How Twitter censored medical experts, and "might have" led people to make wrong decisions