Age, race, weight, and gender impact normative values of bone mineral density.

BACKGROUND: Osteoporosis and fractures represent a major public health issue. Accurate normative reference bone mineral density (BMD) values are vital for diagnosing osteoporosis. The generalizability of the T-score method across gender, race, and age in clinic decision-making has been debated. Our aim was to identify the best statistical model to derive normative BMD values in both men and women in the multiethnic United States population.
METHODS: The Third National Health and Nutrition Examination Survey was used as a data source. Gender- and race/ethnicity-stratified data analyses and modeling were conducted on 9779 persons (ages 20 to 65 years) who reported no conditions or medications likely to affect bone metabolism. Sampling and design effects were addressed using STATA 10. Model comparisons were conducted by partial F tests and residual plots.
RESULTS: Polynomial regression provided a statistically significant better fit than linear regression in predicting normative BMD in both men and women. Age-centered polynomial models provided the best model for predicting normative BMD values.
CONCLUSION: The gender- and race-specific lower limit of normal values obtained created a new classification method of low BMD, which might mitigate some of the T-score limitations in men and minority populations.