Interobserver and Intraobserver Reliability of Clinical Assessments in Knee Osteoarthritis.

OBJECTIVE Clinical examination of the knee is subject to measurement error. The aim of this analysis was to determine interobserver and intraobserver reliability of commonly used clinical tests in patients with knee osteoarthritis (OA). METHODS We studied subjects with symptomatic knee OA who were participants in an open-label clinical trial of intraarticular steroid therapy. Following standardization of the clinical test procedures, 2 clinicians assessed 25 subjects independently at the same visit, and the same clinician assessed 88 subjects over an interval period of 2-10 weeks; in both cases prior to the steroid intervention. Clinical examination included assessment of bony enlargement, crepitus, quadriceps wasting, knee effusion, joint-line and anserine tenderness, and knee range of movement (ROM). Intraclass correlation coefficients (ICC), estimated kappa (κ), weighted kappa (κω), and Bland-Altman plots were used to determine interobserver and intraobserver levels of agreement. RESULTS Using Landis and Koch criteria, interobserver κ scores were moderate for patellofemoral joint (κ = 0.53) and anserine tenderness (κ = 0.48); good for bony enlargement (κ = 0.66), quadriceps wasting (κ = 0.78), crepitus (κ = 0.78), medial tibiofemoral joint tenderness (κ = 0.76), and effusion assessed by ballottement (κ = 0.73) and bulge sign (κω = 0.78); and excellent for lateral tibiofemoral joint tenderness (κ = 1.00), flexion (ICC = 0.97), and extension (ICC = 0.87) ROM. Intraobserver κ scores were moderate for lateral tibiofemoral joint tenderness (κ = 0.60); good for crepitus (κ = 0.78), effusion assessed by ballottement test (κ = 0.77), patellofemoral joint (κ = 0.66), medial tibiofemoral joint (κ = 0.64), and anserine tenderness (κ = 0.73); and excellent for effusion assessed by bulge sign (κω = 0.83), bony enlargement (κ = 0.98), quadriceps wasting (κ = 0.83), flexion (ICC = 0.99), and extension (ICC = 0.96) ROM. CONCLUSION Among individuals with symptomatic knee OA, the reliability of clinical examination of the knee was at least good for the majority of clinical signs of knee OA.