This analysis extends prior work by Magaziner and colleagues that showed significantly lower BMD of the contralateral hip and greater decline in the year following fracture in older women with hip fracture than in age-matched controls [12
]. The continuing BMD decline is alarming because the un-fractured hip was in all likelihood as fragile as the one that did fracture. The present study takes a closer look at that data by examining the underlying femur geometric strength and its trajectory of change to obtain a clearer perspective on the mechanical consequences of the BMD decline. As in the prior study, the results were compared to age matched controls drawn from participants in the Study of Osteoporotic Fractures (SOF) who had not suffered a prevalent or incident hip fracture
In the cross-sectional comparison we observed a number of important geometric differences at post-fracture baseline between BHS3 fracture cases and SOF controls. Our goal was to gain some insight into how much structurally weaker the femurs of BHS3 women were compared to age matched controls who do not fracture their hips. The adjusted BMD differences in BHS3 women averaged 9–14% lower at the three HSA regions than controls. The underlying mechanical implications of the BMD differences however are complicated by the opposing effects of OD on BMD and on bending strength (SM). In the definitive study of femur geometry in the SOF by Kaptoge et al, the ability of BMD to predict hip fracture could not be explained by mass differences alone (BMC or CSA) but required the additional decrement due to a wider bone diameter. Paradoxically a wider bone should be stronger (higher SM) but femoral necks were significantly wider in BHS3 compared to SOF controls consistent with the findings of Kaptoge and colleagues in SOF and Rivadeneira and colleagues in the Rotterdam study [18
]. We did not observe wider femurs in BHS3 at the intertrochanter and shaft regions which may possibly reflect differences in fracture populations between these samples/studies. In neither the Kaptoge nor the Rivadeneira papers did section modulus predict fracture as well as BMD as would be expected from engineering beam theory, suggesting that the wider diameter may tend to preserve SM at the expense of cortical stability. In the present study the largest differences between BHS3 and SOF controls were in the buckling ration (BR), as was also evident in the SOF and Rotterdam studies [18
]. The rates of change in BR were greater than that of any other parameter in BHS3 compared to SOF controls suggesting that buckling susceptibility continues to increase (as is also evident in BMD decline) in the year following fracture.
In the longitudinal comparisons, the BHS3 women showed significant declines in SM (although not significant at NN and shaft), BMD, CSA and significant increases in BR in the year following fracture. Outer diameter showed apparent increases at all regions but only reached significance in the 1 year period at the shaft where method precision is highest [20
]. These results are consistent with adaptive changes to reduced loading activities in the post fracture period among BHS3 women. In SOF controls, decline in BMD at the NN region was entirely due to change in OD; loss of NN bone (CSA) was non-significant. At the IT region, the decline in BMD was due to a combination of significant loss of CSA and a significant increase in OD. These opposing changes appear to preserve SM.
It is also important to note that there was little evidence of recovery among BHS3 women and the decline of some parameters appears to be greater in the first 6 months following fracture (–). Although there appears to be some recovery between 6 and 12 months, a full recovery is not achieved and the femur is geometrically weaker compared to baseline. This is consistent with findings from a study by Mikkola et al. that showed bone mineral mass and geometry of the tibia in women and men who sustained a hip fracture were systematically lower than non-hip fracture controls even an average of 3.5 years post- fracture [21
]. These trends may be explained by the increase in bone turnover observed after fracture, perhaps secondary to a reduction in loading activity. Studies of biochemical markers of bone turnover showed that bone remodeling is significantly increased after fracture where resorption exceeds bone formation during the first 4–7 months following fracture, and though more consistent, bone markers continue to be elevated several years after fracture. These trends are most pronounced in hip fracture cases and may be further confounded by immobilization of the lower limb that typically follows this type of fracture [22
]. The continued elevation of remodeling that occurs long after the fracture may lead to additional loss of bone mass and strength that predisposes women to a higher risk of future fracture. In contrast, a longitudinal study by Dirschl et al. reporting change in BMD over 6 years following fracture showed that BMD lost during the first year after fracture was completely recovered after 6 years, and was even higher than it was at the time of fracture [26
]. However results were from a small sample of men and women and it is not clear if these long term trends would be similar in indices of bone strength.
Changes in femur geometry over time reflect adaptation to prevalent loading conditions, and the small changes among SOF controls suggest that their loading conditions remained relatively constant when no fracture occurs. This is clearly not the case in the recovery period following fracture, where loading is significantly diminished and when faster rates of change in CSA and SM following fracture are detected. Adaptation to load on the femur may be more evident in bone geometry than in BMD, which has important implications for rehabilitation strategies post-fracture [27
]. In addition, several pharmacologic intervention studies including treatment with PTH and Raloxifene have shown greater changes in bone geometry than in BMD [29
]. Primary outcomes for rehabilitation interventions post-fracture include improved functioning, but little is known about the effects of physical rehabilitation on BMD or bone strength. It may be that a combination of exercise and pharmacologic treatment will yield the most benefit.
Although the results of this study are consistent with the previous analyses of contralateral hip BMD among these women who have suffered a hip fracture, the annual BMD loss reported previously was up to two times greater than the BMD loss observed in this analysis [12
]. This may in part be due to differences in the analytic samples. However, differences in BMD loss between the two analyses are most likely due to the difference in methods used to measure BMD. BMD in the previous analysis was measured directly by conventional scanner software whereas in the present analysis, BMD was measured using HSA software. The HSA narrow neck region is roughly comparable to the conventional femoral neck; correlations between conventional DXA and HSA BMD measures are high although they are not equivalent due to difference in neck location and in algorithm edge detection. We ran the same models using DXA measured BMD and found the results to be significant and comparable to BMD measured by HSA (DXA BMD: −3.2% in BHS3 vs. −0.6% in SOF; HSA BMD: −2.8% in BHS3 vs. −0.8% in SOF).
There were several limitations to this study. Although we were able to adjust for height, weight, and age at baseline, we were unable to account for additional factors that may influence change in bone strength following a hip fracture. It could be that femur weakening was exacerbated if women who suffered a hip fracture were sicker (had multiple co-morbidities) than their age-matched controls. However in the previous analysis of BMD change the year after fracture, neither co-morbidities nor disability as measured by difficulty with ADLs were found to be confounders and were left out of final models. In addition, we were unable to accurately account for the influence of bone altering medication such as hormone replacement therapy (HRT), which was still a common therapy at the time women in these studies were scanned. HRT use was self reported in the BHS3 and SOF studies and women in either study could only be categorized equivalently as ever taking HRT or not at all. We found that a significantly larger proportion of women who recently fractured their hip had ever taken HRT compared to SOF controls, but after adjustment, we found that results did not change. One cannot be certain that the contralateral hip was as fragile as the hip that fractured; moreover changes in mechanical load due to altered stance on the intact hip during the recovery period may in part be responsible for some of the changes in geometry evident here. These complex effects cannot be evaluated with the limited data in this study. Finally, in order to make comparisons at 2, 6, and 12 months, we assumed a linear trajectory over 4 years for the SOF cohort as estimated from individual changes in the SOF subjects during that time span. Although one should be cautious when interpreting these results, the estimate decline in BMD over 12 months obtained in our study is similar to prior reports of BMD change over time [32
]. While rates of change may vary with age, the average BMD decline observed in older women from a population based cohort study adjusted for age and BMI was −0.0045 g/cm2
per year which is similar to the estimates we obtained for SOF controls in this analysis [33
]. Other studies reported a decline in femoral BMD between 0.9 and 1.43% which is also similar to the estimates we obtained [32
]. However, if trajectories are not linear in this sample, our models may be over or underestimating the true difference between women with and without a hip fracture.
In conclusion, we have shown that geometric strength of the contralateral hip decreased significantly faster during the year following hip fracture compared to age matched controls from the SOF study who did not experience a hip fracture. The majority of the geometric decline in BHS3 women occurs within the first 6 months at the NN and IT regions where fragility fractures occur. Given the increased risk of suffering a second fracture in the contralateral hip, this analysis provides a potential explanation of the mechanisms by which these fractures occur. Results of the present study point to further loss of bone strength as another consequence of hip fracture that should be considered when identifying strategies for post fracture care.