OBJECTIVETo systematically assess imaging diagnostic tests for fatty liver and provide a decision-making basis for clinical diagnosis and screening.

METHODSElectronic searches were conducted on the Chinese Biomedical Database, PubMed, and EMBASE, combining with manually searching of Chinese literature. All searches were completed until November 2002. All studies which evaluated imaging diagnostic test of human fatty liver were included. Data of diagnostic accuracy in the included studies were extracted, and methodological quality of the studies was assessed independently by two reviewers according to the established quality standard. Quantitative analysis or qualitative description were performed based on available data.

RESULTSOf 13 studies that met the eligibility criteria, 10 studies evaluated the diagnostic accuracy of B-mode ultrasound, 3 studies evaluated contrast-enhanced (helical) CT. To assess 7 diagnostic test studies for fatty liver that used liver biopsy as reference test: the pooled sensitivity of B-mode ultrasound was 0.89 (95% confidence interval 0.87-0.92), specificity was 0.94 (95% confidence interval 0.92-0.96) and the Q value was 0.90 by adjusted SROC method. To assess 2 diagnostic test studies for fatty liver that used CT as reference test: the pooled sensitivity, specificity, and Q value were 0.92 (95% confidence interval 0.89-0.96), 0.88 (95% confidence interval 0.84-0.92), and 0.90 respectively by adjusted SROC method.

CONCLUSIONSB-mode ultrasound method can be regarded as an effective method for fatty liver diagnosis and screening. The methodologic quality of diagnostic test needs to be improved.