The cause of Germany’s low Covid-19 mortality is potentially disturbing
Of the published Covid-19 statistics, Germany appears to be an outlier. We have been asking ourselves why Germany's mortality rate is so low, just three. The total number of reported cases is similar to that of France or Spain, currently around 2000. Is there something wrong with the way Germany measures these data? We think not. As this analysis suggests we should not be relating today’s deaths to today’s new cases but to new cases of several weeks ago. Three deaths would be perfectly consistent with a range of 150-300 infections. But today’s reported number of infection would suggest that in Germany, too, the mortality rate is likely to rise sharply soon. The mortality rate is logically a lagging indicator of the rate of infections. The lags may be longer in Germany because the medical system is coping well. The lags will be shorter in Italy, where some people are now dying even without receiving treatment. We assume that the number of deaths in Italy will rise above 1000 today. That number would be consistent with a much larger than reported actual infection rate, more likely to be 100,000 than 10,000. The above-cited analysis gives even larger numbers.
The German data may, in fact, be more accurate than others. In any case, we should be very careful when doing Covid-19 statistics based on reported infection rates. From a public policy point of view, it is much more important at this stage to understand the dynamics, and the timing of lockdown measures. Over-reliance on reported cases may give a false sense of security. The mortality numbers might be a better indicator of the spread of the disease.
The good news for Germany is that it has one of the best healthcare systems in the world. But Germany will have greater difficulty than Italy to impose self-isolation measures. This is why the data are not as benign for Germany as they may appear at first sight.
The Lancet yesterday published a much-noted article on the effect of lock-down measures. We know from the analysis of data from the Spanish flu of 1918 that cities that banned public meetings had a lower infection rate than those that did not. But, as the article argues, these measures are unlikely to get the job done. We would advise anyone interested in this subject to read the analysis carefully. We can only summarise its main points.
The key task is to reduce the average number of transmissions per infected person, denoted R-zero. The R-zero of Covid-19 is thought to be about 2.5. This is consistent with an estimate that 60% of the population would eventually contract the disease. There is a back-of-the-envelope formula that relates R-zero with the total spread: 1-1/R0. If you plug in 2.5 you get the reported of 60%. If you plug in 1.5, you get 33%. In the absence of a vaccine, the only effective measure we have is a reduction in R-zero.
The authors argue that, based on the available evidence, the most effective public-sector measures would be extreme isolation - in other words, the measures taken in China and now in Italy as well. The measures taken by governments currently, like school closures, are unlikely to be very effective. Limiting public gatherings will help, but this will not reduce R-zero by much because the available evidence suggest that the virus spreads through prolonged contact. The authors are cautious in their conclusions, and honest about the many things they still don’t know about the virus. But their firm conclusion is that a critical factor will be the length of time between the onset of symptoms and self-isolation. Individual behaviour may in the end be more important than government action, unless of course that action is as extreme as China’s. The one prediction we would make is that it won’t be.