As I said in another comment - the only data the pharma company gets to "take home" is anonymized (partially, ie. a number instead of a name) and blinded (you don't know if they received drug or placebo)
The company must reconcile the patient data with the trial data (on premise), in a scenario like: "liver enzymes abnormal after administration of vaccine, is there anything in the patient data that could clarify that (apart from the vaccine itself)". Someone has to painstakingly go through all that data.
After the study is concluded, won't they get the non-blinded (still anonymous) version?
During efficacy/side-effect analysis you should definitely use only blinded data. After confirmatory statistical analysis, it seems reasonable to go over the data in an exploratory pass to possibly discover/hypothesize why/when/how the medicine works/fails/interacts. Or do they just get the aggregate report?
