Abstract
OBJECTIVE: We aimed to investigate the association between the combination of shenfu injection (SFI) and sepsis bundle therapy and clinical outcomes in critically ill patients with sepsis in China. METHODS: This retrospective cohort study was conducted using Electronic Health Records (EHR) from Ruijin Hospital, affiliated with Shanghai Jiaotong University School of Medicine. Adult critically ill patients diagnosed with sepsis between January 2013 and December 2023 were included in the analysis. The decision to administer SFI was made by the attending physician. The primary outcome was 28-day mortality. Secondary outcomes included ICU length of stay (LOS), hospital LOS, ventilation-free days, vasopressor-free days, procalcitonin clearance rate, lactate clearance rate, and D-dimer serum levels. Propensity score matching was used to minimize bias when comparing patients who received SFI treatment to those who did not. Sensitivity analyses were performed to exclude patients who died within 1, 2, 3, and 7 days of ICU admission. Additionally, a 7-day landmark analysis was conducted to evaluate the effect of SFI treatment on 28-day mortality in patients who survived at least 7 days. RESULTS: Following adjustment for confounding variables via a propensity score model, 154 paired patients were derived from patients who received SFI and those who did not. The infusion of SFI were associated with a lower rate of 28-day mortality rate (9.74% [15/154] vs. 22.73% [35/154]; P = 0.002), prolonged vasopressor-free days (19.35 ± 6.86 vs. 16.64 ± 8.91; P = 0.030) and significantly lower D-dimer levels (3.13 ± 4.25 vs. 4.19 ± 5.11; P = 0.050). No significant difference was found in procalcitonin clearance rate (P = 0.316) or the lactate clearance rate at 72 h. CONCLUSION: Among critically ill patients with sepsis, adjuvant SFI infusion was associated with improvements in 28-day mortality and organ function recovery. The results need to be verified in further randomized controlled trials to validate these findings and address potential biases, such as immortal time bias. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12879-026-12782-0.