Am. Schott et al., How hip and whole-body bone mineral density predict hip fracture in elderly women: The EPIDOS prospective study, OSTEOPOR IN, 8(3), 1998, pp. 247-254
We conducted a population-based cohort study in 7598 white healthy women, a
ged 75 years and over, recruited from the voting lists. We measured at base
line bone mineral density (BMD g/cm(2)) of the proximal femur (neck, trocha
nter and Ward's triangle) and the whole body, as well as fat and lean body
mass, by dual-energy X-ray absorptiometry (DXA). One hundred and fifty-four
women underwent a hip fracture during an average 2 years follow-up. Each s
tandard deviation decrease in BMD increased the risk of hip fracture adjust
ed for age, weight and centre by 1.9 (95% CL 1.5, 2.3) for the femoral neck
, 2.6 times (2.0, 3.3) for the trochanter, 1.8 times (1.4, 2.2) for Ward's
triangle, 1.6 times (1.2, 2.0) for the whole body, and 1.3 times (1.0, 1.5)
for the fat mass. The areas under the receiver operating characteristic (R
OC) curves were not significantly different between trochanter and femoral
neck BMD, whereas ROC curves of femoral neck and trochanter BMD were signif
icantly better than those for Ward's triangle and whole-body BMD. Women who
sustained an intertrochanteric fracture were older (84 +/- 4.5 years) than
women who had a cervical fracture (81 +/- 4.5 years) and trochanter BMD se
emed to be a stronger predictor of intertrochanteric ([RR = 4.5 (3.1, 6.5)]
than cervical fractures ([RR = 1.8 (1.5, 2.3]).
In very elderly women aged 80 years and more, hip BMD was still a significa
nt predictor of hip fracture but the relative risk was significantly lower
than in women younger than 80 years.
In the 48% of women who had a femoral neck BMD T-score less than -2.5, the
relative risk of hip fracture was increased by 3, and the unadjusted incide
nce of hip fracture was 16.4 per 1000 woman-years compared with 1.1 in the
population with a femoral neck BMD T-score greater than or equal to-1.