The Formulation And Application Of Diagnostic Models Based On Clinical Chemistry Items In Liver Disease Associated With Viral Hepatitis

Objective: To improve the diagnostic ability of routine laboratory items in liver diseases associated with viral hepatitis through constructing predict models consisting of these items. To assess the diagnostic value of nine new laboratory items by comparing them with routine items. Methods (1)Assessment of routine items and formulation of models. We collected data of 606 patients' cases and 131 healthy cases between May 1997 and August 2003 as the training set and examined serum of 400 patients and 74 healthy people between June 2004 and March 2005 as the validation set. Eleven items (TP, ALB, TBIL, DBIL, ALT, AST, ALP, GGT, TBA, LDH, CHE) were examined with auto biochemical analyzer. We used logistic regression to construct two models, one for discriminating between liver disease patients and healthy people, one for discriminating between chronic hepatitis and liver cirrhosis. The diagnostic value of items and models was assessed by the area under the receiver-operating characteristic (ROC) curve. (2) Assessment of new items. The nine items (AFU, PAB, LAP, ASTm, GLDH, ADA, 5'-NT, MAO and AFP) of 400 patients and 74 healthy people were examined with auto biochemical analyzer. Statistical analysis was by receiver-operating characteristic (ROC) curve. (3) Correlation of laboratory items and liver biopsy. Correlation was evaluated by the spearman correlation coefficient. Results (1) The larger areas under ROC of routine items respectively are GGT(0.977), ALT(0.9661), DBIL(0.958) in acute hepatitis , ALT(0.929), AST(0.916), ALB(0.875) in chronic hepatitis, ALB(0.97), TBA(0.969), AST(0.945) in liver cirrhosis, DBIL(0.999), TBIL(0.998), TBA(0.997) in severe hepatitis, ALB(0.98), GGT(0.955), AST(0.949) in liver cancer. The model to discriminating between liver diseases patients and healthy people consists of seven items (ALB, S/L, ALP,...