Letter to the Editor: “Retrospective evaluation of CT findings to shape epidemiological models of COVID-19 outbreak”
by Andrea Ferraris, Marco Calandri, Paolo Fonio (firstname.lastname@example.org)
Since January 2020, COVID-19 virus has rapidly spread in mainland China and into multiple countries worldwide . The first registered case in Italy (apart from two Chinese tourists in Rome), was reported the February 21st in Codogno, Lombardy.
At that date, only 47 cases were reported in Europe. After February 21st, the epidemic curve increased rapidly: as of March 5th, there were 4,250 cases all over the Continent (3,850 only in northern Italy) . As for the Italian epidemic outbreak, the first registered COVID-19 cases were infected due to local transmission, without the possibility of doing an effective contact tracking to identify the index case  .
Furthermore, in consideration of the sudden epidemic slope in the north of Italy, questions have been raised concerning the presence of a not inconsiderable number of undetected infected individuals before of February 21st.
A precise characterization of the initial epidemic spread and transmission dynamics is more than desirable in consideration of the need of improving statistical and epidemiological models  .
Nevertheless, to obtain this information can be difficult: RT-PCR from nasopharyngeal swabs cannot diagnose a previous infection and serological tests are not informative on the precise time of infection. Can incidental CT-findings of thorax CT acquired before the first registered case be informative in this regard?
We retrospectively reviewed 4,488 thorax CT scans performed between the first December 2019 and the 20th February 2020 at the largest hospital of the Piedmont region, the second most affected region of Italy. Thirty-six CT scans (0.8%) unveiled CT findings (classified according to the structured report released by the Italian Society of Medical and Interventional Radiology, SIRM ) compatible with COVID-19 disease, at that time classified as unspecific interstitial pneumonia.
Review of clinical data confirmed for 30/36 of them (83.3%) a symptomatology compatible with COVID-19 disease (at least one symptom among fever, cough or respiratory distress). According to clinical registries, of the 36 Patients, 13 died (36%): three of them for acute respiratory distress syndrome.
The control group was defined considering the 4,682 thorax CT scans performed from December 1st 2018 to February 20th 2019. Of these, only 10 CT scans were found with CT findings compatible with COVID-19 (0.21%). The incidence of these findings was different from the study group (p = 0,001).
Incidence of CT findings compatible with COVID-19 in the period before epidemic outbreak, was four times as much as the number of similar findings in the same period of the year before.
Firstly, although not specific, these data support for the first time the hypotheses of the presence of unknown COVID-19 cases before the 21st of February in the north of Italy.
Secondly, after internal communication of these findings, the patients of the 2020 cohort found positive to the CT scan re-evaluation have been called back to undergo serological tests and fulfill an epidemiological questionnaire. We believe that this approach, to retrospectively take advantage of the sensibility of CT findings in order to select a population for targeted serological tests, can be a reproducible approach in order to improve the comprehension of the first stage of the epidemic spread and to subsequently modify epidemiological models and public health actions.