the confirmed cases/M inhabitants and the CFR rather than IMR estimates.

A further limitation is the relatively small study sample.

Estimates of the case fatality ratio (CFR) and infection fatality ratio (IFR) made in real time can be biased upwards by the under-reporting of cases and downwards by failure to account for the delay from confirmation-to-death.

The virulence of pathogens depends largely on the previous immunity of the host, whether through cross-reaction or due to previous infection or vaccination. Clinical trials should aim to identify the roles of both highly transmitted and not highly transmitted viruses and pathogens. These studies should also answer the question of whether a high load of a low virulent viral infection among not previously exposed subjects to the virus leads to a high mortality rate.

Factors such as mobility, social distancing policies, population density, and host factors can interfere greatly with the number of cases. The scope of this study was not designed to look for causes in the variance in the number of cases at different times and in different places. We test how the total number of cases affects the MR and CFR. That the number of cases, as a factor, determines the MR and CFR is a novel finding. The underlying this might be explained by the viral load and density of infection. Although this has been studied under certain conditions previously, the contribution of the number of cases in disease pathogeneses should be examined in depth again. During this pandemic, many works have tried to determine why CFR differs rather than looking at why the number of cases/M inhabitants differs. It was suggested, for example, that when the viral load is high, severe COVID-19 disease can result and multisystem inflammatory syndrome (MIS-C), a Kawasaki disease (KD)-like syndrome suggested to be related to the overdrive of innate immunity ^22,23,24,25, can occur in children²⁶. Family and community clusters of severe COVID-19 infection have been reported early in this pandemic. ^27,28

Reported excess deaths estimates were thought to represent misclassified COVID-19 deaths or potentially those indirectly related to the COVID-19 pandemic. It was suggested that this excess number was related to the pandemic itself and not to disease, i.e., it was attributed to lack of facilities during the pandemic.²⁹According to our estimates, this excess could represent a portion of the total cases calculated through IFR/CFR and the total deaths/M inhabitants.

Previously, it has not been easy to explain the extraordinarily high mortality rates during certain epidemics, and this has been a concern for scientists. For example, there have been high mortality rates during measles epidemics in the Pacific Islands, such as in Fiji in 1875 and Rotuma in 1911, with mortality rates of 20% and 13% of the total residents, respectively. The mortality rate in the Faroe Islands in 1846 was nearly 10 times higher than that during the 1911 epidemic in Rotuma.³⁰ One of the amazing things in these epidemics is that there was no direct evidence of hypervirulent strains of the measles virus or genetic predispositions to fatal outcomes after measles infection. ³¹

This makes us consider the role of high number of cases in initiation of high MR and CFR.

Again, in the 1918 influenza epidemic, virulence was notable when the number of deaths exceeded 20 million worldwide, with approximately half a million of these occurring in the United States.³²Noteworthy evidence from the 1918 epidemic was that one-quarter of the American population had clinically recognizable cases of flu during the epidemic, giving the impression of a high attack rate.³³ Before the 1918 epidemic, one has to go back to the black death (bubonic plague) of 1346 to find a similarly devastating epidemic.³²

During the COVID-19 pandemic, clinicians have struggled to understand why some infected patients experience only mild symptoms while others exhibit progressive, fatal disease.²⁷ I think this new evidence enable us to understand these amazing situations.

Advantages and conclusions

During previous epidemics and the current pandemic, there has been no model or equation based on the mortality rate to predict the total number of cases or the CFR.

During outbreaks, routine case detection of asymptomatic patients consumes time, effort, equipment, finance, and can increase the risk of health personnel getting and transmitting the infection.

We have provided uniform equations that can be applied everywhere with flexible CFR/IFR values.

The number of infected persons can be calculated in novel two ways: either through the CFR and through MR equations. We can also get it from CFR /IMR standered definitions.

We think that these equations could possibly be applied for every epidemic, as they are based on the estimation of mortality/M inhabitants, i.e., the curve and equations could be the same. The estimated number of equivalent confirmed cases should be adjusted to the total number of true cases.

The total number of infections is not a major determinant factor of the number of deaths in pre-estimated CFR modules. However, in our study, the number of deaths/M inhabitants was shown to be a determinant of the current CFR and the total number of confirmed cases.

An increase in the number of deaths/M inhabitants coincides with increases in the number of cases/M inhabitants and the CFR. This means that an increase in the total number of deaths outnumbers the increase in total infections, which leads to an increased CFR. The proxy indicator for an increase in the total number of cases is the number of deaths/M inhabitants and vice versa.

Our proposed new definition for an epidemic is an increase in the CFR from its standard level.

These findings will help in the development of infection control policies to break the chain of the pandemic and help to understand the philosophy of the pandemic.

Health systems should focus on decreasing the number of total cases, since MR and CFR increased with an increase in the total number of cases. All health systems could have the same fatality rate.

Better understanding of the pandemic behavior through showing that an increased CFR with an increased number of cases supports the viral overload theory.

Conflict of interest

There are no conflicts of interest worth mentioning.

There is no funding source.