Background and Objective:
The COVID outbreak has required many person-hours of infection control and occupational health personnel within hospitals to identify the community or nosocomial origin of the infection, as well as case contacts between patients and healthcare professionals. Furthermore, nosocomial COVID represents a care-related infection and, as such, it can be one of the causes of complaints and/or requests for compensation by patients and caregivers. Having a rapid and reliable system for identifying the origin of COVID cases has the following advantages:The reduction of the commitment of dedicated healthcare personnel;Early identification of cases with possible image damage or medico-legal implications;The definition of an internal and external benchmark. We tried to develop an algorithm that distinguished community, probably community, probably nosocomial, and nosocomial COVID cases, adapting the case definitions by the source of infection published by ECDC.

Methods:
The following information was extracted from the electronic health records:Type of hospitalization: urgency, elective;Date of admission;Positivity date for 1st positive test for COVID; Based on that, the algorithm has assigned the different COVID+ cases to one of the 4 categories. The algorithm, validated by infection controll personnel, was then implemented.

Results:
The validation of the algorithm carried out on a n. 65 of COVID+ cases showed an initial per cent agreement of 56.9 %. The different attribution of the category was initially found for probably community cases, whose correct identification was facilitated by the graphical interface of the application.

Conclusion:
automatic methods for defining cases on parameters that are easily identifiable in the electronic health records can be an effective low-cost analysis tool with a high impact on the organization of the hospital staffs time commitment, as well as a tool to support daily clinical activity and medium-long term business planning.