Abstract: Objective The aim of this study was to investigate the difference in predicting efficacy of significant liver fibrosis by noninvasive liver fibrosis index in different hospitals.Methods Clinical data of patients with chronic HBV infection from two hospitals were retrospectively analyzed,and the liver biopsies were done in all the patients. The APRI,FIB-4 and liver stiffness measure（LSM） were obtained. The diagnostic performance was analyzed by ROC curve.Results We enrolled 327 patients （group A） and 250 patients （group B） from two hospitals respectively;there were significant differences as respect to liver pathological data and LSM in the two groups（both P<0.001）,while there were no significant differences respect to APRI or FIB-4 in the two groups （P=0.547 and P=0.578）;as for distinguishing liver fibrosis S0-1 and ≥S2,the APRI in group A was set at 0.14 and 0.18,in group B was 0.15 and 0.24,the FIB-4 in group A was 0.98 and 1.26, in group B was 0.93 and 1.50,while the LSM in group A was set at 5.2 kPa and 6.8 kPa,and in group B was 7.2 kPa and 9.0 kPa（P<0.001）,respectively;the cut-off-values of APRI in group A and group B were 0.105 and 0.145,of FIB-4 were 0.675 and 0.775,and of LSM were 4.650 and 6.345,respectively,their sensitivities and specificities in predicting significant liver fibrosis of patients from the two hospitals were at about 85% and 24% to 46%. Conclusion The diagnostic performance of APRI,FIB-4 and LSM in predicting significant liver fibrosis might varies,because of some uncontrollable factors. Hence,the cut-off-value of each noninvasive index might be set in every hospital.

Key words: Chronic hepatitis B, Liver fibrosis, Aspartate aminotransferase to platelet ratio index, Fibrosis-4 index, Liver stiffness measurement, Diagnosis