I think we need to add a convention for when source data does not contain date and/or time information. The datetime field is now mandatory in the clinical data tables. However, many data sources do not contain time information for clinical events. And the conventions do not provide any guidance. If the source data has November 26, 2018 as the date for a drug exposure we know/can assume the patient was exposed before 11:59pm/23:59 on November 26, 2018. Likewise when the day of month is NULL at source, the patient had the clinical event in the source data by the last day of the month.
Thoughts or debate?