Here is a paper showing how these tests have 99% sensitivity in the lab, but this drops to 91% when the test is read in the field by a volunteer, and down to 75

Here is a paper showing how these tests have 99% sensitivity in the lab, but this drops to 91% when the test is read in the field by a volunteer, and down to 75% when the test is read by the subject themselves: https://archive.is/20081206110632/http://archfami.ama-assn.org/cgi/pmidlookup?view=long&pmid=9755740 https://archive.is/20081206110632/http://archfami.ama-assn.o... Moreover, if it's such that even a trace of a line indicates a high confidence of pregnancy, users may misinterpret a faint result as somehow meaning "kinda" pregnant.