Examiner dependence on physical diagnostic tests for the detection of splenomegaly: a prospective study with multiple observers.

Objective: To determine the reliability and validity of various physical diagnostic techniques (including three methods of palpation and three methods of percussion) in detecting ultrasonographically identified splenomegaly.

Design: Prospective, double-blind study.

Setting: University hospital.

Patients: Twenty-seven hospitalized male patients with suspected human immunodeficiency virus (HIV) infection.

Interventions: Three methods of palpation (bimanual, ballottement, and palpation from above) and three methods of percussion (as described by Nixon, Castell, and Barkun et al.) were performed on each patient by eight examiners. Splenic ultrasonography was performed within 96 hours of admission.

Measurements And Main Results: The prevalence of splenomegaly by ultrasonography (defined as a spleen > or = 13 cm on the longitudinal scan) in this population was 33.3%. The sensitivity and specificity of each method of palpation and percussion varied by examiner. The ranges of sensitivity across examiners for the three methods of palpation and the three methods of percussion were 0%-64.3% and 7.7%-75%, respectively. The ranges of specificity across examiners for the three methods of palpation and the three methods of percussion were 50%-100% and 60%-100%, respectively. Likelihood ratios pooled across observers revealed that for palpation, palpation from above, and percussion, Castell's method had the highest likelihood ratios [LR = 2.66 and 1.97, respectively; 95% CI = 1.52-4.64 and 1.22-3.19, respectively]. A combination of tests (either palpation or percussion) increased the diagnostic accuracy.

Conclusion: Physical diagnostic techniques for the detection of splenomegaly are relatively insensitive but specific. In this study there was high interobserver variability, which did not appear to be associated to the level of experience. Combining tests increases diagnostic accuracy.