The Italian risk of making COVID decisions based on wrong metrics
We have already discussed how the encouraging effects of the nation quarantine on hospitals, especially on intensive care, and the greater presence of people with mild symptoms should be considered only transitory and that they will be less in case of reopening.
Furthermore, a lot of confusion was generated due to the use of a metric to check the trend of the epidemic defined as follows:
actual_positive_daily = new_positive_daily — recovered_daily — deaths_daily
This metric takes into account two factors placed at different times. The new_positives_daily are the new positives found today, as an effect of a recent past, while the recovered_daily and deaths_daily are the result of a contagion that occurred much earlier in time.
First, first note that this metric turns out to be negative if the number of daily deaths exceeds the daily contagion rate which is surprising. If there are many deaths and many infections, the metric tells us that everything is in place. At least surprising.
Secondly, note that actual_positive_dailyy can give smaller and smaller numbers, even if today there is a growth of new daily infections (factor new_positives_daily ) together with today’s growth of the dead or of the recovered who have been infected in the past (recovered_daily, deaths_daily).
In other words, this metric can give an excessive impression of improvement when there is no actual reduced improvement.
This is exactly what is happening in these days, and, unfortunately, this metric has been widely adopted during the civil protection press conferences and taken up and amplified by newspapers, giving rise to a little prudent sense of excessive improvement that everyone is talking about. It is appropriate that this metric is no longer used and that the confusion it has generated is definitively clarified.
In this situation, the only metric that matters is represented by the new_positive_daily shown in figure 1 while the metric actual_positive_daily should not be taken into account. The difference between the two metrics is shown in the table.
Different number of swab tests in Italy