Analytical versus clinical sensitivity and specificity in pregnancy testing.

A typical urine pregnancy test with the enzyme-linked immunoassay for human chorionic gonadotropin (Tandem Icon human chorionic gonadotropin) was performed as was a qualitative serum radioimmunoassay for human chorionic gonadotropin when the positive cutoff for the latter was set at 30 IU/L. There was 99.5% concordance between the two methods when performed on 871 simultaneous urine and serum specimens. Although the serum radioimmunoassay can detect levels of human chorionic gonadotropin as low as 10 IU/L, in this study the majority of patients with serum levels between 10 and 30 IU/L were not pregnant on clinical follow-up. Of nine patients with adequate clinical follow-up, a negative urine pregnancy test result, and a serum human chorionic gonadotropin level between 10 and 30 IU/L, only one proved to be pregnant. The Tandem Icon human chorionic gonadotropin urine test is sufficient for routine pregnancy testing and to rule out the need for immediate intervention in cases of ectopic pregnancy.