结核性脑膜炎的临床早期诊断特征探索

摘要

ObjectivesTo investigate methods for early diagnosis of tuberculous meningitis (TBM) by comparing the clinical and laboratory features of TBM and acute bacterial meningitis (ABM).MethodsThe clinical and laboratory data of 178 TBM patients and 87 of ABM patients, who were hospitalized from January 2005 to May 2014, were retrospectively analyzed. Logistic analysis was applied to screen the factors related to the diagnosis, and ROC curve was applied to explore the cutoff value. Then an early diagnosis model for TBM was developed.ResultsFactors related to the early diagnosis of TBM included headache duration≥6 days, peripheral blood WBC count<14×109 /L, CSF WBC count<600×106 /L and CSF neutrophil percentage<0.75. The AUC of the early diagnosis model developed based on the 4 factors was 0.999 (95%CI: 0.998-1.000), with a high diagnosis value. The misjudgment rate of this early diagnosis model was 1.15%, lower than that of Thwaites’ model (8.85%).ConclusionsThe model developed for the early diagnosis of TBM has a high diagnosis value with a high diagnosis coincidence rate. It’s worthy of further clinical verification and application.%目的：通过比较结核性脑膜炎和急性细菌性脑膜炎患者的临床特征和实验室指标，探索结核性脑膜炎的早期诊断方法。方法对2005年1月—2014年5月住院确诊的173例结核性脑膜炎和87例急性细菌性脑膜炎患者的临床特征和实验室指标进行回顾性分析，采用logistic回归分析筛选与诊断相关的影响因素，并采用ROC曲线求诊断界值，建立结核性脑膜炎早期诊断模型。结果结核性脑膜炎早期诊断的相关因素包括：头痛时间≥6 d、外周血WBC计数＜14×109/L、脑脊液（cerebrospinal fluid, CSF）WBC计数＜600×106/L、CSF中性粒细胞比例＜0.75。用这4个指标建立的结核性脑膜炎早期诊断模型对应的ROC 曲线下面积为0.999（95%CI：0.998～1.000），具有较高的诊断价值。通过回代法进行验证，模型的误判率为1.15%，低于Thwaites模型的误判率（8.85%）。结论本研究建立的结核性脑膜炎诊断模型具有较高的诊断价值，诊断符合率较高，值得进一步进行临床验证与应用。