摘要：

目的: 探讨重症结核性脑膜炎患者远期预后相关影响因素。方法: 纳入2012年6月至2019年12月在天津市海河医院行抗结核治疗的163例重症结核性脑膜炎患者，治疗后1年采用改良Rankin量表(mRS)评价预后，单因素和多因素逐步法Logistic回归分析筛查重症结核性脑膜炎患者远期预后不良相关影响因素，并绘制受试者工作特征曲线(ROC曲线)评价其对预后的预测价值。结果: 根据mRS评分分为预后良好组(mRS评分≤ 2分，95例)和预后不良组(mRS评分 > 2分，68例)。预后不良组患者年龄(t = - 5.884，P = 0.000)，高血压(χ² = 15.769，P = 0.000)、冠心病(校正χ² = 6.785，P = 0.009)、四肢无力(χ² = 9.544，P = 0.002)、脑积水(χ² = 23.004，P = 0.000)、脑梗死(χ² = 17.417，P = 0.000)比例，脑脊液乳酸(Z = - 2.405，P = 0.016)均高于预后良好组，入院时Glasgow昏迷量表(GCS)评分低于预后良好组(Z = - 6.750，P = 0.000)。Logistic回归显示，高龄(OR = 1.064，95%CI：1.031 ~ 1.098；P = 0.000)和脑脊液高乳酸水平(OR = 1.404，95%CI：1.065 ~ 1.852；P = 0.016)是重症结核性脑膜炎患者远期预后不良的危险因素，入院时高GCS评分是远期预后良好的保护因素(OR = 0.539，95%CI：0.405 ~ 0.716；P = 0.000)。进一步绘制ROC曲线，年龄、入院时GCS评分和脑脊液乳酸水平预测重症结核性脑膜炎患者远期预后不良的截断值分别为48.50岁、12.50分和6.99 mmol/L，曲线下面积分别为0.743(95%CI：0.665 ~ 0.822，P = 0.000)、0.802(95%CI：0.732 ~ 0.871，P = 0.000)和0.611(95%CI：0.517 ~ 0.704，P = 0.016)，灵敏度和特异度分别为63.20% 和76.80%、58.80% 和86.30%、42.60% 和91.60%；上述3项指标联合应用的曲线下面积为0.892(95%CI：0.841 ~ 0.944，P = 0.000)，灵敏度为80.90%、特异度88.40%，其预测效能优于单一指标。结论: 高龄和脑脊液高乳酸水平是重症结核性脑膜炎患者远期预后不良的危险因素，入院时高GCS评分为保护因素，临床医师应重点关注这些因素以改善患者预后。

关键词: 结核, 脑膜, 预后, 危险因素, Logistic模型

Abstract:

Objective: To investigate the influencing factors related to long - term prognosis in patients with severe tuberculous meningitis (TBM). Methods: One hundred sixty-three patients with severe TBM admitted to Tianjin Haihe Hospital from June 2012 to December 2019 were included. After one year of anti - tuberculous therapy, the prognosis of the patients was evaluated by modified Rankin Scale (mRS). Univariate and multivariate stepwise Logistic regression analyses were used to analyze the influencing factors of long - term dismal prognosis of severe TBM. The predictive value was evaluated by receiver operating characteristic (ROC) curve. Results: In the dismal prognosis group (mRS > 2, n = 68), patients' age (t = - 5.884, P = 0.000), hypertension (χ² = 15.769, P = 0.000), coronary heart disease (adjusted χ² = 6.785, P = 0.009), limb weakness (χ² = 9.544, P = 0.002), hydrocephalus (χ² = 23.004, P = 0.000), cerebral infarction (χ² = 17.417, P = 0.000), cerebrospinal fluid (CBF) lactic acid level (Z = - 2.405, P = 0.016) were higher than those in favourable prognosis group (mRS ≤ 2, n = 95), while Glasgow Coma Scale (GCS) score on admission was lower than that in favourable prognosis group (Z = - 6.750, P = 0.000). Logistic regression showed that old age (OR = 1.064, 95%CI: 1.031-1.098; P = 0.000) and high CSF lactic acid level (OR = 1.404, 95%CI: 1.065-1.852; P = 0.016) were risk factors for long- term dismal prognosis of severe TBM, and high GCS score on admission was a protective factor for long-term favourable prognosis of severe TBM (OR = 0.539, 95%CI: 0.405-0.716; P = 0.000). ROC curve showed the cut - off value of age for predicting long - term dismal prognosis of severe TBM on admission was 48.50 years, the area under the curve (AUC) was 0.743 (95%CI: 0.665-0.822, P = 0.000), the sensitivity was 63.20% and the specificity was 76.80%. The cut - off value of GCS score was 12.50, the AUC was 0.802 (95%CI: 0.732-0.871, P = 0.000), the sensitivity was 58.80% and the specificity was 86.30%. The cut - off value of lactic acid level in CSF was 6.99 mmol/L, the AUC was 0.611 (95%CI: 0.517-0.704, P = 0.016), the sensitivity was 42.60% and the specificity was 91.60%. The AUC of the combined application of the 3 indexes was 0.892 (95%CI: 0.841 -0.944, P = 0.000), the sensitivity was 80.90% and the specificity was 88.40%, which showed that the prediction efficiency was better than that of a single index. Conclusions: Advanced age and high lactic acid level in CBF are risk factors for dismal long - term prognosis of severe TBM. High GCS score on admission is a protective factor. Clinicians should focus on these factors to improve the prognosis of patients.

Key words: Tuberculosis, meningeal, Prognosis, Risk factors, Logistic models