Anticoagulation in sepsis remains controversial because sepsis involves both dysregulated inflammation and coagulation. This retrospective cohort study evaluated the effects of unfractionated heparin (UFH) and low‑molecular‑weight heparin (LMWH) compared with no anticoagulation in patients with sepsis, assessing survival, bleeding, and changes in multiple coagulation and inflammatory biomarkers.
The study analyzed 1,362 sepsis patients stratified into three groups: no anticoagulation, UFH, or LMWH. Propensity score matching was applied to balance baseline characteristics between treated and untreated groups. After matching, analyses included 109 patients for the UFH comparison and 394 patients for the LMWH comparison against the no anticoagulation group.
UFH was administered intravenously and LMWH subcutaneously at routine standard doses; the abstract does not report specific dosing schedules or protocol details. Twenty‑two biomarkers were measured at baseline and on days 1, 3, and 7 after anticoagulation initiation. Survival and bleeding were assessed using Kaplan‑Meier curves. Generalized linear mixed models identified biomarkers associated with treatment, and mediation analysis tested whether biomarker changes explained part of the survival benefit.
Twenty‑two biomarkers encompassing coagulation and inflammatory pathways were collected at predefined time points: baseline (pre‑treatment) and post‑anticoagulation on days 1, 3, and 7. The study specifically reports results for several key markers including FDP, D‑dimer, IL‑6, IL‑8, and CRP. Statistical adjustments for multiple comparisons (false discovery rate) were applied when identifying significant changes.
After propensity score matching, both UFH and LMWH were associated with significantly improved survival compared with the no anticoagulation group (p < 0.001). Importantly, neither UFH nor LMWH increased the risk of bleeding within 90 days in this cohort. Survival differences were evaluated with Kaplan‑Meier analysis; detailed survival curves and absolute effect sizes are reported in the full manuscript but are not included in the abstract.
Patients treated with UFH demonstrated statistically significant reductions in multiple biomarkers: FDP, D‑dimer, IL‑6, IL‑8, and CRP (FDR‑adjusted p < 0.001). These findings indicate that UFH was associated with both anticoagulant activity (lower FDP and D‑dimer) and broader anti‑inflammatory effects (reductions in IL‑6, IL‑8, and CRP).
The observed anti‑inflammatory changes were temporally assessed at days 1, 3, and 7, suggesting a measurable early effect after UFH administration. The authors used generalized linear mixed models to identify biomarkers correlated with UFH exposure; full model covariates and effect estimates are presented in the complete article but are not detailed in the abstract.
In the LMWH group, significant reductions were found in FDP and D‑dimer only. This pattern suggests that LMWH produced a clear anticoagulant response without the broader anti‑inflammatory biomarker suppression observed with UFH in this cohort. The absence of statistically significant reductions in IL‑6, IL‑8, or CRP with LMWH indicates the measured survival benefit for LMWH may derive mainly from anticoagulant mechanisms rather than suppression of the measured inflammatory mediators.
Mediation analysis was performed to evaluate whether biomarker changes explained part of the survival benefit associated with anticoagulation. For UFH, reductions in IL‑6 and IL‑8 significantly mediated a portion of the survival effect: IL‑6 accounted for 30.9% of the total effect (p = 0.012) and IL‑8 accounted for 29.6% (p = 0.045). These results indicate that suppression of these inflammatory cytokines partly explains UFH's association with improved survival.
For LMWH, mediation analysis did not find any of the measured biomarkers to significantly mediate the survival benefit. The implication in this dataset is that LMWH's survival association is not explained by reductions in the specific inflammatory or coagulation markers assessed beyond its anticoagulant effects on FDP and D‑dimer.
In this propensity score–matched retrospective cohort of patients with sepsis, both UFH and LMWH were associated with improved survival compared with no anticoagulation and did not increase 90‑day bleeding risk. UFH demonstrated broader anti‑inflammatory biomarker suppression (including IL‑6 and IL‑8) and part of its survival benefit was mediated by reductions in these cytokines. LMWH reductions were limited to coagulation biomarkers (FDP, D‑dimer), and its survival association was not mediated by the biomarkers measured.
These findings suggest differential biologic effects between UFH and LMWH in sepsis — UFH may provide both anticoagulant and anti‑inflammatory benefits, whereas LMWH may act predominantly through anticoagulation. Clinicians should interpret these results within the limitations of a retrospective design and propensity matching. Specific dosing, timing, and safety details beyond 90‑day bleeding are not reported in the abstract and require review of the full text for practice implications.
The study is registered in the Chinese Clinical Trial Registry: ChiCTR2100045167, retrospectively registered on 06 April 2021.