Between IVN and TRI took place a meeting on potential use of a model created by TRI in Romanian Trial. The model was developed to predict an E. coli infection and this could be extended to infectious agents such as viruses. In this situation this model could be considered for the design of the Romanian trial. During the meeting, the availability of medical data from centralized databases or at the hospital level was signaled.

In Romania, there are 2 main data flows:

1) Communicable Diseases Data: paper/excel files, not a very high level IT system. Goes to a central database (CDC Romania). This data are hard to access because contains personal information (e.g., name).

2) Hospital data: encounters of patients of communicable diseases who stayed in hospitals. This data are easier to access but costs a lot.

This goes to central database, validated/corrected/stored in multiple tables. Data is at patient level e.g., age, diagnosis, procedures, status (admission/discharge). It does not contain info like treatment with antibiotics or symptoms, which the model in discussion requires. It does have other information like comorbidities, diagnoses, date of admission/discharge.

Data would be anonymized before get it.

Data submitted to EU level is aggregated: nothing like person information, age or gender; just how many patients of x kind encountered in a period of time.

In order to make most out of Romanian data, IVN should wait until the model is clearer and variables needed are described properly so that they can reassess what data needs to be sought. Only after this should they try and get relevant data.

Regarding communicable diseases data: any model/ database could be a problem when is linked with hospital acquired infections. While this might be best for the model in question, Romania (and likely other countries) has a problem of distorting data because of legal ramifications of higher infection rates (law punishes hospital management if the hospital acquired infection rate is high and they do not report or distort these kinds of statistics).

Addressing hospital acquired infections Romanian team had concerns until to read the documentation about the model, but now thinks this data is appropriate.