Gait Analysis at Your Fingertips: Accuracy and Reliability of Mobile App Enhanced Observational Gait Analysis in Children with Cerebral Palsy

Abstract

Recipient: Donald Kephart, MD 
 Gait Analysis at Your Fingertips: Accuracy and Reliability of Mobile App Enhanced Observational Gait Analysis in Children with Cerebral Palsy
 Donald T. Kephart, MD1; S. Reed Laing, BA2; Anita Bagley, PhD, MPH2; Jon R. Davids, MD3; Vedant A. Kulkarni, MD4, (1) UC Davis Medical Center, Sacramento, CA, (2) Shriners Hospitals for Children - Northern California, Sacramento, CA, (3) Shriners Hospitals for Children, Sacramento, CA, (4) Shriners Hospitals for Children - Northern California; UC Davis School of Medicine, Sacramento, CA
 Purpose: Three-dimensional gait analysis (3DGA) allows for quantification of gait deviation that can inform decision-making for orthopaedic surgery in children with cerebral palsy (CP). Where 3DGA is unavailable, observational gait analysis (OGA) guided by the Edinburgh Visual Gait Score (EVGS) has been shown to have acceptable reliability. The addition of mobile application slow-motion video analysis may improve the accuracy and reliability of the OGA method. This study prospectively evaluates the accuracy and reliability of mobile app enhanced OGA when compared to the gold standard of 3DGA in children with CP.
 Methods: All subjects gave their informed consent for this IRB approved prospective study. Slow-motion video was captured on an iPhone 8S while simultaneous 3DGA was acquired using a 12-camera infrared system at a children's hospital Motion Analysis Center. Using the Dartfish Express app on an iPad Pro, two observers made 17 quantitative measurements per limb guided by the EVGS [Figure 1]. Inter-class correlation coefficient (ICC) was used to compare reliability between observers and between methods. Pearson Correlation was used to assess the impact of transverse plane deviations on the accuracy of sagittal plane measurements.
 Results: Ten subjects with CP were recruited for the study (7M, 3F; GMFCS II = 8, GMFCS III = 2; Mean 12.4y, range 7.7y to 16.4y). All subjects had significant gait deviation as measured by the Gait Profile Score (GPS), with a mean GPS of 14.10 (SD 4.18), nearly three times greater than the reference normal of 5.2 (SD 1.9). There was excellent overall reliability between raters (ICC 0.95) using mobile app enhanced OGA, and good overall reliability between OGA and 3DGA (ICC 0.89). For individual measurements, the reliability was excellent (ICC > 0.9) for 6 measures, good (ICC 0.75 – 0.9) for 5 measures, moderate (ICC 0.5 – 0.75) for 5 measures, and poor (ICC < 0.5) for 1 measure. The mean error between OGA measurement and 3DGA was 7.02°(SD 6.86°), with foot progression angle (FPA) and knee progression angle (KPA) having among the highest mean errors (12.85° and 10.33°, respectively) [Table 1]. Out of plane measurements substantially affected accuracy of the OGA method. If the FPA or KPA was greater than 20°, the accuracy of mid-stance ankle and knee position had substantially lower reliability and accuracy.
 Conclusion: Mobile-enhanced OGA has good reliability between raters, allowing for enhanced communication about gait deviations in children with cerebral palsy using widely available technology. When compared to 3DGA, mobile-enhanced OGA has clinically acceptable measurement errors in the sagittal plane but should be used with caution to quantify transverse plane deviation. When rotation of the body segment exceeds 20 degrees, sagittal plane measurements become out of plane, resulting in decreased accuracy and reliability.