The scratch test is unreliable for detecting the liver edge.

To determine the validity and reliability of the scratch test method for estimating the liver span below the right costal margin, we performed a prospective double-blind study using multiple examiners at different levels of training. Twenty-two patients were examined by 11 observers using only the scratch test. Measures of liver edge length below the right costal margin using the scratch test were compared with those by ultrasound. The validity of the scratch test was determined by simple linear regression and the concordance correlation coefficient. There was very poor correlation between the scratch test estimates and ultrasound measurements for all examiners. Measurement reliability, estimated using generalizability theory, for seven observers who examined the same nine patients was 0.68. The reliability of a single examiner was 0.24. The scratch test method for measuring the liver edge span below the right costal margin was neither a valid nor reliable method of physical examination.