Diagnostic Accuracy of Physical Examination Maneuvers for Knee Effusion: An Ultrasound Validation Study

Abstract

To assess the accuracy of physical examination tests for knee effusion using ultrasound (US) as the reference standard, and to explore whether BMI affects test performance. Consecutive consenting patients with knee pain were recruited from an academic Rheumatology clinic. Each had a knee US and clinical exam, including inspection, bulge sign (BS), balloon/cross-fluctuance (CF), and patellar tap (PT) tests by an experienced rheumatologist. The sonographer was blinded to the clinical findings. Based on published literature, a pathologic knee effusion was defined as a hypoechoic collection measuring >3.2 mm in the lateral recess with quadriceps contraction.[1] Patients (N=125) had a mean age of 54 + 16 years and 75% were female. Most patients had RA (45%), other inflammatory arthritis (19%), or CTD (14%). Replaced knees (N=4) were excluded. Clinical evidence of effusion was present in 92/246 knees, including 77 with a CF test, either alone (N=42) or in combination with a BS (N=19) or PT (N=16). In 14 knees with a BS alone, 9 were false positive. One knee had a PT alone which was false positive. The prevalence of pathologic knee effusion on US was 35% (87/246). The physical exam (BS, CF &/or PT + for effusion) had moderate sensitivity 71% (95% CI 61, 81), specificity 81% (95% CI 74, 87), PPV 67% (95% CI 57, 77) and NPV 84% (95% CI 77, 89), with an overall accuracy of 78%. False positive results (N=30) were significantly more frequent in patients with BMI > 25 (83%). False negative results occurred mainly with smaller effusions (<6.6 mm). The CF test had the highest agreement (80%) with US (kappa 0.54, p<0.00005) and the best overall performance (ROC area) compared to the BS and PT (p<0.05). The agreement between the physical exam and US was lower if BMI > 25 (kappa 0.44) compared to BMI <25 (kappa 0.64). The physical exam had moderate diagnostic accuracy for knee effusion. The CF test had the best overall performance in detecting knee effusions in patients with rheumatic diseases, a finding that has not previously been reported. The PT did not add value to the clinical exam in this study. Elevated BMI was associated with reduced accuracy and PPV of the physical exam. Accuracy of knee effusion diagnosis is improved with US, particularly in patients with elevated BMI or small effusions, and is useful for arthrocentesis planning. [1.] Terslev L. Ultraschall Med 2012;33:E173-E178.