Yes in this case you watched to much TV.
Switching the complete arms must be MORE then just a database thing. Here is why:
When the trial STARTS (and that is when the switching happened according the filing) you do not know the patients. So there is noting to switch that is like patients ID or so and switch them between arms.
The patients will populate the arms as the trials go and the randomizing was said to be 1:1:1
So each time a patients comes in the company in charge of randomizing receives the patient info tags it A, B or C and sends it to CSM. CSM then add the patient to the indicated arm, collects the patients weight from the treatment centre and starts to ship SOC+Bavi, or SOC+ 1 or 3 mg/kg Bavi that it makes from the bulk. During the trial weight may get updated and dosage adapted.
So the switching must have been done in an other way if it was done in 2010. And that is in the ARMS properties. When an arm is created CSM will assign treatment to it. SOC+placebo or SOC+1 or 3mg/kg Bavi. All labels will now be printed accordingly.
If you do the switch there, nobody will ever ask or see UNLESS the data is audited by a second person. Procedure wise you could switch it AFTER the audit because the audit is a QUALITY control measure NOT a tempering measure. Nobody assumes someone will change the data after the audit.
A good clinical trial management system would make it impossible to change after audit, maybe the CSM system didn't have that feature. But if it did, the only way to bypass it is to grep (if linux) or SQL query (if an SQL database) the database directly. And that needs you to logon. And again in a good system a project manager cannot get that kind access and is supposed to use the user interface screens. A programmer or other ITC person may.
And this will NOT have been a big change to make.
If you are correct wook, an auto change back mechanism would need some batch to run on a date-timer or there must be a pc of the clinical trial management software (or maybe a script) that runs regularly and at a certain moment, based on a system date check, execute the code to toggle both arms again. Normally CSM wouldn't have any access to the source code of such software because it is very probably something (a package) they bought.
In a script you could have an access problem as scripts run under the access rights of the users that started them (in which case it would be easy to trace who did it) or under system rights in which case again the project manager alone can not have done this.
So that would leave two options. The NEXT program manager that took over is in on it or some ITC person with the needed access rights is involved.
News 
Market Data 
Discover 
AI
