Does bone density predict fractures comparably in men and women?

Abstract

At the recent National Institutes of Health Consensus Development Conference on Osteoporosis Prevention, Diagnosis, and Therapy (http://odp.od.nih.gov/consensus/cons/111/111_statement.htm), there was some tension between two concepts that, upon first glance, appear contradictory: First, there is a greater relative risk of fracture per standard deviation (SD) change in bone mineral density (BMD) in women than men; and, second, the absolute risk of fracture in women and men with a given BMD level is the same. Are both of these statements true? And, if so, are these two notions really mutually exclusive? Resolution of this issue has a bearing on whether World Health Organization diagnostic criteria for osteoporosis should be the same for men as for women and whether gender should be a consideration in new approaches to bone density assessment that focus on absolute fracture risk [1–3]. In this Brief Review, we consider the available evidence bearing on these two questions and suggest that the two concepts are not contradictory. We also specify additional data that should help clarify several unresolved issues that remain. That the relative increase in fracture risk per 1 SD decline in areal BMD (g/cm) is greater in women than men has been suggested by a number of studies where the two sexes were assessed comparably [4–7]. For example, in a population-based case–control study in Rochester, MN, the age-adjusted relative risk of an osteoporotic fracture (hip, spine or distal forearm due to minimal or moderate trauma 535 years of age) was 2.4 per 1 SD decrease in femoral neck BMD in women but only 1.1 in men [7]. Other investigators have found more comparable sex-specific relative risks for osteoporotic fractures in general [8] and for spine [9–11] and hip fractures [12,13] in particular. In the Rotterdam study, for example, the age-adjusted relative risk of hip fracture per 1 SD decrease in femoral neck BMD was 2.5 for women and 3.0 for men [12]. However, the standard deviation for subjects 570 years of age, in whom most hip fractures occur, was substantially greater in the men (0.141–0.143 g/cm) than the women (0.128–0.130 g/ cm) in Rotterdam. By contrast, the Rochester analysis employed standard deviations from normal 20to 29year-old subjects, and these happened to be less in men than women at the femoral neck (0.109 g/cm vs 0.119 g/ cm). Thus, sex-specific fracture risks per 1 SD change are partly dependent upon the size of the standard deviations, and it appears that the studies which found equivalent sex-specific relative risks all used larger standard deviations for men than for women when making the calculations. The use of different standard deviations may therefore confound direct comparisons of men and women, but these relative risks can be difficult to interpret and they are not closely correlated with absolute fracture risk in any event [14]; the latter (see next) may be more relevant clinically. Data also exist which show that absolute fracture risk is comparable in women and men who have the same BMD level. In the Hawaii Osteoporosis Study, for example, the incidence of new morphometric vertebral fractures over 6–10 years of follow-up was the same for women and men with a given calcaneus BMD (Fig. 1) and similar results were seen at the distal forearm [10]. The risk of fractures generally over 5 years of follow-up was also practically identical among elderly Finnish men and women who had the same size-adjusted calcaneus BMD level, although women were likely to experience a greater number of fractures [15]. Similarly, when fracture risk was estimated from baseline femoral neck BMD in the Rotterdam study, annual hip fracture incidence over Osteoporos Int (2001) 12:707–709 2001 International Osteoporosis Foundation and National Osteoporosis Foundation Osteoporosis International