By now many have heard about the debate over convalescent plasma as an effective treatment for Covid-19. On Aug. 23, the FDA issued an Emergency Use Authorization stating that convalescent plasma—extracted from the blood of recovered patients—“may be effective” in hospitalized patients and can be used in the current emergency. The agency stopped short of fully authorizing or licensing convalescent plasma. But on Sept. 1 a National Institutes of Health panel argued that the data were insufficient “to recommend either for or against” use of plasma.
How can two federal agencies come to different conclusions on a topic of such great importance in the middle of an epidemic? Understanding the dispute requires a little history and also thinking about how experts evaluate evidence.
In late March 2020, the FDA allowed physicians to apply to give patients convalescent plasma on a case-by-case basis for compassionate use. The basis for this decision was that such plasma has been effective in other viral diseases, including those caused by other coronaviruses, and has shown encouraging results in small studies on Covid-19. With no alternative therapies, physicians flooded the FDA with requests. The FDA in early April developed an Expanded Access Program, which allowed treatment under the auspices of a Mayo Clinic research study of plasma safety, while the agency awaited firmer data from randomized trials. We are both co-authors of the Mayo Clinic study.
Many hundreds of hospitals enrolled their patients in the expanded program, and more than 1,000 physicians across the country administered convalescent plasma to tens of thousands of patients. Government and academic scientists from the Mayo Clinic and elsewhere mined this huge set of data for insights about plasma’s efficacy. This approach was unprecedented, as the database was set up primarily to monitor safety. The criticism has been that plasma was being used routinely without the benefit of randomized trials and without the usual government approval process.
Here’s what was gleaned from the data: First, consistent with everything known about plasma, treatment early in the course of disease was critical. Second, in patients treated early, mortality rates were lowest when plasma contained high levels of antibodies, highest in patients who received low antibody plasma, and intermediate when levels of antibody were in the middle.