Could Italy host the new coronavirus before China?

The pandemic of the COVID-19 disease caused by the SARS-CoV-2 virus has been believed to originate in China and spread later to other parts of the world. It is well acknowledged that the first diseased individuals appeared in China as early as in December 2019, and possibly even earlier in November. It has also been well established that the virus stroke Italy later in January or in February 2020, distinctly after the outbreak in China. Paper by Apolone et al. published in a local Italian medical journal in November 2020 however exposed this chronology to doubt. By fitting early part of the epidemic curve with the exponential model and extrapolating it backwards in time, we could estimate the day-zero of the epidemic and calculated its confidence intervals in Italy and China. We also calculated how probable it is that Italy encountered the virus prior 1 January 2020. We determined an early portion of the epidemic curve representing unhindered exponential growth which fit the exponential model with high determination >0.97 in both countries. We suggest that the day-zero in China and Italy was 8 December (95% CI: 3 Dec., 20 Dec.) and 22 January (95% CI: 16 Jan., 29 Jan.), respectively. We could with high confidence reject that Italy encountered the virus earlier than China (p <0.01). Based on our analysis we oppose the findings published by Apolone at al. and view the proposed pre-pandemic presence of the virus in Italy as very unlikely.


45
The global pandemic of the COVID-19 disease caused by the SARS-CoV-2 virus will certainly be tracked 46 among the most significant global events in the recent history with a death toll nearing 2.5 Million as of 47 the end of January 2021 [1]. Its humanistic, economic, and political impact has not only shaped the 48 ongoing efforts to contain its spread but also relationships between countries, their citizens, and       110 has been however published to date.

111
We want to respond to speculation that the virus could be in Italy before the outbreak in China, which Italy, which we believe describes the best the epidemic progression within a naive population with 119 minimum imposed measures affecting its intrinsic dynamics. It was repeatedly shown that epidemic 120 curve in its early portion can be well fitted by simple phenomenological models as intrinsic growth is the 121 determining variable [13,14]. Considering the interval from January through April only, we identified the 122 most linear early portion of new cases projected day-by-day on log-linear plot, using a semiautomatic 123 process. In SAS v. 9.4 for MS Windows, we stepwise rolled a 10-day window over the data and

151
In addition, we looked at 95% and 99% confidence interval for the d 0 estimates in both countries to 152 judge the probability that Italy hosted the virus before China.

155
Performing the log-linear analysis on the Italian data, we identified the interval from 28 February through

164
As a sensitivity analysis we trimmed the early data from the Chinese curve which departs from the log-

193
We made an attempt to revise the speculation that Italy could host SARS-CoV-2 prior China has 194 confirmed the outbreak. However may this seem to be built on a weak foundation to anybody who

210
The work by Russo et al. suggests 14 January as the day-zero in Italy, which is eight days earlier than 211 our estimate and within our 99% confidence interval (11 January, 3 February). The difference could be 212 explained by the delay between the onset of the disease and the date of diagnostics. In our model we 213 did not adjust our estimated day-zero for this delay as, for the early period of the outbreak, we found no 214 reliable data to support it. The Italian outbreak was initially concentrated in the north of the country and 215 hence the data we used for fitting come predominantly from this regional outbreak and does not 216 compound several parallel outbreaks. It was first later in March and April when the epidemics spread 217 over the entire country and abroad [16]. This further justifies the use of a simple model over a

236
Much has been written and said about specificity of various SARS-CoV-2 antibody-based assays and it 237 is out of scope of this paper and competence of its authors to deep dive into this topic. One of several 238 possible explanations of the surprising finding by Apolone et al. could consist in the use of a cross-239 reaction of the used antibodies [19].

240
The strength of our study is its robustness consisting in the use of adequate model on a consistent data 241 with minimum assumptions needed. The main limitations of our study is the unknown proportion of 242 undetected infected individuals and the unknown time between the infection and the test date in the 243 early days of the pandemic. Assuming that both would have a substantial effect, it would shift the day-244 zero by days. We believe that this error would rather constantly affect both countries and hence more 245 or less preserve the order.

247
In our work no medical or research intervention was imposed on humans or animals. We did not 248 specifically address individual human subjects.