Sept. 24 (UPI) -- COVID-19 antibody test accuracy could be as low as 61%, depending on the type, and when and how they are used, according to an analysis published Thursday by the journal PLOS Pathogens.
The 10 tests evaluated in the study proved between 61% and 87% accurate at correctly identifying blood samples as positive for antibodies for the new coronavirus, the data showed.
The same tests were between 82% and 100% accurate at identifying samples negative for antibodies, the researchers said.
The tests were more accurate with "serum" samples -- or blood collected from a vein -- than they were using "finger prick" samples, and with samples collected from patients up to three weeks after symptoms of the virus first appeared, study co-author Suzanne Pickering told UPI.
"It's important to use the right test in the right situation at the right time, and to thoroughly evaluate tests before use," said Pickering, a research associate in the department of infectious diseases at King's College London.
"Tests vary, and we found that some of the tests we trialed in our study gave very good results," she said.
Antibodies are cells produced by the human immune system to help the body fight infection. Those who recover from COVID-19 thus produce antibodies against the virus.
Researchers are evaluating the use of convalescent blood plasma donated by people who have recovered from the disease -- and transfused into infected patients -- as a possible treatment for COVID-19.
The assumption is that their antibodies would bolster the immune systems of patients battling the virus.
For the analysis, Pickering and her colleagues assessed the accuracy of the 10 testing platforms using 110 blood samples collected from 87 confirmed COVID-19 patients and 50 pre-pandemic virus-negative samples.
They developed their own sensitive and specific antibody assay and used it to conduct "unbiased, head-to-head comparisons of the 10 testing platforms.
All of the tests evaluated gave the best results when used on patients 20 days or more after the start of symptoms, with most accurately identifying samples as positive for antibodies 95% of the time, the researchers said.
In addition, antibody levels were higher in individuals with severe illness compared to those with asymptomatic or mild disease, they said.
"We wanted to perform a thorough, transparent comparison of [COVID-19] antibody tests, in order for this to be useful to medical and diagnostics communities," Pickering said.
"We found that several antibody tests, including the type of rapid lateral flow tests that have received negative publicity in the past, performed well when used carefully."