PURPOSE: To investigate the association between consecutively measured thromboelastographic (TEG) tracings and outcome in patients with severe sepsis.

METHODS: Multicentre prospective observational study in a subgroup of the Scandinavian Starch for Severe Sepsis/Septic Shock (6S) Trial (NCT00962156) comparing hydroxyethyl starch (HES) 130/0.42 vs. Ringer's acetate for fluid resuscitation in severe sepsis. TEG (standard and functional fibrinogen) was measured consecutively for 5 days, and clinical data including bleeding and death was retrieved from the trial database. Statistical analyses included Cox regression with time-dependent covariates and joint modelling techniques.

RESULTS: Of 267 eligible patients, we analysed 260 patients with TEG data. At 90 days, 68 (26 %) had bled and 139 (53 %) had died. For all TEG variables, hypocoagulability according to the reference range was significantly associated with increased risk of death. In a linear model, hazard ratios for death were 6.03 (95 % confidence interval, 1.64-22.17) for increased clot formation speed, 1.10 (1.04-1.16) for decreased angle, 1.09 (1.05-1.14) for decreased clot strength and 1.12 (1.06-1.18) for decreased fibrinogen contribution to clot strength (functional fibrinogen MA), showing that deterioration towards hypocoagulability in any TEG variable significantly increased the risk of death. Patients treated with HES had lower functional fibrinogen MA than those treated Ringer's acetate, which significantly increased the risk of subsequent bleeding [HR 2.43 (1.16-5.07)] and possibly explained the excess bleeding with HES in the 6S trial.

CONCLUSIONS: In our cohort of patients with severe sepsis, progressive hypocoagulability defined by TEG variables was associated with increased risk of death and increased risk of bleeding.