Abstract
The daily Covid-19 fatality rate is modelled with a trend line based on nominal day-to-day reproduction rates and a cosine to take account of weekly fluctuations. The fatality trajectory represented by this trend line can be projected from the number of daily infections by assuming a time lapse between symptom onset and death between 17 and 19 days and a nominal time-dependent fatality rate. The time trajectory of this fatality rate suggests a change of the infection dynamics at April 3, with an increase from 2.5% to 6% within 20 days perhaps indicating spread of infection to more vulnerable people. Later in summer, the nominal fatality rate decreases down to 1% in mid-July raising the question whether Covid-19 is intrinsically less lethal in summer. Although the time trajectories of infections and fatality are pronouncedly different, the reproduction rates obtained therefrom are similar indicating that the infection dynamics may reasonably well be deduced from the potentially biased reported infection rate if it is biased consistently, i.e. the same way, over an extended period of time. The administrative measures to contain the pandemic seem not to have an immediate effect on the infection dynamics but well the ease of restrictions. An effect of mask wearing on decreasing lethality cannot be excluded.
1 Introduction
In an earlier paper, we have presented a mathematical procedure to lay a trend line through the reported number of daily infections in Germany, Sweden, and France based on a day-to-day reproduction rate and, if appropriate, a cosine to account for periodic weekly fluctuations (Mergel D 2020 1, Mergel D 2017). This procedure achieves an effective noise reduction and allows to calculate an effective reproduction rate at the time of symptom onset of the infecting persons. It further allows to construct the infection time trajectory from the effective reproduction rate so that different progressions of the epidemic can be modelled. Hypothetically implementing part of the dynamics in Sweden onto that of Germany leads to about 8000 more fatalities up to the end of April.
It has often been argued that the reported number of confirmed infections is not representative of the number of true infections due to peculiarities of the testing procedure so that the only reliable data are case fatalities.
In this paper, we:
- lay a trend line through the reported number of Covid-19 fatalities with the same procedure based on nominal reproduction rates
- predict the fatality trajectory from the infection trajectory assuming a time-dependent lethality
- check whether the reproduction and lethality trajectories are influenced by the administrative lock-down measures listed in Table 1.
Lockdown measures of the German administration
The nomenclature used in this paper is summarized in Table 2.
Data structure and nomenclature
2 Trend line for the daily fatality rate
Figure 1 shows the daily Covid-19 fatality rate nF from march 9 (t=9) to July 26 (t = 148) as reported by the Robert-Koch-Institut (RKI 23-07-2020, 2) together with a trend line nFcalc modelled with a nominal day-to-day reproduction rate rT of the type developed for the daily infections with Covid-19 (Mergel 2020, 1; Mergel 2017). The procedure to get the trend line is as follows:
Daily fatality rate nF(t), trend line nFcalc(t), and nominal day-to-day reproduction rate rT(t), R2adj due to Equation 5
A seed point (tS, nS) is chosen near the supposed maximum of the curve and the trend line is modelled, with nCalc(tS) = nS, for t > tS as:
and for t < tS as:
The power parameter d is set to 4, d = 4, because this is the most probable time span between the symptom onsets of the infect-ing and the infect-ed person (Böhmer M and others 2020). The position tS of the seed point is kept fixed but the value nS is a fit parameter.
The function rT(t) is modelled as a polyline with 9 vertices where the start and stop instants, t = 9 and t=148 are kept constant and the other 16 parameters are treated as adjustable fit parameters.
Figure 2 displays the residuals nFres (t) = nF (t) −nFcalc (t) of the fit of Figure 1 together with a trend line nFcos intended to represent weekly fluctuations:
Residuals of the fit Figure 1 together with the cosine resCalc defined in Equation 3
The sum of the square deviations of the experimental nFres and the modelled nFcos residuals is taken as the target to be minimized with the 1 value of the seed point, the 16 parameters of the vertices and the 2 parameters A and tN of the model residuals. So, the trend line in Figure 1 minimizes the sum of the square deviations of the residuals from the cosine model and therewith maximizes the adjusted coefficient of determination R2adj defined as:
Where vTot is the total variance of the nF and vRes the adjusted residual variance (adjusted for the decreased degree of freedom due to the number of fit parameters) calculated as:
Where nPoints is the number of data points (here 140) and the nDdof the number of fit parameters (here 1 + 16 + 2 = 19). For the fit in Figure 1 and Figure 2, R2adj = 0.95 is achieved.
3 Projecting the fatality rate from the infection rate
For people dying of Covid-19, the time between symptom onset and death is reported to be 17, 18, 19 days in 90% of cases (Verity R 2020 and others). So, our model for projecting the fatality rate is:
With lT being a nominal lethality, called nominal because the reported number of infections is generally regarded to be systematically underestimated. Setting lT = const. = 6% yields the projection nFproj displayed in Figure 3.
nFpred is the fatality rate projected from the trend line nIcalc of the infection rate, here with constant lethality (6 %) and time to death 18 days.
Although the time shift seems to fit with the onset of fatality and the approximately exponential decay after t = 50, the characteristic shape of the fatality time trajectory is not captured. There seems to be a marked difference between an early and a later stage of the epidemic. We therefore try a second approach by modelling the lethality time-dependent, using again a polyline as for the infection rate and the fatality rate. The result can be seen in Figure 4. The polyline, designated lT, consists of 12 vertices. Now, the fatality time trajectory is reasonably well modelled.
Fatality rate nFproj projected from the trend line nIcalc of the infection rate with the time dependent lethality leT2000 as fit curve, “2000” means that the numbers have been multiplied by 2000 to fit the secondary vertical axis
A similar fit is obtained when assuming an average time to death of 20 days leading to a better coincidence with the initial stage (t = 20) of the fatality rate. The two nominal lethality rates are displayed together in Figure 5. They represent the probability to die for a person with symptom onset at t.
Nominal lethality rates for assumed average time to death alternatively 18 or 20 days
Their more prominent common features are:
- lethality dynamics changes for symptom onset at t = 34 (April 3) with rapid increase from 2.5% to above 6% for t > 50.
- nominal lethality decreases after t = 72 (May 11) down to 1.2% at t = 138 (July 17).
The increase after April 3 may be because the infection has spread from younger and healthier people being infected in superspreading leisure time events to more vulnerable persons.
Nominal lethality may decrease because:
- testing has captured more people with mild diseases;
- vulnerable population is better protected;
- Covid-19 lethality is intrinsically lower in summer time.
The vertical dashed lines indicate the administrative measures listed in Table 1 and intended to contain the spread of the virus or keep it contained. There is no evidence of an immediate influence of these measures on the lethality trajectory. However, a positive effect of mask wearing on decreased lethality (due to decrease of viral load?) cannot be excluded if its appearance is delayed by competing effects, e.g. continued spread of the epidemic to more vulnerable persons.
4 Reproduction rates
A nominal effective reproduction rate is calculated from the trend line through the fatality trajectory as (Mergel 2020, 1):
Figure 6 compares this reproduction rate being shifted by tFatal = 18 days to earlier dates, RF(t-tFatal), with the effective reproduction rate obtained to model the daily number of infections Reff(t).
Effective reproduction rates obtained from the reported number of daily infections (Reff) and from the reported number of fatalities (Rf)
Despite differences in detail, there is a remarkable similarity in the main characteristics:
- rapid decrease towards 1, before administrative measures to restrict private life are in effect;
- effective reproduction rate of about 0.8 in the long run.
The overshoot of Reff around t = 107 is due to a superspreading effect at that time (Tönnies 2020) leading to a peak in the reported number of daily infections.
The similarity of Reff(t) from infections and Rf(t) from fatalities suggests that the dynamics of the infection process may well be captured by analyzing the infection rates although the number of reported infections is generally considered not representative of the infection demographics. However, if it is biased consistently, the same way in a time span, the effective reproduction rate calculated from the infection trajectory arguably describes the infection dynamics sufficiently well. A change of the testing procedure or an isolated outbreak lead only to a temporary disturbance.
Administrative lockdown measures do not seem to have an immediate positive effect on the infection dynamics, but the ease of restrictions at t = 107 arguably a negative effect.
5 Conclusions
A trend line nFcalc through the Covid-19 fatality rate can sensibly be modelled similar to the trend line nIcalc for the infection rate, with a seed point representing the strength and a day-to-day reproduction rate representing the dynamics. The effective reproduction rates calculated from the two trend lines are similar if nFcalc is shifted by 18 days, the average time from symptom onset to death, towards earlier times.
The coincidence of the two reproduction rates indicates that the effective reproduction rate calculated from the number of daily infections is representative of the infection process even if the testing procedure does not capture all cases. It is sufficient that testing bias is consistent, the same way over a period of time.
There is no evidence that administrative measures to contain the epidemic influence the infection dynamic represented by the reproduction rates. Caveat: absence of evidence is not evidence of absence! This does not necessarily mean that social distancing is useless but arguably indicates that self-regulatory processes are at least as important as administrative measures.
The fatality rate can be projected from the infection rate by assuming an average delay of 18 or of 20 days between symptom onset and death, and a nominal time-dependent lethality. A sharp increase of the nominal time-dependent lethality after April 3 from 2.5% to 7% indicates a change in the infection process probably due to spread to a more vulnerable fraction of the population.
The nominal time-dependent lethality decreases from 7% at mid-May to 1% at the mid-July. This is still above the lethality estimated from serological studies (Bendavid E and others 2020, Mergel D2020-2, Streeck H and others 2020). Statistically regarded, there are three possible reasons for the decrease:
- testing has captured more people with mild diseases;
- vulnerable population is better protected;
- Covid-19 lethality is intrinsically lower in summer time.
It cannot be excluded that face mask covering has its stake but more detailed analyses and epidemiological expertise are necessary to decide on this issue.
Footnotes
Email: Dieter.Mergel{at}Uni-Due.de
Subject Area
Reviews and Context
1
Comment
0
TRIP Peer Reviews
0
Community Reviews
0
Automated Services
0
Blogs/Media
Author Videos