There’s a difference between incorrect and incomplete.
Facts, like how many people in a trial of 23 recovered, can be shared and useful early on even though they’re incomplete. That’s different from attempting to draw a strong conclusion about what works from n=23, that might be an incorrect conclusion.
The form that the incomplete data took was part of the problem.
Adding numbers to a spreadsheet is one thing. When the same data takes the form of an NJEM or Lancet paper, it definitely implies--and some cases, outright says--more than "FYI: 37/53 people got better when we did [this]." Part of this is how papers are ordinarily written, but I think the authors (and editors) should have stopped all over the word "improvement" in those manuscripts.
Facts, like how many people in a trial of 23 recovered, can be shared and useful early on even though they’re incomplete. That’s different from attempting to draw a strong conclusion about what works from n=23, that might be an incorrect conclusion.