Increase in Serum Testosterone following Contact with Blood Cells

Whilst validating a testosterone assay, the normal range for females was found to be substantially higher than that derived when the original method was developed (M J Wheeler, St Thomas' Hospital, London). There was an excellent agreement between the results obtained from 13 samples with serum testoster­ one in the range 0·8-4·0 nmol/L, which were analysed by both laboratories (r=0·992, n=13). This indicated that the higher normal range was not due to an assay bias. Subsequent investiga­ tions, described here, strongly suggest that the higher normal range results from a difference in specimen handling. Blood was collected from eight females attending a subfertility clinic and from five female laboratory staff. Each blood sample was divided between two plain glass tubes and, on receipt in the laboratory, the serum from one tube was separated and stored at - 20°C until analysis. The other tube was left at room temperature (2o-25°C) for between 1 and 5 days before separation of serum and storage at -200C. Testosterone was measured by radioimmu­ noassay following ether extraction of the