Am. Schott et al., ULTRASOUND DISCRIMINATES PATIENTS WITH HIP FRACTURE EQUALLY WELL AS DUAL-ENERGY X-RAY ABSORPTIOMETRY AND INDEPENDENTLY OF BONE-MINERAL DENSITY, Journal of bone and mineral research, 10(2), 1995, pp. 243-249
We measured the heels of 43 women who had recently sustained a hip fra
cture and 86 age matched controls, using an Achilles(R) ultrasound dev
ice. Average BUA, SOS, and Stiffness were significantly lower in fract
ured patients (p < 0.0001). We also estimated ultrasound parameters fo
r patients as a function of controls and found the mean BUA to be -1.0
9 SD compared with controls, the mean SOS -0.89 SD, and the mean Stiff
ness -0.98 SD. Femoral BMD measured at the neck, Ward's triangle, and
the trochanter with a DPX Plus(R) was also significantly lower in frac
tured patients (p < 0.0001). The increased risk of hip fracture associ
ated with low ultrasound values was estimated with logistic regression
analysis for each bone parameter, adjusted for height and weight. The
adjusted regression coefficients associated with BUA, SOS, Stiffness,
and BMD were all significant (p < 0.0001) demonstrating the influence
of all ultrasound and DXA parameters on the risk of hip fracture. Aft
er adjusting the logistic regressions for BMD neck, BUA, SOS, and Stif
fness were still significant independent predictors of hip fracture. S
ensitivity and specificity of all measures were analyzed with the area
under the ROC curve which were for BUA, 0.77 +/- 0.04; for SOS, 0.75
+/- 0.04; for Stiffness, 0.78 +/- 0.04; and for BMD, 0.74 +/- 0.04. We
determined the range for the best compromise between sensitivity and
specificity of BUA, 97-98 dB/MHz; SOS, 1482-1487 m/s; Stiffness 59-62%
Young Adult; and of BMD, 0.64-0.69 g/cm(2). The area under the ROC cu
rves of BUA, SOS, Stiffness, and DXA were compared and no statisticall
y significant difference was found. Patients with trochanteric fractur
es had significantly lower ultrasound parameters than those with cervi
cal fractures (SOS and Stiffness p = 0.001, BUA p = 0.009), whereas no
significant difference was found for DXA parameters (BMD neck, BMD Wa
rd, and BMD trochanter p greater than or equal to 0.5). Some previous
studies showed a lower bone density in patients with trochanteric frac
tures and suggested that the two types of fractures might correspond t
o a different process. Our results suggest that ultrasound is better c
orrelated to the type of fracture than DXA, discriminates subjects wit
h hip fracture equally well as DXA, and does indeed provide an indicat
ion of fracture risk independent of BMD.