Non-linear pattern of age-specific bone mineral density and related factors in women: data from the 5th Korea National Health and Nutrition Examination Survey.

Abstract

Age-specific bone mineral density (BMD) of Korean women showed a nonlinear pattern. According to our study, this pattern could be mainly attributed to the nonlinear effects of body composition. Interventions targeting these factors in premenopausal women could be critical for achieving peak BMD. The patterns of age-specific bone mineral density (BMD) of femur and spine in women showed a non-unidirectional decrease, with ups and downs around the point of achieving peak BMD, that are not evident in men. We investigated the factors associated with these age-specific BMD in women using linear and nonlinear models. A total of 4681 women recruited during the 5th Korean National Health and Nutrition Examination Survey 2010-2011 (KNHANES V) were included. A range of related factors were considered including body compositional, reproductive, and socio-behavioral factors. Age-specific BMD was demonstrated using a generalized additive model (GAM) with smoothing spline transformation. Age, lean body mass, fat mass, and age at menarche were significant predictors of BMD in premenopausal women. The GAM plot of BMD according to age revealed two peaks for total femur, femoral neck, and lumbar spine. After adjustment for body compositional factors, the peaks flattened. In the full model, the pattern for lumbar spine showed a single peak around the mid-40s. The GAM showed a generally better performance than the linear model for BMD at all three sites. Reduced models that included only lean body mass and fat mass best explained the variance of BMD in GAMs. The BMD of Korean women in reproductive period could be largely attributed to the nonlinear effects of age and body compositional factors. Intervention targeting these factors could be critical for maintaining peak BMD.