To compare three fluoroscopic methods for determining femoral rotation.

Native femoral version was measured by computed tomography in 20 intact femurs from 10 cadaveric specimens. Two Steinmann pins were placed into each left femur above and below a planned transverse osteotomy which was completed through the diaphysis. Four surgeons utilized the true lateral (TL), neck-horizontal angle (NH), and lesser trochanter profile (LTP) techniques to correct the injured femur's rotation using the intact right femur as reference, yielding 120 measurements. Accuracy was assessed by comparing the angle subtended by the two Steinmann pins before and after manipulation and comparing against version measurements of the right femur.

Absolute mean rotational error in the fractured femur compared to its uninjured state was 6.0° (95% CI, 4.6-7.5), 6.6° (95% CI, 5.0-8.2), and 8.5° (95% CI, 6.5-10.6) for the TL, NH, and LTP techniques, respectively, without significant difference between techniques (p = 0.100). Compared to the right femur, absolute mean rotational error was 6.6° (95% CI, 1.0-12.2), 6.4° (95% CI, 0.1-12.6), and 8.9° (95% CI, 0.8-17.0) for the TL, NH, and LTP techniques, respectively, without significant difference (p = 0.180). Significantly more femurs were malrotated by >15° using the LTP method compared to the TL and NH methods (20.0% vs 2.5% and 5.0%, p = 0.030). Absolute mean error in estimating femoral rotation of the intact femur using the TL and NH methods compared to CT was 6.6° (95% confidence interval [CI], 5.1-8.2) and 4.4° (95% CI, 3.4-5.4), respectively, with significant difference between the two methods (p = 0.020).

The true lateral (TL), neck-horizontal angle (NH), and the lesser trochanter profile (LTP) techniques performed similarly in correcting rotation of the fractured femur, but significantly more femurs were malrotated by >15° using the LTP technique. This supports preferential use of the TL or NH methods for determining femoral version intraoperatively.