Fig. 1a shows the temporal evolution of the cumulative number of deaths for the selected regions (Lombardia, Veneto, Piemonte Emilia Romagna, Marche, Campania, Lazio, Liguria, Abruzzo). In Fig. 1b, the same data are normalized per million inhabitants. The cells in the control table contain values that can be edited by the user. The “Factor” column allows the user to rescale the vertical axis, multiplying it by the entered value, or to make the region disappear from the chart by entering the value 0. The “Shift” column allows to apply a positive translation in terms of days.
What do these charts tell us?
Leaving all the default values unchanged (Factors=1 Shifts=0), these graphs allow to get a general picture of the trends in the different regions. The first graph clearly shows that the number of deaths in Lombardy is much higher than any other region, reaching almost 50% of the total and exceeding by almost an order of magnitude that of Veneto. All the regions of the centre-south Italy have been much less affected, as is evident even when the normalised to the population numbers are taken into account. It is interesting to note from the second graph how, although it was one of the first regions to be affected, the number of deaths normalized to the population for Veneto is much lower than all other regions of the North.
By changing the values in the table, the user can highlight the striking similarities in the shape of the curves, by translating and/or heating the data until they overlap. This makes it possible to obtain some estimates, to be considered prudently, regarding the evolution of the epidemic in some regions, on the basis of what happened in those affected before the spread.
An example can be achieved by applying to the Lombardia curve in the Graph of Fig.1b (data referred to April 22) a scale factor F = 0.45 and a time translation of S = 2 days. This is achieved by erasing the other regions by erasing (F = 0), an excellent overlap with the curve of the Marche region. The image is shown here. If this trend is respected, in 3 days the number of deaths per million inhabitants in the Marche region will be just under half of what Lombardia has now.
Fig. 2 shows the trend of the daily percentage increase in deaths based on the total number of deaths for different Italian regions. The Italian regions have been selected with the most significant spread of the contagion, in conjunction with the regions where there are operational units of our Institute.
We first observe that this is not a time-dependent plot: the time variable has been removed. It shows how effectively in absolute terms the various regions have managed to contain the lethality of the epidemic (zeroing out the daily growth rate of deaths) before reaching a too large number of deaths.
Comments (data from April 20):
All regions follow a more or less similar pattern with numerous irregularities most likely related to the way data is acquired (delays in the processing of swabs, lack of information from health institutions (RSA), etc.). For example, in Fig. 2 we recognize three separate groups of behavior in crossing the 10% daily increase threshold: Campania, Lazio and Abruzzo cross this level with just over 100 deaths, the regions in the Northern Italy with a number of deaths between 500 and 1000, with the considerable exception of Lombardia with more than 4000 deaths. However, if this trend is considered in relation to the population of each individual region as in Fig. 3, we can identify a similar and still more clear situation. For Campania and Lazio, the number of deaths per million inhabitants is about 20, in all the other regions the threshold of 10% daily increase is crossed when the number of deaths per million inhabitants is between 100 and 200. Finally, only for Lombardia, this event took place for more than 400 deaths per million inhabitants.