LAUSR

BACKGROUND  Anticoagulant therapy has seldom been achieved in randomized trials targeting nonspecific overall sepsis patients. Although the key components to identify the appropriate target in sepsis may be disseminated intravascular coagulation (DIC) and high disease severity, the interaction and relation of these two components for the effectiveness of therapy remain unknown. OBJECTIVE  This article identifies the optimal target of anticoagulant therapy in sepsis. METHODS  We used a prospective nationwide cohort targeting consecutive adult severe sepsis patients in 59 intensive care units in Japan to assess associations between anticoagulant therapy and in-hospital mortality according to DIC (International Society on Thrombosis and Haemostasis [ISTH] overt and Japanese Association for Acute Medicine DIC scores) and disease severity (Acute Physiology and Chronic Health Evaluation II [APACHE II] and Sequential Organ Failure Assessment scores). Multivariable Cox proportional hazard regression analysis with nonlinear restricted cubic spline including a two-way interaction term (treatment × each score) and three-way interaction term (treatment × ISTH overt DIC score × APACHE II score) was performed. RESULTS  The final study cohort comprised 1,178 sepsis patients (371 received anticoagulants and 768 did not). The regression model including the two-way interaction term showed significant interaction between intervention and disease severity as indicated by the ISTH overt DIC score and APACHE II score (p = 0.046 and p = 0.101, respectively). Three-way interaction analysis revealed that risk hazard was suppressed in the anticoagulant group compared with the control group in the most severe subset of both scores. CONCLUSION  Anticoagulant therapy was associated with better outcome according to the deterioration of both DIC and disease severity, suggesting that anticoagulant therapy should be restricted to patients having DIC and high disease severity simultaneously.