Background: The diagnosis of liver cirrhosis is important in the evaluation and management of patients. Liver biopsy is the gold standard but it is invasive. Ultrasonography is a non-invasive and useful modality in assessing the liver for certain conditions but its sensitivity and specificity in diagnosing cirrhosis is unknown locally.
Aim: To assess the accuracy of ultrasonography in diagnosing compensated liver cirrhosis in daily clinical practice outside the context of clinical trials.
Methods: All the liver biopsies were identified from the Pathology Logbook retrospectively from January 1998 to March 2001. Only patients who had both liver biopsy and ultrasonography with no clinical evidence of cirrhosis were included. Patients with incomplete data, hepatoma or liver secondaries were excluded. Ultrasonographic diagnosis of cirrhosis was based on nodularity or irregularity of the liver surface, small liver size, coarse echotexture and increase attenuation by using the 3.5 to 5 MHz transducers.
Results: A total of 151 liver biopsies were performed during this period. Eighty-eight patients who had both ultrasound and liver biopsy were analysed. Seventeen patients had ultrasonographic diagnosis of cirrhosis but only six cases were proven by a liver biopsy. On the other hand, 10/16 cases of biopsy-proven cirrhosis were "missed" by ultrasound. Thus, the sensitivity of ultrasonography in diagnosing cirrhosis was 37.5% and the specificity was 84.7%. The positive and negative predictive values were 35.3% and 85.9% respectively.
Conclusion: Low frequency ultrasonography is not a sensitive test for the diagnosis of liver cirrhosis in daily clinical practice.