If that's the case. The sampling strategy is wrong. The point of rct is that you even out extreme edge cases and try to find a statistical validation for the population. You can't just call a trial unreliable without a valid data
And the sample for covishield,Pfizer and moderna vaccine was pretty diverse. For your example, how do we know there wasn't some pre existing complicacy (it goes both ways)
Also SAE reporting/rare disease isn't something hidden atleast for fda or Ema. They have very structured way of reporting. I really doubt that it will not be recorded and patient history won't be checked. If it was a widespread issue it is bound to come to limelight.
The data isn't there to identify this a direct correlation. Not a single peer reviewed paper has proven this as direct correlation.