基于受试者工作特征曲线和逐步判别分析法探索实验室指标对慢性乙型肝炎中医证候诊断的价值

目的：探索75个常用临床实验室指标对慢性乙型肝炎中医肝胆湿热、肝郁脾虚证的诊断价值。方法：选取慢性乙型肝炎病人422例，其中肝胆湿热证300例，肝郁脾虚证122例。检测常用实验室指标，包括肝肾功能、凝血功能、乙型肝炎标志物、乙型肝炎病毒DNA定量检测、血常规、激素指标、细胞免疫指标、体液免疫指标、血脂、甲胎蛋白、肝纤维化指标等共75项。采用受试者工作特征曲线（receiver operating characteristic curve，ROC）查看单一指标对慢性乙型肝炎肝胆湿热证、肝郁脾虚证的诊断效率，采用逐步判别分析模型检测差异指标组合对慢性乙型肝炎证候的判别效率。结果：白蛋白、凝血酶原时间、免疫球蛋白A、免疫球蛋白M、尿素氮、血尿酸、嗜碱粒细胞、嗜碱粒细胞百分数、血小板平均体积在慢性乙型肝炎肝胆湿热证与肝郁脾虚证之间存在组间分布差异。上述差异指标的ROC曲线下面积范围为0．42～0．62。对差异指标采用逐步判别分析法建立证候判定模型，其模型自身验证总误判率35．3％，刀切法总误判率为35．3％。结论：无论单一差异指标还是差异指标组合判定模型均未达到较好的判定慢性乙型肝炎中医证候的目的，提示单一采用临床指标判定中医证候具有一定的局限性。

Diagnostic value of clinical indices in syndrome differentiation of chronic hepatitis B: an exploration based on receiver operating characteristic curves and stepwise discriminant analysis

OBJECTIVE： To explore the diagnostic value of 75 of traditional Chinese medicine syndromes such as I spleen deficiency in patients with chronic hepatitis B. commonly used clinical laboratory markers for differentiation ver and gallbladder damp-heat and liver depression and METHODS： A total of 422 patients with chronic hepatitis B （CHB） were enrolled, including 300 patients with damp-heat in liver and gallbladder syndrome, and 122 patients with liver depression and spleen deficiency syndrome. Seventy-five commonly used clinical markers were selected, including liver and kidney function, clotting function, the quantitative detection of hepatic B virus （HBV） markers, HBV-DNA, blood count, hormones levels, cellular immunity indicators, humoral immunity indicators, lipid panel, protein electrophoresis, alpha-fetoprotein and liver fibrosis indicators. Receiver operating characteristic （ROC） curve was used to detect the diagnostic efficiency of single differential indicators, and stepwise discriminant analysis model was used to explore the discrimination efficiency of differential indices between two TCM syndromes in CHB. RESULTS： The differential indices between two CHB Chinese syndromes were albumin, prothrombin time, immunoglobulin A, immunoglobulin M, blood urea nitrogen, blood uric acid, basophils, basophil percentage and mean platelet volume. The area under ROC curve （AUC） of these indices was between 0.42 and 0.62, and the total false positive rate of own validation of stepwise discriminant analysis model, which was estab- lished by differential indices combination, was 35.3%, and the jackknife total error rate was 35.3%. CONCLUTION： Neither single differential index nor multiple differential indices determinant models provided appropriate determination of the TCM syndromes of patients with chronic hepatitis B, suggesting that clinical indicators have limited value in determining traditional Chinese medicine syndromes.