RCT is the gold-standard way to establish cause and effect in humans. If there’s no significant difference in x between the experimental and control group in a well-designed RCT, I think it’s appropriate to conclude that the experimental condition does not cause x. If I’m wrong, please feel free to correct me.

That said, I did mention that the Pfizer investigators merely dismissed the all-cause mortality numbers without providing any statistical analyses at the end of the article.