Effect of female database use for T-score derivation in men

Whether to use male or female databases to obtain T-scores in men remains controversial. This study evaluated the impact of deriving male T-scores using female databases in 350 men aged 22.8-93.5 (mean 67.5+/-12.2) yr who were referred for clinically indicated dual-energy X-ray absorptiometry exams. Spine, femur, and nondominant radius scans were obtained in routine clinical manner using a GE Healthcare Lunar Prodigy densitometer. Analyses were performed using software version 9.30. Initially, the GE Healthcare Lunar male normative database was used to calculate T-scores. Subsequently, scans were reanalyzed using female databases; GE for the spine and radius, and NHANES III for the femur. Using the manufacturer's male database, T-scores (mean [range]) of the L1-4 spine, femur neck, total femur, and .3 radius were 0.0 [-4.6 to +8.5], -1.6 [-4.3 to +2.3], -1.1 [-4.0 to +3.3], and -0.7 [-5.3 to +2.9], respectively. On reanalysis with female databases, T-scores "improved" (p<0.0001) with a positive bias of 0.34, 0.33, 0.58, and 1.20, respectively at the above 4 sites. Using female databases, the proportion of men classified as having normal bone mass increased from 22% to 33% and those identified as osteoporotic decreased from 29% to 17%. If pharmacologic treatment were prescribed at a T-score <-2.0, use of the female databases would reduce those treated for low bone mass from 46% to 32%. In conclusion, using female databases to derive male T-scores results in "improvement" of diagnostic classification for a substantial number of men with fewer being classified as having low bone mass.