Nonlinear association between bone mineral density and all-cause mortality: the Dong-gu study.

ABSTRACT

There was a U-shaped association between hip BMD and all-cause mortality, with the lowest mortality in the 90th percentile in males. However, there was an inverse linear relationship in females. In contrast, the association between lumbar spine BMD and mortality was less evident in males, with no association in females. INTRODUCTION: Bone mineral density (BMD) is reported inversely associated with mortality. Although some previous studies provided evidence for nonlinear associations, these were not adequately assessed in most previous works. METHODS: We evaluated the nonlinear relationship between BMD and mortality in Asians. Our study involved 8629 participants in the Dong-gu study from 2007 to 2010. Cox proportional hazard regression was used to calculate hazard ratios (HRs) according to BMD categories after adjusting for potential confounders. During a follow-up of 6.7 ± 1.4 years, 712 participants died. RESULTS: There was a U-shaped association between hip BMD and all-cause mortality, with the lowest mortality in the 90th percentile in males. However, there was an inverse linear relationship in females. In males, compared with the 75th to 95th percentile group, the < 2.5th percentile group had a 3.89 (95% CI 2.41-6.28)-fold higher risk and the 2.5th to 5th percentile group had a 2.51 (95% CI 1.25-5.04)-fold higher risk. The HR was 2.51 (95% CI 1.25, 5.04) in the > 97.5th percentile group. In females, compared with that in the 75th to 95th percentile group, the HR was 2.33 (95% CI 1.24, 4.39) in the < 2.5th percentile group. In contrast, the association between lumbar spine BMD and mortality was less evident in males, with no association in females. CONCLUSION: In conclusion, this study shows that the association between BMD and mortality varies by gender and that high and low BMD are predictors of all-cause mortality in males.