Reliability of the “Sydney,” “Sunnybrook,” and “House Brackmann” Facial Grading Systems to Assess Voluntary Movement and Synkinesis after Facial Nerve Paralysis

Abstract

OBJECTIVE: To investigate the extent of within-system reliability and between-system correlation for the “Sydney” and “Sunnybrook” systems of grading facial nerve paralysis, and to examine the interobserver reliability and agreement of the “House Brackmann” grading system. STUDY DESIGN: A fixed-effects reliability study in which 6 otolaryngologists viewed videotapes of patients with facial nerve paralysis. SETTING: University and medical Centers. PATIENTS: Patients with unilateral lower motor neurone facial nerve dysfunction greater than 1 year after onset, none of whom had undergone surgical reanimation procedures. INTERVENTION: Twenty-one patients with facial nerve paralysis were videotaped while they performed a protocol of facial movements. Six otolaryngologists viewed the videotapes and scored them with the Sydney and Sunnybrook systems, and then gave a House Brackmann grade. MAIN OUTCOME MEASURE: The 3 systems of grading facial nerve paralysis were evaluated and compared with the use of intraclass correlation coefficients, Pearson's weighted kappa, and percentage exact agreement values. RESULTS: The Sydney and the Sunnybrook systems had good intrasystem reliability and high intersystem association for the assessment of voluntary movement. Grading of synkinesis was found to have low reliability both within and between systems. The House Brackmann system had substantial reliability as shown by weighted kappa but had a percentage exact agreement of 44%. CONCLUSIONS: For clinical grading of voluntary movement, there is good correlation between ratings given on the Sydney and Sunnybrook systems, and within each system there is good reliability. The assessment of synkinesis was far less reliable within, and less related between, systems. Although the reliability of the House Brackmann system was found to be high, examination of individual grades revealed some wide variation between trained observers.