Effect of radiographic malpositioning on patellar height measurements: implications for diagnosis and surgical planning. A retrospective observational study.

Abstract

Purpose: Patella alta stands as a significant predisposing factor for patellofemoral instability. Patellar heights indexes (PHI) require precise lateral knee radiographic projections for accurate computation. However, within clinical contexts, a notable proportion of radiographic images are taken with certain degrees of excessive rotation and/or tilting. The primary aim of this investigation was to assess the impact of suboptimal radiographic positioning on the determination of patellar height, utilizing the Blackburne-Peel (BP), Caton-Deschamps (CD) and Insall-Salvati (IS) indexes. Secondarily, it was evaluated whether any index is more sensible to suboptimal radiographic positioning, and how inter and intra observer reproducibility are affected.

Methods: Thirty-three patients with strictly lateral radiographs and another one demonstrating a noticeable degree of tilt and/or rotation were included in the study. Four orthopaedic surgeons specialized in knee surgery and four Orthopaedics residents conducted measurements on each radiograph in a randomized sequence at two different time points. A linear mixed-effects model was applied, with the quality of the radiograph (adequate projection vs malrotation, tilt, or both), observer expertise (consultant or resident), and observation time regarded as fixed effects, while consultant and patient were treated as random effects.

Results: Statistically significant differences were obtained between strict lateral and tilted radiographs in the BPI, with an overestimation of up to 0.0937; between strict lateral and malrotated radiographs in the ISI, showing an overestimation of up to 0.0696 and between tilted and/or rotated radiographs in the CDI, with overestimation reaching up to 0.0813. No significant differences were observed between resident and consultant observers in any of the indexes. Good inter-observer consistency was achieved.

Conclusion: This study showed statistically significant differences in the determination of the three PHIs. Although these differences were small, they may predispose to diagnostic errors and inaccurate surgical planning in cases requiring surgical correction. Future studies quantifying the degrees of malrotation and/or inclination may further clarify these findings.