Is the Clinician's Eye a Valid and Reproducible Tool for Diagnosing Patella Alta on a Lateral Knee Radiography?

Abstract

Introduction: Validity and reproducibility of the clinician's eye (CE) to diagnose patella alta (PA) on a lateral knee radiography (radiograph) is unknown. Methods: Cross-sectional study of 46 lateral knee x-rays. Three blind observers used CE, Insall-Salvati (IS), modified Insall-Salvati (mIS), and Caton-Deschamps (C-D) to determine patellar height. Sensitivity and specificity of each observer was compared with the musculoskeletal radiologist's C-D measurements. Intraobserver and interobserver agreement were assessed with intraclass correlation coefficient and Fleiss k, respectively. Time needed to estimate patellar height for every method was recorded in seconds. Statistical differences between observers were calculated with a generalized estimating equation. Analysis of variance and post hoc Bonferroni test compared duration of each method (P< 0.05). Data were analyzed using Stata 15 (StataCorp). Results: CE, IS, mIS, and C-D's sensitivity and specificity values are as follows: 77%, 92%; 94%, 52%; 67%, 58%; and 53%, 89%, respectively. Intraclass correlation coefficient and Fleiss k of CE, IS, mIS, and C-D values are as follows: 0.66 and 0.43, 0.88 and 0.68, 0.54 and 0.09, and 0.68 and 0.59, respectively. CE was the second most sensitive and most specific method for diagnosis of PA, with moderate intraobserver and interobserver agreement. IS was the most sensitive method with good intraobserver and interobserver agreement. CE was significantly faster (P< 0.05) than all other conventional radiographic ratios. Conclusion: CE's sensitivity increases with observer's experience and is highly specific. If normal patellar height is diagnosed, no other ratios are necessary, even in the less experienced clinician. Intraobserver and interobserver reproducibilities were moderate and only inferior to the IS ratio. In case patellar height is uncertain with the CE, the IS ratio is the most sensitive and reproducible method to confirm the diagnosis of PA.