目的 对比降钙素原（PCT）指导组与常规治疗组在脓毒症患者抗感染治疗中的整体疗效，并通过分析PCT指导抗感染治疗策略在不同临床特征亚组中的交互作用与获益差异，为临床个体化应用提供真实世界依据。 方法 选取2021年1月至2024年6月中山大学孙逸仙纪念医院收治的脓毒症患者515例，采用回顾性队列研究，依据临床实际采用的监测与治疗方式分为PCT指导组（245例）与常规治疗组（270例）。两组基线年龄、性别、基础疾病均衡可比；PCT指导组每24 ~ 48 h动态监测PCT ≥ 3次，以PCT峰值下降 ≥ 80%或连续2次PCT < 0.5 ng/mL作为核心停药阈值，结合临床调整抗感染治疗；常规治疗组则根据传统指标及经验调整。再对比两组整体疗效与安全性核心指标，同时按年龄、序贯器官衰竭评估（SOFA）评分、是否合并糖尿病进行亚组分析，采用基于回归模型的交互项检验法分析各临床特征与PCT指导策略的交互效应。 结果 PCT指导组抗菌药物使用时长［408.01 （228.06，887.00） h vs. 579.01 （288.99，1 290.77） h， P = 0.001］显著短于常规治疗组，停药时PCT水平［0.31 （0.09，1.41） ng/mL vs. 0.97 （0.22，6.04） ng/mL， P < 0.001］、抗菌药物费用［6 790.44 （2 393.00，21 712.71） 元 vs. 12 231.96 （3 396.04，27 211.60） 元，P = 0.004］及28 d死亡率（16.7% vs. 28.1%，P = 0.003）显著低于常规治疗组，两组日均抗菌药物费用、肾功能异常发生率差异无统计学意义（均P > 0.05）。亚组及交互作用分析显示，年龄与PCT指导策略在抗菌药物使用时长存在显著交互作用（P for interaction = 0.039），糖尿病状态与PCT指导策略在住院时间存在显著交互作用（P for interaction = 0.003），SOFA评分与PCT指导策略无显著交互作用（均 P for interaction > 0.05）。 结论 动态监测PCT可有助于缩短抗菌药物使用疗程、降低医疗成本并改善短期预后；该策略在年龄< 65岁、无糖尿病患者中获益更显著，老年及合并糖尿病者需结合临床综合评估使用。

Objective To compare the overall efficacy of procalcitonin （PCT）-guided antimicrobial therapy with that of conventional antimicrobial therapy in patients with sepsis， and to analyze the interaction and differential benefits of the PCT-guided strategy among subgroups with different clinical characteristics， thereby providing real-world evidence for individualized clinical application. Methods A total of 515 patients with sepsis who were admitted to Sun Yat-sen Memorial Hospital of Sun Yat-sen University from January 2021 to June 2024 were included in this retrospective cohort study. According to the actual monitoring and treatment strategies implemented in clinical practice， the patients were divided into a PCT-guided group （n = 245） and a conventional treatment group （n = 270）. The baseline characteristics， such as age， sex， and underlying diseases， were well-balanced between the two groups. In the PCT-guided group， the PCT levels were dynamically monitored every 24 - 48 hours for at least three times. The core antibiotic discontinuation threshold was defined as a ≥ 80% decline in the peak PCT level or two consecutive PCT values < 0.5 ng/mL， and the antimicrobial therapy was adjusted accordingly in combination with the clinical assessment. In contrast， the conventional treatment group received antibiotic adjustments based on traditional indicators and clinical experience. Subsequently， the overall core efficacy and safety outcomes were compared between the two groups. Subgroup analyses were carried out based on age， Sequential Organ Failure Assessment （SOFA） score， and the presence of diabetes mellitus. A regression model-based interaction term test was employed to examine the interaction effects of each clinical characteristic with the PCT-guided strategy. Results The PCT-guided group exhibited a significantly shorter duration of antibiotic use ［408.01 （228.06， 887.00） h vs. 579.01 （288.99， 1 290.77） h， P = 0.001］ when compared with the conventional group. The PCT level at the time of antibiotic discontinuation ［0.31 （0.09， 1.41） ng/mL vs. 0.97 （0.22， 6.04） ng/mL， P < 0.001］， antibiotic costs ［6 790.44 （2 393.00， 21 712.71） yuan vs. 12 231.96 （3 396.04， 27 211.60） yuan， P = 0.004］， and the 28-day mortality rate （16.7% vs. 28.1%， P = 0.003） were also notably lower in the PCT-guided group. No significant differences were detected between the two groups in terms of the average daily antibiotic cost or the incidence of kidney dysfunction （all P > 0.05）. Subgroup and interaction analyses indicated that there was a significant interaction between age and the PCT-guided strategy regarding antibiotic duration （P for interaction = 0.039）， and between diabetes status and the PCT-guided strategy regarding the length of hospital stay （P for interaction = 0.003）； no significant interaction was identified between the SOFA score and the PCT-guided strategy （all P for interaction > 0.05）. Conclusions Dynamic monitoring of PCT can assist in shortening the duration of antibiotic therapy， reducing medical costs， and enhancing short-term prognosis. This benefit is more pronounced in patients aged less than 65 years and without diabetes. Elderly patients and those with diabetes should undergo comprehensive clinical evaluation.