基于临床指标构建的列线图模型评分对活动性肺结核与非活动性肺结核的诊断价值

The diagnostic value of a nomogram scoring model based on clinical indicators for active and inactive pulmonary tuberculosis

目的基于临床指标构建一种新的列线图模型,区分活动性肺结核(active pulmonary tuberculosis,APTB)与非活动性肺结核(inactive pulmonary tuberculosis,IPTB)。方法选取2019年1月至2022年12月南通市第六人民医院收治的622例肺结核患者,以2∶1随机分为训练集421例和验证集201例。比较训练集中APTB与IPTB患者的临床资料和血生化指标,多因素Logistic回归分析筛选APTB的预测因素,使用R软件建立列线图。结果训练集与验证集诊断APTB患者271例(64.4%)和116例(57.7%),差异无统计学意义(χ^(2)=2.566,P=0.109)。APTB与IPTB患者比较,APTB有典型症状和合并糖尿病病例数增多,白细胞计数、中性粒细胞百分比、血小板计数、血小板与淋巴细胞比值、中性粒细胞与淋巴细胞比值、高敏C反应蛋白(hs-CRP)、hs-CRP与白蛋白比值、hs-CRP与前白蛋白比值、hs-CRP与淋巴细胞比值(hs-CRP to albumin ratio,HSCLR)、红细胞沉降率、单核细胞与高密度脂蛋白比值(monocyte to high-density lipoprotein ratio,MHR)、降钙素原、乳酸脱氢酶(lactate dehydrogenase,LDH)、碱性磷酸酶和腺苷脱氨酶明显升高(P<0.05);而平均红细胞体积、红细胞比容、血红蛋白、平均红细胞血红蛋白浓度、平均血小板体积、血小板体积分布宽度、尿素氮、血钠、LDH与腺苷酸脱氢酶比值和白蛋白等有明显降低(P<0.05)。平均红细胞体积、红细胞沉降率、白蛋白、腺苷脱氨酶、MHR和HSCLR是APTB的独立预测因素(P<0.05)。ROC曲线计算训练集与验证集的AUC分别为0.956和0.876,提示模型区分度较好。校正曲线与决策曲线显示模型有较好的吻合度与临床净获益比。根据列线图建立的评分系统经ROC曲线计算AUC为0.899,最佳临界值为15分,敏感度为85.6%,特异度为93.7%。结论利用临床最易获得的血生化指标构建列线图评分模型对指导临床鉴别诊断APTB与IPTB有较好的应用价值。

Objective To construct a new nomogram model based on clinical indicators to distinguish between active pulmonary tuberculosis(APTB)and inactive pulmonary tuberculosis(IPTB).Method:A total of 622 patients diagnosed with tuberculosis in the Sixth People's Hospital of Nantong from January 2019 to December 2022 were retrospectively summarized,and were randomly divided into 421 patients in the training set and 201 patients in the verification set by 2:1.The clinical data and blood biochemical indexes of APTB and IPTB patients in the training set were compared,the predictive factors of APTB were screened by multivariate Logistic regression analysis,and the R software was used to establish a column graph.Result There was no significant difference between 271(64.4%)and 116(57.7%)patients diagnosed with APTB in the training set and the verification set(χ^(2)=2.566,P=0.109).Compared with APTB and IPTB patients,APTB has typical symptoms and an increased number of diabetic cases,white blood cell count(WBC),percentage of neutrophils,platelet count(PLT),platelet-lymphocyte ratio(PLR),neutrophil to lymphocyte ratio(NLR),highly sensitive C-reactive protein(hs-CRP),hs-CRP to albumin ratio(HSCAR),hs-CRP to prealbumin ratio(HSCPR)The ratio of hs-CRP to lymphocyte(HSCLR),erythrocyte sedimentation rate(ESR),monocyte to high-density lipoprotein(MHR),procalcitonin(PCT),lactate dehydrogenase(LDH),alkaline phosphatase(ALP)and adenosine deaminase(ADA)were significantly increased.Mean erythrocyte volume(MCV),erythrocyte specific volume(HCT),hemoglobin(Hb),mean erythrocyte hemoglobin concentration(MCHC),mean platelet volume(MPV),platelet volume distribution width(PDW),urea nitrogen,blood sodium,LDH to adenylate dehydrogenase ratio(LAR)and albumin were significantly decreased(P<0.05).MCV,ESR,albumin,ADA,MHR and HSCLR were independent predictors of APTB(P<0.05).The area under the curve(AUC)of the receiver operating curve(ROC)calculation training set and verification set were 0.956 and 0.876,respectively,indicating good model differentiation.The calibration curve and decision curve showed a good agreement between the model and the clinical net benefit ratio.A scoring system was developed based on the nomogram.The AUC calculated by ROC was 0.899,the optimal critical value was 15 points,the sensitivity was 85.6%,and the specificity was 93.7%.Conclusion Using the most readily available blood biochemical indexes to construct the nematoglyphic scoring model has good application value to guide the clinical differential diagnosis of APTB and IPTB.