OBJECTIVE: To establish reference intervals for serum free testosterone for DPC's Free Testosterone assay. METHODS: We used data from healthy subjects and patients to determine reference intervals by parametric and non-parametric methods after partitioning by sex and age. RESULTS: In males, there was a significant decrease in free testosterone concentrations with age. Reference intervals derived from a combination of 2075 "healthy" and patients' results gave similar values by parametric and nonparametric methods. However, the subgroups failed the test for Gaussian distribution. For each decade from 20 years on and > or = 60 years, the intervals based on 2.5th and 97.5th percentiles were: 24.1-94.8, 25.0-89.3, 23.4-81.7, 22.5-80.4, and 21.5-74.3 pmol/L respectively, in females, there was little change with age. The frequency distribution of 1915 patients was positively skewed, and showed a wider range than "healthy." Using square roots of values gave a Gaussian distribution. The central 95% intervals based on 187 "healthy" subjects were: 0.5-8.1 and 0.0-6.4 pmol/L for 20-59 and > or = 60 years, respectively. CONCLUSION: Developing reference intervals for free testosterone was complicated by the need to partition data by gender and age, difficulty in establishing disease in subjects and presence of physiological and other factors which can affect concentration in health and disease.
OBJECTIVE: To establish reference intervals for serum free testosterone for DPC's Free Testosterone assay. METHODS: We used data from healthy subjects and patients to determine reference intervals by parametric and non-parametric methods after partitioning by sex and age. RESULTS: In males, there was a significant decrease in free testosterone concentrations with age. Reference intervals derived from a combination of 2075 "healthy" and patients' results gave similar values by parametric and nonparametric methods. However, the subgroups failed the test for Gaussian distribution. For each decade from 20 years on and > or = 60 years, the intervals based on 2.5th and 97.5th percentiles were: 24.1-94.8, 25.0-89.3, 23.4-81.7, 22.5-80.4, and 21.5-74.3 pmol/L respectively, in females, there was little change with age. The frequency distribution of 1915 patients was positively skewed, and showed a wider range than "healthy." Using square roots of values gave a Gaussian distribution. The central 95% intervals based on 187 "healthy" subjects were: 0.5-8.1 and 0.0-6.4 pmol/L for 20-59 and > or = 60 years, respectively. CONCLUSION: Developing reference intervals for free testosterone was complicated by the need to partition data by gender and age, difficulty in establishing disease in subjects and presence of physiological and other factors which can affect concentration in health and disease.