While the risk of osteoporotic fractures, especially hip fractures, has consistently been reported to be higher in those with low BMI, fracture risk in overweight and obese individuals has not been well characterized. In this large prospective study of older men who were normal weight or heavier, most hip (62%) and nonspine fractures (68%) occurred in overweight or obese men, who represented 72% of the cohort. We also found that obesity was associated with an increased risk of hip and other fractures after adjustment for BMD, but this association appeared to be due at least in part to confounding by deficits in physical performance. Similar deficits have been linked to fall rates(27
) and might mediate the obesity-fracture relationship. However, adjustment for a history of falls in the previous year, which has been consistently predictive of incident falls,(28
) did not alter associations. In light of our finding that many fractures occur in men of higher BMI, public health and clinical strategies must be developed to identify overweight and obese individuals at greatest risk. Moreover, since obesity appears to impart particular risk for fracture after BMD adjustment, the mechanisms that are responsible must be determined.
Whereas some previous analyses have noted a tendency for increased weight to be associated with higher fracture risk,(6
) the MrOS study design made it possible to examine fracture and BMI associations in ranges more typical of in men in the United States Very few MrOS men were in the clinical underweight category (<18.5 kg/m2
= 6), and the majority of participants were overweight or obese, which is reflective of US men in this age group.(26
) Many other studies that previously examined the association between fracture and BMI have been characterized by relatively lower weights, and the mean baseline BMI in MrOS (27.4 kg/m2
) is higher than the mean for men in a recent meta-analysis of seven studies of BMI and fracture(4
) that focused on the relationship of low weight and fracture risk. Our findings are not incompatible with studies that have reported a higher risk of fracture for men with low BMI than for men with a normal BMI. In fact, in the meta-analysis of BMI and fracture in men, the association was observed to be nonlinear, with greatest risk in low BMI and no additional reduction of risk among men with greater than normal BMI.(4
) Whereas the importance of low weight as a risk factor for fracture should not be ignored, the proportion of men 70 to 79 years of age in the United States who are underweight is small [<5% from the National Health and Nutrition Examination Survey (NHANES)], and more than half are overweight or obese.(26
) The chronic disease risks associated with obesity have long been recognized, and our results support the conclusion that clinical and public health messages for obesity prevention should be compatible with fracture-prevention messages.
BMD is moderately positively correlated with BMI, and low BMD has been considered one of the major causes of the increased risk of fracture in those with low weight. Similarly, higher BMD in larger people is posited to mitigate fracture risk. On the other hand, the correlation between BMI and BMD is moderate, many overweight and obese individuals have relatively low BMD, and bone strength may not increase in proportion to increases in total or fat mass.(31
) Moreover, fat gain has been associated with higher rates of BMD loss,(32
) and visceral fat in particular has been shown to be negatively correlated with bone structure and strength.(33
) Thus increased adiposity may have deleterious effects on bone strength that could be important in determining fracture risk.
In addition to the possible adverse effects of fat on bone strength, our results suggest that obesity also imparts a considerably higher risk of hip and other fractures after BMD adjustment. Some data suggest that obesity may impair physical function and increase the risk of falls.(34
) On the other hand, in women, increased adiposity may protect against hip fracture by reducing the force exerted on bone in a fall.(36
) However, we reported previously that soft tissue thickness around the hip did not protect against fracture in men.(37
) This observation suggests the possibility of gender differences in the relation of excess adiposity to fracture risk.
In obese men, disruption of the hypothalamic-pituitary axis, including androgen deficiency, has been reported, and higher BMI is be associated with lower serum vitamin D levels.(38
) These are plausible mediators of an obesity-fracture association; however, only a subset of men in the MrOS cohort had these measures available. Furthermore, we have reported previously that low testosterone was not associated with increased fracture risk in the MrOS cohort.(40
) We also reported an increased risk of hip fracture in men with low serum vitamin D that was attenuated with adjustment for BMD.(41
) Further investigation into neuroendocrine alterations in obesity as they relate to fracture risk may provide additional insight.
The limitations of this study include the use of a relatively healthy volunteer cohort in which there were few underweight men, a group that is known to be at particularly high risk of fracture. While this prevented us from examining the risk of underweight in our analyses, in view of the weight distribution in the US population of older men, we were most interested in the effects of overweight and obesity. Also, 90% of men were non-Hispanic whites, which may limit generalizability. Different associations with body composition and fracture between women and men have consistently been reported, and our conclusions cannot be applied to fracture risk in women. Weight loss has been shown to be associated with BMD loss(42
) and increased hip fracture risk,(43
) but we lacked detailed information about weight and weight change prior to enrollment. As in any association study, there may be unmeasured confounders that explain the effect of BMI on both BMD and fracture. Finally, although we examined associations with fractures in different regions of the body, we lacked power to further dissect fracture sites. For example, associations may differ by whether the hip fracture is trochanteric(44
) or by whether an arm fracture occurs in the forearm versus the proximal humerus.(9
In summary, in this large cohort of older men, most of whom had a BMI above normal, most hip and other nonspine fractures occurred in those who were overweight or obese, which reflected the distribution of BMI in the cohort. In addition, obesity was an important contributor to the risk of fracture, including hip fracture, among men with similar BMD. The combination of hip fracture and obesity, both of which adversely affect physical function,(45
) may be particularly likely to lead to disability or institutionalization.(45
) The interplay of BMI, BMD, physical performance, and fracture requires additional study in the context of a growing population of overweight and obese older men in the United States.