What Is the Accuracy of Shoulder Range of Motion Measurements on Physical Exam?

Abstract

The purpose of this study is to investigate a new method for measuring shoulder range of motion (ROM) in an orthopedic practice utilizing a smartphone application to improve accuracy from physical exam typically used in research. Our aim is to evaluate the application, Physio2Go (P2G), which uses a virtual goniometer, assessing validity by comparing its measurements to those taken by a universal goniometer (UG). Two observers of varying clinical experience, a research assistant and research fellow, compared measurements. Statistically, we used the intra-class correlation coefficient (ICC), standard error of measurement (SEM), and the Pearson correlation coefficient (PCC). Following validation we tested P2G in symptomatic postoperative shoulder patients measuring forward flexion (FF) and external rotation (ER). We compared P2G measurements to visual estimation (VE) done by a fellowship trained orthopedic surgeon. Statistically we used ICC, Bland-Altman plots with 95% limits of agreement (LOA), and scatter plots. We examined the impact of the application using Welch’s t-test comparing pre-to-postoperative ROM improvements using the values obtained by P2G and VE. We found high intra-rater reliability of P2G for both observers, substantial correlation between UG and P2G measurements, highly correlated inter-observer reliability for UG and P2G, and statistically significant PCC values (p < 0.05). As expected, ROM measurements of symptomatic patients comparing P2G and VE measurements demonstrated lower correlation. Bland-Altman plots demonstrated wide confidence intervals; scatterplots and histograms confirmed low agreement among measurement methods. Clinical application demonstrated varying statistical significance depending on whether measurements were done by P2G or VE. Our study found that P2G provided superior reliability compared to the customary physical exam routinely used for orthopedic research.