Reliability of a technique for determining sagittal knee geometry from lateral knee radiographs

The purposes of this study were to describe an analytical technique for determining selected 2-dimensional geometric characteristics of a knee joint using lateral knee radiographs and to examine the inter- and intra-analyst reliability of this technique. Five lateral knee radiographs of different knee flexion angles (25–85° at intervals of 15°) were obtained from five subjects. Two graduate and five undergraduate students with knowledge of knee anatomy served as the analysts and were asked to identify certain landmarks from the radiographs. The coordinates of these landmarks were used to determine the effective moment arm of the quadriceps force, patellar tendon length, patella height, patellar mechanism and patellar tendon angles, and tibiofemoral and patellofemoral joint spaces. For each radiograph, intraclass correlation coefficients (ICCs) were computed for combinations of 2–7 analysts. Using all seven analysts, the ICCs ranged from 0.9967 to 0.9985 for different radiographs. When fewer analysts were used, the average ICCs were 0.9975 (6 analysts), 0.9974 (5 analysts), 0.9974 (4 analysts), 0.9974 (3 analysts), and 0.9973 (2 analysts). Four of the analysts re-analyzed the radiographs 2–3 months after the initial analyses. Intra-analyst ICCs ranged from 0.9842 to 0.9999. Overall, the high ICC values indicate excellent inter- and intra-analyst reliability. The proposed technique is reliable and may be used for both clinical and research purposes. The relatively small reductions in ICCs when fewer analysts were used suggest that a single experienced analyst is sufficient for clinical assessment. However, 2–3 analysts are recommended for research purposes.