An accessible hip DXA scan tracked the strength of screws placed in the neck of the femur about as closely as newer bone-quality assessments in a laboratory study of cadaver bones. Femoral-neck DXA BMD, a measure of bone mineral density, and the 3D-Shaper reading had the highest reported correlation with screw pullout strength, with a correlation coefficient of 0.95 and a reported p-value below 0.001. The authors interpret the result as evidence that hip-specific DXA performed as well as the newer imaging approaches. Because the test used cadaveric femurs, the finding describes a mechanical association in bone rather than a patient-level surgical rule.
How the test was run
The research team used 10 paired proximal femurs, for 20 femurs in total, from donors aged 56 to 96. One male and one female donor were represented in each decade, and the specimens had no known conditions expected to alter bone mineral density. Each femur underwent all of the study’s bone-quality assessments, including hip-specific DXA, TBS Ortho, 3D-Shaper and qCT. The researchers then placed a cannulated femoral-neck screw using a standardized procedure, recorded insertional torque digitally, and mechanically tested the screw.
In the pullout test, each screw was drawn out at 5 millimeters per second. Failure was marked by a force drop of at least 20 percent from the maximum, and the peak force before that drop was used as the measure of screw strength. Screws within a given femur were tested in random order to limit loosening. The authors planned the sample around 10 paired femurs and 20 specimens, saying that with 80 percent power the design could detect a correlation of at least 0.44 as statistically significant.
Several measures moved with strength
Several measures moved closely with pullout force. DXA at the proximal femur had a correlation coefficient of 0.92; qCT at the femoral head, 0.89; 3D-Shaper at the proximal femur, 0.88; TBS Ortho at the proximal femur, 0.85; and qCT at the femoral neck, 0.82. The paper reports that all of these imaging correlations had p-values below 0.001 after adjustment for multiple comparisons. A positive correlation here means specimens with higher readings tended, in the study’s ranking, to require more force before the screw pulled out.
Looking at the share of variation accounted for by a mixed-model estimate produced a similar picture. The 3D-Shaper measure at the femoral neck had the highest reported value, accounting for 94.2 percent of peak pullout-force variance, with a reported pseudo-R2 of 0.94. DXA BMD at the femoral neck had a pseudo-R2 of 0.93, and qCT at the femoral head had 0.92. DXA at the proximal femur, qCT at the femoral neck, 3D-Shaper at the proximal femur and TBS Ortho at the proximal femur each had a reported pseudo-R2 of 0.91. These are model-based estimates of explained variance, and the paper reports no confidence intervals for the pseudo-R2 values.
Insertional torque also tracked pullout strength. The torque reading had a Spearman coefficient of 0.86 and a reported pseudo-R2 of 0.89, with the model accounting for 89.2 percent of the variation in peak pullout force. That result concerns the mechanical test itself. It does not show that using torque readings changes fixation failure or patient outcomes.
What the study cannot answer
The authors’ broader conclusion was that hip DXA BMD performed as well as TBS Ortho, 3D-Shaper and qCT, and that its accessibility could matter for surgical decision-making. That conclusion is about how the measures compared in this experiment. It does not establish that one modality is superior or equivalent for predicting fracture risk in patients.
The limitations narrow the result further. The experiment assessed screw pullout, but not push-through, shear or cutout, and cyclical push-through testing was abandoned after the cannulated screws bent under the required forces. There was also no standardized cadaveric DXA measurement approach, while TBS Ortho and 3D-Shaper analyses required manual modifications. The work therefore leaves open whether the same relationships hold in living patients or in other clinical settings.
Paper data and sources
Original title: Predicting femoral neck screw pullout strength using DXA and novel assessments of bone microarchitecture.
Authors: Jarod Moyer, Richard Behlmer, Diane Krueger et al.
Journal/Repository: Archives of osteoporosis
Status: Peer-reviewed
First online: 2026-08-21
DOI: 10.1007/s11657-026-01763-0
Original paper · Full text