Introduction: Computed tomography (CT) is considered the method of choice for detecting rotational malalignment of the femur. However, it is unclear how reliable the method is, and what the causes are of potential inaccuracies.
Materials and methods: To address these issues three observers measured the CT images of the femur of 76 patients on two separate occasions. The images were made during follow-up of a unilateral femoral shaft fractures. Rotational malalignment was determined by comparing the torsion angle of the injured to the noninjured leg.
Results: The pooled intraobserver variance was 3.9 degrees and interobserver variance 4.1 degrees. Of the two measurements of one observer 95% were up to 10.8 degrees different, and between observers 95% of the measurements were up to 15.6 degrees different.
Conclusions: CT measurements of rotational malalignment of the femur are not accurate. This is due principally to the difficulty in defining a line through the axis of the femoral neck. The accuracy can be improved by taking the average of two measurements.