CNN medical expert blows a huge hole in Trump's plasma pitch

According to CNN senior medical expert Elizabeth Cohen, Donald Trump's much-heralded announcement to allow the use of convalescent plasma in the treatment of COVID-19 is rife with problems and needs much more extensive research before doctors move forward with it.

Speaking with host Jim Sciutto, Cohen -- whose view of the therapy was echoed afterward by Dr. Richard Besser who once headed the CDC -- explained in simple terms the problematic nature of Trump's proposal.

"Jim, what the studies show is that, while this might work, we don't know if it works, and what the studies really show is that the Trump administration is cherry-picking data to make this treatment look better than it might actually be," she began. "I'm going to get nerdy, but here it is: they chose a death rate what happens seven days out after treating people, that's okay but why not look 30 days out? The seven-day death rate looks better! It's a better number so they chose that one, when they should have chosen the 30-day one, so let's take a look at what the actual numbers show."

"The Trump administration says 35 percent decreased death rate when you treat people with convalescent plasma earlier in the hospitalization versus later on," she continued. "That's when you look at the seven-day death rate. You look at the 30-day death rate it's only 23.6 percent lower, so that was a difference not pointed out by the Trump administration."

"They chose the better number that is less reliable and this part is even more important," Cohen continued. "This death rate can be explained, this lower death rate can be explained by other factors. You don't need to be a medical genius to know that earlier treatment is always better. If these patients were getting plasma earlier, what else were they getting earlier? They were in the hospital earlier, so they were getting Remdesivir, steroids, other things earlier, that could explain why they had better survival, not necessarily the plasma."

Watch below: