参附注射液在脓毒性休克治疗中的精准用药探索

目的
基于回顾性队列研究探讨参附注射液在脓毒性休克治疗中的精准用药。

方法
以广东省中医院脓 毒症专病数据库中 2014 年 9 月至 2024 年 9 月诊断为脓毒性休克的患者为研究对象。将患者根据诊断后参附注 射液的使用情况分为对照组、参附注射液组。根据患者参附注射液的使用剂量（＜60 mL·d-1 ，60～100 mL·d-1 ，＞ 100 mL·d-1 ）、开始用药时间（诊断后 24 h，诊断后 24～72 h，诊断后＞72 h）和使用疗程（＜3 d，3～7 d，＞7 d） 设置亚组。采用倾向性评分匹配（PSM）方法校正混杂变量，通过多因素 Logistic 回归模型调整相关协变量（如年 龄、性别、基线生命体征数据、基础疾病及实验室检查结果等）；根据结局指标的变量类型选择回归分析方法， 对分类变量使用单因素和多因素 Logistic 回归分析。

结果
共纳入参附注射液组 416 例，对照组 2 859 例， PSM 匹配后两组各 413 例，基线特征基本平衡 ［标准化平均差异（SMD）＜0.1 或 P＞0.05］。与对照组比较，参 附注射液组的住院天数、住院死亡率均显著降低（P＜0.05）；单因素 Logistic 回归分析显示，参附注射液可显著 降低脓毒性休克患者死亡风险 ［比值比（OR）=0.601，95% 置信区间（CI）（0.393，0.928），P=0.017］，经多因素 Logistic 回归模型校正后这种保护作用仍然显著。进一步亚组分析发现，不同亚组参附注射液的治疗效果存在 异质性，基线 SOFA 评分＜6 分的患者 ［OR= 0.531，95%CI（0.325，0.867），P=0.011］ 及基线 SOFA 评分为 6～12 分的患者 ［OR = 0.171，95%CI（0.040，0.733），P=0.017］ 均显著获益，交互效应的 P 值= 0.048。女 性、年龄 65～80 岁及基线 MAP＞80 mm Hg、基线乳酸水平 2.5～4 mmol·L-1 、基线乳酸水平＞4 mmol·L-1亚组 均可观察到死亡风险显著降低，但亚组分析的交互作用检验结果为阴性（交互效应 P＞0.05）。不同剂量亚组 （P=0.014）与不同开始用药时间亚组（P=0.015）组间的住院死亡率差异均具有统计学意义；与未使用参附注射液 组比较，＞100 mL·d-1 剂量亚组（P=0.048）及诊断后 24 h 开始用药亚组（P=0.010）的住院死亡率明显降低。

结论
基线 SOFA＜6 分及 SOFA 评分为 6～12 分的脓毒性休克患者均可能从参附注射液的使用中获益；剂 量＞100 mL·d-1及诊断后 24 h 内使用参附注射液可能更适用于脓毒性休克患者。

Objective
To investigate the precision medication of Shenfu Injection in the treatment of septic shock based on a retrospective cohort study.

Methods
Patients diagnosed with septic shock from the Sepsis Specialty Database of Guangdong Provincial Hospital of Chinese Medicine between September 2014 and September 2024 were included. Patients were divided into control group and Shenfu Injection group based on post-diagnosis usage. Subgroups were established according to dosage （＜60 mL·d-1 ，60–100 mL·d-1 ，＞100 mL·d-1 ），initiation time （within 24 h，24-72 h，＞ 72 h post-diagnosis），and treatment duration （＜3 d，3-7 d，＞7 d）. Propensity score matching （PSM） was used to adjust for confounding variables，and multivariate logistic regression models were applied to adjust for covariates （e.g.， age， sex， baseline vital signs， comorbidities， and laboratory results）. Regression analysis methods were selected based on outcome variable types， with univariate and multivariate logistic regression used for categorical variables.

Results
A total of 416 patients in the Shenfu Injection group and 2 859 in the control group were included. After PSM， 413 patients were matched in each group， with balanced baseline characteristics [standardized mean difference （SMD）＜0.1，or P＞0.05]. Compared with the control group，the Shenfu Injection group showed significantly reduced hospital stay and in-hospital mortality （P＜0.05）. Univariate Logistic regression indicated that Shenfu Injection significantly reduced mortality risk in septic shock patients [odds ratio （OR） = 0.601，95% confidence interval （CI） （0.393， 0.928）， P = 0.017]， and this protective effect remained significant after multivariate logistic regression adjustment. Further subgroup analysis revealed heterogeneity in treatment effects： patients with baseline SOFA scores ＜6 [OR = 0.531，95% CI （0.325，0.867），P = 0.011] and those with baseline SOFA scores of 6-12 [OR = 0.171，95% CI （0.040，0.733），P = 0.017，P for interaction = 0.048] showed significant benefits. Significant mortality risk reduction was also observed in females，patients aged 65–80 years，those with baseline MAP ＞80 mm Hg， baseline lactate levels of 2.5-4 mmol·L-1 ， and baseline lactate levels ＞4 mmol·L-1 ， though interaction tests in subgroup analyses were negative （P for interaction ＞0.05）. Significant differences in in-hospital mortality were found across dosage subgroups （P = 0.014） and initiation time subgroups （P = 0.015）. Compared with the non-Shenfu Injection group，the ＞100 mL·d-1 dosage subgroup （P = 0.048） and the within-24 h initiation subgroup （P = 0.010） showed significantly reduced in-hospital mortality.

Conclusion
Septic shock patients with baseline SOFA scores ＜6 or 6－12 may benefit from Shenfu Injection. A dosage ＞100 mL·d-1 and initiation within 24 h post-diagnosis may be more suitable for septic shock patients.