Observer error and sampling variability tested in evaluation of hepatitis and cirrhosis by liver biopsy

In 50 patients with chronic active liver disease, observer and sampling error in histologically evaluating hepatitis and cirrhosis after blind-needle biopsy of the liver was assessed from coded tissue. This was done by repeated readings of the same specimens by the same pathologist, by sequential biopsies from the same patients with cirrhosis, and by multiple simultaneous biopsies from adjacent areas of the liver. Observer error was small. The consistency of grading the individual histologic characteristics of hepatitis was 90%, and the reproducibility of the degree of either hepatitis or cirrhosis was 94%. Sampling error was also trivial in hepatitis, indicating that a single needle biopsy accurately reflects the type and degree of inflammation and necrosis in adjacent areas of the liver. By contrast, sampling error was of considerable magnitude when the presence of cirrhosis in patients known to have the lesion was sought, since confirmation by simultaneous or sequential biopsies was made in only 33% of the cases.