We are a academic CTU, who is debating how to interpret the meaning of the onset date for the event. We wonder whether it should be the first sign and symptoms of the event (i.e. when the event may still be non-serious) or whether it should be when the event meets any of the seriousness criteria.
We thought it was easier for new events which become serious within hours or days and in those cases we would take the onset date as the date of the first sign and symptoms on the SAE form. However we found it more difficult for long-term and pre-existing AEs which worsen during the trial as we debated how to pinpoint the correct onset date (e.g. date of worsening? date of change of the severity grade from 2 to 3, and then maybe change to 4? Events which fluctuate often from grade 2 and 4, which date to consider?). We couldn't find ant clear guidance on the literature and ICH, therefore we welcome your advice.

We would like a system which works for both new and existing AEs.

Many thanks!