Reliability of CT measurement of induced radioulnar step in dogs using a circle superimposition technique

To evaluate the reliability of CT measurement of canine elbow axial radioulnar congruence using a duplicated circle superimposition technique.

CT of six cadaveric canine elbows was performed following either: (1) placement of an external fixator on the radius with ostectomy and radial shortening of 1 to 5 mm; (2) radial lengthening of 1 to 5 mm using a radial lengthening system; or, (3) no axial length adjustment. Four observers, on two occasions, blinded to the amount of radial adjustment, placed a circle of best fit of the ulnar trochlear notch on a sagittal image that was duplicated and transposed to the articular contour of the radial head; then the axial distance between circles relative to the dorsal radius was measured. Statistical analysis assessed differences between actual and observer-assessed radial length adjustment; inter-/intra-observer interclass correlation coefficients and sensitivity/specificity were calculated to evaluate measurement reliability.

There was not a significant difference between actual and observer measures of radioulnar congruence. Inter- (intraclass correlation coefficient=0.98) and intra-observer reliability (intraclass correlation coefficient=0.94 to 0.99) were almost perfect for measurement of radioulnar incongruence when compared to the actual radial adjustment. A Bland-Altman plot revealed a 95% confidence interval range of -1.31 to 1.16 mm. Sensitivity and specificity were high for correct identification of congruent elbows and elbows with 1 or 2 mm incongruency.

The duplicated circle superimposition technique provides a simple and accurate means of assessing radioulnar axial incongruence.