• global haemostasis;
  • sepsis;
  • thrombin generation;
  • thromboelastography;
  • thrombosis


  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Competing interests
  7. Acknowledgements
  8. References

Haemostatic changes in septic patients are complex, with both procoagulant and anticoagulant changes. Thirty-eight patients with severe sepsis and 32 controls were investigated by coagulation screens, individual factor assays, calibrated automated thrombography (CAT), whole blood low-dose-tissue factor activated (LD-TFA) Rotem and LD-TFA waveform analysis. Thirty-six of 38 patients had an abnormal coagulation screen. The mean levels of factors II, V (P < 0·05), VII, X, XI and XII, antithrombin and protein C (P < 0·01) was decreased in sepsis compared with controls. The mean factor VIII and fibrinogen level (P < 0·001) was increased. CAT in platelet rich and poor plasma showed a prolonged lag time (P < 0·02), decreased peak thrombin (P < 0·02) and delayed time to peak thrombin (P < 0·001) in sepsis patients, however, the endogenous thrombin potential was equivalent in sepsis and controls. In LD-TFA Rotem, septic patients had delayed clot times (P = 0·04) but an increased maximum velocity of clot formation (P < 0·01) and area under the clot elasticity curve (P < 0·01). LD-TFA waveform analysis showed a delayed onset time but an increased rate of clot formation (P < 0·005). In conclusion, global tests of haemostasis suggest that in this patient group, activation of haemostasis is delayed but once initiated thrombin generation and clot formation are normal or enhanced.

Sepsis is associated with complex changes in haemostasis. In severe cases, disseminated intravascular coagulation (DIC) may lead to consumption of platelets and coagulation factors resulting in clinical bleeding, whilst in other situations a compensated consumptive coagulopathy may be present, with fibrinogen levels and platelet number preserved or raised (Dempfle, 2004). Altered levels of both procoagulant factors and anticoagulant proteins have been described in patients with sepsis syndrome (Hesselvik et al, 1989; Mavrommatis et al, 2000; Dempfle, 2004; Dhainaut et al, 2005). Some changes, such as decreased levels of coagulation factors, predispose the patient to bleeding whilst others, such as raised factor VIII and fibrinogen and decreased levels of protein C and antithrombin, induce a prothrombotic state.

Critically ill patients with sepsis syndrome often develop multi-organ organ failure. This complication has a complex pathophysiology but is thought to be partly due to microvascular thrombosis (Dixon, 2004). This hypothesis is indirectly supported by studies showing that patients with sepsis have increased markers of activation of haemostasis, such as d-dimer, prothrombin fragment 1 + 2 and thrombin antithrombin complexes (TAT) (Gando et al, 1998; Mavrommatis et al, 2000; Amaral et al, 2004). Improved assessment of the balance between anticoagulant and prothrombotic changes may aid understanding of haemostatic changes associated with sepsis syndrome, give insight into the pathogenesis of multi-organ failure and ultimately aid the clinical management of patients.

Current concepts of haemostasis support the view that haemostasis is activated through low concentrations of tissue factor (Mann et al, 2003; Roberts et al, 2004) and this has been confirmed in in vivo models of sepsis (Taylor et al, 1991). It is also thought that the rate of clot formation, and in particular the rate of thrombin generation, is crucial for formation of a stable fibrin clot (Roberts et al, 2004). Routine coagulation screens are often prolonged in critically ill patients with sepsis syndrome (Dempfle, 2004). However, there is debate about the utility of these tests in assessing haemostasis (Mannucci, 2006; Reverter, 2006) and coagulation screens appear not to accurately reflect a patient's risk of bleeding (Segal & Dzik, 2005). This may be because the endpoint of these assays is early in the haemostatic process or because they are initiated by high concentrations of tissue factor or contact activators that do not reflect the physiological situation. Despite these shortcomings, prolonged coagulation screens are often treated with infusion of fresh frozen plasma (FFP), even though individual coagulation factor levels are often not significantly reduced (Chowdhury et al, 2004).

There has been recent interest in the role of global tests of haemostasis, such as thrombin generation assays, in the investigation of both acquired and congenital haemostatic defects (Al Dieri et al, 2002; Hemker et al, 2003; Luddington & Baglin, 2004). A thromboelastographic method, activated by low concentration tissue factor and measured using the Rotem® machine, has also been described (Sorensen et al, 2003). Updated software enabled the standard clot elasticity trace to be differentiated to give a velocity of clot strengthening, and the area under this curve gives a measure of the total amount of increasing clot strength (Sorensen et al, 2003). A third test is described here, which activates haemostasis through low concentrations of tissue factor in the presence of thrombomodulin and measures rate of fibrin clot formation using the first derivative of the MDA waveform.

Assays that are activated by low concentrations of tissue factor are susceptible to contact activation (Luddington & Baglin, 2004). Although activation of contact factors may occur in sepsis, depletion of these factors will not affect in vivo haemostasis and assays that are contact-activated may be misleading in the setting of sepsis. In the assays presented here, we therefore inhibited contact activation with corn trypsin inhibitor (CTI). We report the results of routine coagulation tests and three global haemostatic assays in critical ill patients with severe sepsis syndrome.

Materials and methods

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Competing interests
  7. Acknowledgements
  8. References


The study was reviewed by the South East Wales Research Ethics Committee and the Cardiff and Vale NHS Trust Research and Development Office. A cohort of 32 anonymised, adult normal controls was recruited and compared with a cohort of 39 adult critically ill patients with severe sepsis syndrome. The patient cohort was recruited from the general intensive care unit (ICU) and were aged 18 years or older. Severe sepsis syndrome was defined according to standard criteria; temperature >38·5°C or <35°C, white blood cell count >11 × 109/l or <4 × 109/l, respiratory rate >20 breaths/min or patient dependent on ventilation, heart rate >90 beats/min and suspected or proven infection, with organ dysfunction, hypoperfusion or hypotension (American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference, 1992). No patients were bleeding or receiving an infusion of activated protein C or therapeutic heparin at the time of sample collection. Patients, or next-of-kin, gave written informed consent. Patients who were unable to give consent were retrospectively informed of the study after recovery and given the opportunity to withdraw. Basic demographic data as well as 28-day mortality was collated from the ICU database.

Blood samples

Blood was obtained from patients via a radial arterial line. The first 20 ml of blood was drawn into a syringe and discarded to avoid heparin contamination. Measurement of anti-Xa heparin levels in the samples showed a raised level in one patient who was excluded from subsequent analyses. A full blood count was performed (Haribo ABX Pentra 120; Haribo ABX, Montpellier, France). Further blood was drawn into a fresh syringe and immediately anticoagulated with 0·109 mol/l trisodium citrate (Sigma, Poole, UK) and 20 μg/ml CTI (Cambridge BioScience, Cambridge, UK). Two further 4·5 ml aliquots of citrated blood were collected. The samples were centrifuged at 83 g for 10 min at room temperature. The supernatant platelet rich plasma (PRP) was collected and the samples re-spun at 1168 g for a further 15 min at room temperature. The plasma supernatant was removed and re-spun at 1168 g to give double-spun platelet poor plasma (PPP). The PRP platelet count was measured (Haribo ABX Pentra 120) and diluted with autologous PPP to give a count of 150 × 109/l. In samples where the platelet count in the PRP was <150 × 109/l, no adjustment was made.

Further PPP samples (not taken into CTI) were aliquoted and frozen at −80°C for later analysis of prothrombin time (PT), activated partial thromboplastin time (aPTT), TT, fibrinogen, factors II, V, VII, VIII, IX, X, XI and XII, protein C and antithrombin using the MDA II analyser (bioMérieux UK Ltd, Basingstoke, UK) according to manufacturer's protocols. These assays are externally quality controlled through the UK National External Quality Assessment Schemes.

Calibrated automated thrombography

Immulon 2HB 96-well round bottomed microtitre plates were used to carry out the fluorogenic reaction. Thrombin generation was measured using a Fluoroskan Ascent plate reader (ThermoLabsystems, Helsinki, Finland) using excitation and emission spectra of 390 and 460 nm, respectively. Fluorogenic substrate (Z–Gly–Gly–Arg–AMC) was obtained from Bachem (St Helens, UK) and was prepared using the method described (Hemker et al, 2003). Dimethyl sulphoxide and bovine serum albumin (BSA) were obtained from Sigma (Poole, UK), HEPES, NaCl and CaCl2 were obtained from VWR International (Lutterworth, UK). The phospholipid vesicles were constructed by an extrusion method and consisted of phosphatidylserine, phosphatidylethanolamine and phosphatidylcholine.

Eighty μl of sample (PRP or PPP) was incubated with 20 μl of tissue factor solution, Innovin (Sysmex UK Ltd, Milton Keynes, UK) pH 7·35 diluted in HEPES/NaCl buffer, to give a final concentration of 5 pM for PPP and 0·5 pM for PRP as previously described (Hemker et al, 2003). Each sample was measured in triplicate with a calibration well that contained 80 μl of PPP or PRP from the relevant patient and 20 μl of thrombin calibrator (600 nmol/l) supplied by Synapse BV, Maastricht, the Netherlands. The plate was warmed in the Fluoroskan for 5 min, followed by the addition of 20 μl of fluorogenic substrate per well. The fluorescent signal was then measured at 15 s intervals until the reaction was complete. The data was analysed using thrombinoscopeTM software (Synapse BV, Maastricht, the Netherlands). Each plate contained a control aliquot of PPP, the inter-assay coefficient of variance was lag time 12%, endogenous thrombin potential (ETP) 8%, peak thrombin 8%, time to peak thrombin 10%.

Low concentration tissue factor-activated thromboelastography

Measurements were performed on a Rotem 05 Coagulation Analyser (Pentapharm®, Munich, Germany). All tests were performed in Rotem cup and pins. Blood samples were tested 30 min after venepuncture (Sorensen et al, 2003). Cup holders were prewarmed to 37°C. 300 ml of whole blood taken into CTI, final concentration 20 μg/ml, was dispensed into the sample cups according to the EXTEG program. Clotting was initiated by the addition of 20 μl 200 mmol/l CaCl2 and tissue factor (Innovin) at final concentrations of 0·35 pmol/l as previously described (Sorensen et al, 2003). calcuro software (Pentapharm GmbH, Munich, Germany) was used to calculate derived parameters. As the whole blood nature of the assay makes quality assurance difficult, the inter-assay reproducibility was assessed using freeze-thaw PRP that was aliquoted and run with each sample. This gave a coefficient of variance of clot time 9%, mean clot firmness 4%, alpha angle 2%, maximum velocity 10%, time to maximum velocity 9% and area under the clot firmness curve 3%.

Low concentration tissue factor-activated MDA waveform global haemostatic assay

Measurements were performed on a MDAII coagulation analyser (bioMérieux, Lyon, France). Briefly 50 μl of PPP taken into 20 μg/ml CTI was incubated at 37°C with 2 pmol/l tissue factor and 0·75 nmol/l Thrombomodulin (American Diagnostica Inc., Stamford, CT, USA) for 220 s. Clot formation was initiated by the addition of CaCl2 (VWR International, Lutterworth, UK) containing 1 mmol/l phosphorylcholine (Sigma-Aldrich Co. Ltd, Gillingham, UK) and monitored at a wavelength of 460 nm. The maximum rate of fibrin polymerisation was calculated from the first derivative of the light transmittance (T) against time and is the value reported (Braun et al, 1997).


The data were analysed using The Statistical Package for the Social Sciences (spss) version 12·1. Parameters measured on control samples were normally distributed; however, many parameters measured on patient samples were not (analysed by the Kolmogorov–Smirnov test). Comparisons between patients and controls were therefore performed with a Mann–Whitney U-test and correlations were performed using a Spearman rank test. P-values <0·05 were considered significant.


  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Competing interests
  7. Acknowledgements
  8. References


The patients were recruited from the general ICU if they fulfilled the criteria for severe sepsis syndrome (American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference, 1992). Patient characteristics at the time of enrolment into the study are shown in Table I. None of the patients was bleeding at the time of blood sampling.

Table I.   Patient characteristics and outcome.
 Median (range)
  1. The clinical status of patients when recruited into the study is shown. APACHE, acute physiology and chronic health evaluation; SAPS, simplified acute physiology score; ICU, intensive care unit.

Age (years)61 (19–82)
APACHE II18 (6–30)
APACHE III61 (26–94)
SAPS II43 (21–69)
Number of organ failures1 (0–4)
Organ failure score17 (6–36)
Length of stay on ICU11 (1–60)
28 day mortality34%

Routine coagulation tests and coagulation factor levels

The patients with severe sepsis syndrome had abnormal coagulation screens compared with normal controls (Table II). Of the 38 patients with sepsis, 36 (95%) had an abnormal coagulation screen and 23 (60%) had either PT or aPTT >1·5 times the midpoint of the normal range, a commonly used trigger for the use of FFP in patients undergoing operative procedures (American Society of Anesthesiologists Task Force on Blood Component Therapy, 1996). None of the patients had a low fibrinogen.

Table II.   Routine coagulation tests and individual factor assays in patients with sepsis compared with controls.
 Control subjects, Mean (SD)Sepsis patients, Mean (SD)
  1. The mean and SD for routine coagulation tests and individual coagulation factors and anticoagulant proteins are shown. C reactive protein and albumin were not performed on normal controls, the normal range stated is the local laboratory range.

  2. *P = 0·03; **P < 0·001.

PT (s)11·7 (0·5)20·1 (5·9)**
APTT (s)27 (3·4)50·1 (36·3)**
Fibrinogen (g/l)2·8 (0·57)5·7 (2·6)**
FII (iu/dl)100 (12·1)60 (30·2)**
FV (iu/dl)116 (22·9)95 (56·4)*
FVII (iu/dl)130 (31·1)55 (30·7)**
FVIII (iu/dl)107 (31·5)257 (113)**
FIX (iu/dl)101 (16·5)111 (51·8)
FX (iu/dl)123 (16·6)72 (40·8)**
FXI (iu/dl)116 (15·7)77 (40·7)**
FXII (iu/dl)125 (27·8)52 (24·1)**
Protein C (%)127 (20)63 (34·3)**
Antithrombin (iu/dl)103 (8)63·1 (28·7)**
C Reactive protein (mg/l)<6181 (98·5)
Albumin (g/l)35–5021·7 (7·4)

The abnormal coagulation screens were caused by decreased levels of coagulation factors II, V, VII, X and XII leading to prolongation of the time to clot formation in these assays (Table II). In the 38 patients with sepsis, 12 (32%) had all haemostatically relevant coagulation factors within the normal range and only eight (21%) had one or more factor below a theoretic haemostatic threshold (<15 iu/dl for factor VII and <30 iu/dl for other coagulation factors). There were 18 patients with a low level of factor XII, which does not contribute to the haemostatic risk but would have led to an abnormal aPTT.

In contrast, the mean factor IX level was not significantly reduced compared with normal controls. Factor IX was not below 30 iu/dl in any patients and was raised in 11 patients. Furthermore, factor VIII was significantly increased in the sepsis patients compared with normal controls (P < 0·001). Factor VIII was not reduced in any patients and was above the laboratory normal range (150 iu/dl) in 32 (84%) patients. Similarly, fibrinogen was increased (above 4 g/l) in 24 of the 38 (63%) patients.

There was a lower mean level of the anticoagulant proteins antithrombin and protein C. In 20 (53%) patients, the level of protein C was below the normal range (<70 iu/dl) and in 25 (66%) patients, the antithrombin level was low (<75 iu/dl). The mean platelet count in the sepsis group was 164 × 109/l (range 15–613) compared with 221 × 109/l (range 169–403) for the controls.

There was a strong correlation between the fibrinogen level and C reactive protein (CRP) (r = 0·51, P < 0·01) suggesting that the raised fibrinogen was secondary to an acute phase reaction. There was no correlation between factor VIII and CRP. Antithrombin has previously been suggested to be negatively correlated with an acute phase reaction (Niessen et al, 1997), but no correlation with CRP was observed in our study. Albumin correlated with factors II (r = 0·67), V (r = 0·52), VIII (r = 0·42), IX (r = 0·85), X (r = 0·58), XI (r = 0·73) and XII (r = 0·6) (P < 0·01), in keeping with these proteins being synthesised in the liver. Albumin also strongly correlated with a decreased antithrombin (r = 0·69) and protein C (r = 0·58) (P < 0·01).

Although almost all patients had abnormal coagulation screenings, the variable levels of procoagulant and anticoagulant proteins means that it was unclear whether an individual patient was at increased risk of bleeding during invasive procedures or of microvascular thrombosis predisposing to multi-organ failure. Global tests of haemostasis in this group of patients were performed.

Thrombin generation assays

Calibrated automated thrombography (CAT) was performed in PRP and PPP and the pattern of results was the same. In critically ill patients with sepsis compared with normal controls, the lag time to thrombin generation was prolonged, peak thrombin was decreased and time to peak thrombin prolonged. The mean total amount of thrombin generated in the group of patients with sepsis as measured by the ETP, however, was equivalent to normal controls (Table III). There was, however, a wide range of thrombin generation in the patient group and thrombocytopenia, leading to decreased platelet count in the PRP, was an important cause of decreased ETP. Representative thrombin generation in PRP and PPP is shown in Fig 1.

Table III.   Calibrated automated thrombography in patients with sepsis compared with controls.
 Control subjects, mean (SD)Sepsis patients, mean (SD)
  1. The mean and SD of thrombin generation assays in platelet rich (PRP) and poor plasma (PPP) in patients compared with controls are shown. The prolongation of the lag times and time to peak thrombin generation suggest a delay in the activation of haemostasis but the normal endogenous thrombin potential (ETP) implies that once thrombin generation has been initiated a similar total amount of thrombin to normal is produced.

  2. *P < 0·02; **P < 0·001.

 Lag time (min)17 (8)29 (22)*
 ETP (nmol/l·min)1395 (488)1334 (512)
 Peak thrombin (nmol/l)76 (40)56 (31)*
 Time to peak (min)32 (12)48 (27)**
 Lag time (min)2·4 (0·9)5·0 (5·4)**
 ETP (nmol/l·min)1681 (281)1642 (452)
 Peak thrombin nmol/l454 (100)344 (146)**
 Time to peak (min)4·2 (1·2)6·7 (6·7)**

Figure 1.  The variability of thrombin generation within the sepsis patient group. (A) Representative thrombin generation for four patients with sepsis (solid lines) compared with a control (dashed lines) in platelet poor plasma. (B) Representative thrombin generation studies for four different patients with sepsis (solid lines) compared with a control (dashed lines) in platelet rich plasma. Most patients had a delayed lag time (note different x-axis scales) but, once initiated, total thrombin generation, as defined by area under the thrombin generation curve, was equivalent to normal. However, some patients had markedly decreased thrombin generation, especially in platelet rich plasma.

Download figure to PowerPoint

These results suggest that, although some patients had a decreased ETP implying a potentially increased risk of bleeding, as a group, once activated, thrombin was generated in normal amounts.

Low concentration tissue factor activated thromboelastography

Compared with normal controls, critically ill patients with sepsis syndrome as a group had a prolonged time to initial clot formation (clot time). However, the alpha angle and mean clot firmness were increased. The first derivative of the clot firmness curve revealed an increased maximum velocity of clot formation and increased area under the clot firmness curve in patients compared with controls (Table IV). These results are similar to those derived from the thrombin generation assays in that there was a delay in activation of haemostasis but, once initiated, clot formation proceeded normally or was exaggerated.

Table IV.   Low concentration tissue factor-activated thrombography in patients compared with controls.
 Control subjects, mean (SD)Sepsis patients, mean (SD)
  1. The mean and SD of each parameter is shown in patients with sepsis and controls. The prolonged clot time indicates a delay in initiation of haemostasis but the raised mean clot firmness, alpha angle, maximum velocity of clot formation and area under the clot firmness curve all support a procoagulant state.

  2. *P < 0·05; **P < 0·005; ***P < 0·0005.

Clot time (s)818 (271)1163 (784)*
Mean clot firmness (mm)51 (12)67 (17)***
Alpha angle (°)47 (15)55 (21)**
Maximum velocity (mm/s)6·5 (3·0)11·2 (7·5)**
Time to maximum velocity (s)1040 (334)1042 (633)
Area under clot firmness curve51 (12)62 (24)***

Low concentration tissue factor-activated MDA waveform assay

The mean (SD) time to clot formation was 75·8 s (21·0) in the control patients compared with 97·4 s (36·9) in the sepsis group (P < 0·01). The group of normal controls had a mean (SD) clot transmission velocity result of 0·99 (0·39)/Ts compared with the sepsis group 1·83 (0·54), P < 0·005. These results were again similar to the CAT and Rotem assays because they suggest that, as a group, the patients with severe sepsis had delayed initiation of clot formation but once activated haemostasis was exaggerated.


  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Competing interests
  7. Acknowledgements
  8. References

The data reported here demonstrated that critically ill patients with sepsis syndrome have abnormal routine coagulation screens and delayed initiation of haemostasis, as measured by three low concentration tissue factor-activated global haemostatic assays. These assays further showed that, once haemostasis was activated, the propagation and clot formation phase was either normal or enhanced.

The three global haemostatic assays showed a similar pattern of results. The CAT assays demonstrated that, whilst initiation of thrombin generation was delayed and peak thrombin reduced, the total amount of thrombin, as measured by the ETP, was unchanged in the patients with sepsis compared with controls. Similarly, whilst the time to clot in the whole blood, low concentration tissue factor Rotem assay and the MDA waveform assay were also delayed, the rate of increasing fibrin clot elasticity and rate of clot formation was enhanced.

The findings of this study may be explained by the measured levels of individual coagulation factors in the light of currently accepted models of haemostasis. Coagulation is thought to be activated through the tissue factor/factor VII pathway, which generates a small amount of thrombin, insufficient to clot fibrinogen, but required for the activation of factors V and VIII and expression of negatively charged phospholipids on platelets. The rapid burst of thrombin generation required to form a stable clot is then driven through the tenase complex of factor IXa and factor VIIIa and the prothrombinase complex of factor Xa and factor Va localised on the negatively charged phospholipid surface of activated platelets (Mann et al, 2003; Roberts et al, 2004). Activation of haemostasis in sepsis has been shown to be through the tissue factor pathway (Taylor et al, 1991; Levi et al, 2003).

In the sepsis group of patients, it is likely that the decreased levels of factors VII, X and II led to a delay in the generation of sufficient initial thrombin to activate factors V and VIII and to stimulate expression of platelet phospholipids. The major components of the intrinsic pathway (factors IXa and VIIIa), however, were preserved or increased in sepsis patients and hence, once the cofactors had been activated, thrombin and fibrin generation proceeded normally or was enhanced. This mechanism also explains the prolonged PT and aPTT tests, which are sensitive to the initiation rather than the propagation phase of haemostasis.

Measurement of individual coagulation factors and anticoagulants in this study demonstrated a pattern similar to those previously reported (Hesselvik et al, 1989). The procoagulant factors II, V, VII, X and XI as well as the anticoagulant factors antithrombin and protein C, were decreased compared with normal. The level of these factors correlated with serum albumin, suggesting that hepatic dysfunction as well as consumption played a role. Some coagulation factors, such as factor VIII and fibrinogen, were increased whilst factor IX was stable. The raised fibrinogen (but not factor VIII) correlated with CRP confirming that the increase was the result of an acute phase reaction. It is unclear what effect the combined changes in individual coagulation factors and anticoagulant have on global haemostasis, especially as some patients were also thrombocytopenic. As a result, it was not clear from the measurement of individual factors whether patients were at increased risk of bleeding, thrombosis or both. It is possible that some parameters of the global haemostatic tests may give information about the balance between pro- and anti-coagulant changes. The low concentration tissue factor-activated MDA waveform assay described here, for example, measures the rate of fibrin clot formation in PPP in the presence of thrombomodulin. This means the assay is sensitive to the protein C concentration in the sample in addition to the procoagulant clotting factors. The CAT and Rotem assays are not sensitive to the reduced levels of antithrombin and protein C and would underestimate the procoagulant state in patients.

In routine clinical practice, clinicians rely on the PT and aPTT to assess a patient's haemostatic status. Prolongation of these assays is interpreted as evidence of an anticoagulated state and FFP is often infused, particularly at the time of invasive procedures. This contrasts with previous findings of raised markers of activation of haemostasis, such as prothrombin fragment 1 + 2 and TAT, in sepsis syndrome (Gando et al, 1998; Mavrommatis et al, 2000; Amaral et al, 2004). We have previously shown that abnormal coagulation screening tests are not a reliable guide to coagulation factors levels in critically ill patients (Chowdhury et al, 2004) and there continues to be a debate regarding the utility of coagulation screens to predict bleeding during invasive procedures in the context of hepatic dysfunction (Segal & Dzik, 2005; Mannucci, 2006; Reverter, 2006). This issue was further highlighted by our finding that factor XII was the most commonly reduced individual coagulation factor in the cohort of septic patients as previously reported (Hesselvik et al, 1989). Reduced factor XII is an important cause of a prolonged aPTT, but it is not required for haemostasis and so the prolongation of the aPTT related to a low factor XII is misleading and likely to be one of the major reasons why the aPTT is of limited utility in assessing clinical bleeding risk in septic patients.

The prolonged coagulation screening tests and the delayed initiation of haemostasis demonstrated by the three global tests probably reflect similar processes. If this is the case then the initiation time in the global assays may also have limited utility in assessing an individual patient's risk of bleeding and need for blood product support. It is possible that other parameters measured on the global assays that assess the propagation phase of haemostasis may be more useful. Although in our study the septic patients, as a group, had some results suggestive of a prothrombotic state, individuals had parameters of global haemostasis below the normal range, which may indicate an increased risk of bleeding during invasive procedures.

The present study aimed to establish the effect of sepsis on global haemostatic assays. It did not aim to link these results to either bleeding or thrombotic complications in these patients and patients who were bleeding were not recruited into the study. The role of these assays in the management of patients remains to be defined. Further studies are underway to link clinical relevant bleeding endpoints to global assays of haemostasis in this patient population.

Baseline levels and dynamic changes in haemostatic factors have been associated with outcome in critically ill patients with sepsis and used as evidence to support a pathophysiological link between coagulation abnormalities and multi-organ failure (Dhainaut et al, 2005). Biphasic clot waveform has also been shown to predict clinical outcome in sepsis (Toh et al, 2003). There were too few patients included in this study to investigate whether any parameters measured were predictive of prognosis or outcome.

We conclude that global assays activated by low concentrations of tissue factor show that critically ill patients with sepsis have delayed activation of haemostasis but, once initiated, thrombin and clot formation are normal or enhanced. The results may help to explain why routine coagulation screen tests poorly predict bleeding in sepsis. Whether global assays will be useful in clinical practice will only be established once they are linked to relevant end points of bleeding, end organ failure and clinical outcomes.

Competing interests

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Competing interests
  7. Acknowledgements
  8. References

The Rotem machine and consumables used in this study were supplied to the haematology department in Cardiff without charge by Sysmex UK Limited, the company that was marketing the instrument in the UK at the time of the study.


  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Competing interests
  7. Acknowledgements
  8. References

The study was supported by a Research and Development grant from the Cardiff and Vale NHS Trust.

Laboratory tests were performed by N. Macartney, G. Florou J. Knippings, E. Stephenson, L. Hathaway and R. Davies. S. Lees helped to establish the CAT assays. Anthony Wilkes contributed to the statistical analysis of the data.


  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Competing interests
  7. Acknowledgements
  8. References
  • Al Dieri, R., Peyvandi, F., Santagostino, E., Giansily, M., Mannucci, P.M., Schved, J.F., Beguin, S. & Hemker, H.C. (2002) The thrombogram in rare inherited coagulation disorders: its relation to clinical bleeding. Thrombosis and Haemostasis, 88, 576582.
  • Amaral, A., Opal, S.M. & Vincent, J.-L. (2004) Coagulation in sepsis. Intensive Care Medicine, 30, 10321040.
  • American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference (1992) Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Critical Care Medicine, 20, 864874.
  • American Society of Anesthesiologists Task Force on Blood Component Therapy (1996) Practice guidelines for blood component therapy: a report by the American Society of Anesthesiologists Task Force on Blood Component Therapy. Anesthesiology, 84, 732747.
  • Braun, P., Givens, T.B., Stead, A.G., Beck, L.R., Gooch, S.A., Swan, R.J. & Fischer, T.J. (1997) Properties of optical data from activated partial thromboplastin time and prothrombin time assays. Thrombosis and Haemostasis, 78, 10791087.
  • Chowdhury, P., Saayman, A.G., Paulus, U., Findlay, G.P. & Collins, P.W. (2004) Efficacy of standard dose and 30 ml/kg fresh frozen plasma in correcting laboratory parameters of haemostasis in critically ill patients. British Journal of Haematology, 125, 6973.
  • Dempfle, C.-E. (2004) Coagulopathy of sepsis. Thrombosis and Haemostasis, 91, 213224.
  • Dhainaut, J.F., Shorr, A.F., Macias, W.L., Kollef, M.J., Levi, M., Reinhart, K. & Nelson, D.R. (2005) Dynamic evolution of coagulopathy in the first day of severe sepsis: relationship with mortality and organ failure. Critical Care Medicine, 33, 341348.
  • Dixon, B. (2004) The role of microvascular thrombosis in sepsis. Anaesthesia and Intensive Care, 32, 619629.
  • Gando, S., Nanzaki, S., Sasaki, S., Aoi, K. & Kemmotsu, O. (1998) Activation of the extrinsic coagulation pathway in patients with severe sepsis and septic shock. Critical Care Medicine, 26, 20052009.
  • Hemker, H.C., Giesen, P., Al Dieri, R., Regnault, V., De Smedt, E., Wagenvoord, R., Lecompte, T. & Béguin, S. (2003) Calibrated automated thrombin generation measurement in clotting plasma. Pathophysiology of Haemostasis and Thrombosis, 33, 415.
  • Hesselvik, J.F., Blomback, M., Brodin, B. & Maller, R. (1989) Coagulation, fibrinolysis, and kallikrein systems in sepsis: relation to outcome. Critical Care Medicine, 17, 724733.
  • Levi, M., Keller, T.T., Van Gorp, E., Ten Cate, H. (2003) Infection and inflammation and the coagulation system. Cardiovascular Research, 60, 2639.
  • Luddington, R. & Baglin, T. (2004) Clinical measurement of thrombin generation by calibrated automated thrombography requires contact factor inhibition. Journal of Thrombosis and Haemostasis, 11, 19541959.
  • Mann, K.G., Brummel, K. & Butenas, S. (2003) What is all the thrombin for? Journal of Thrombosis and Haemostasis, 1, 15041514.
  • Mannucci, P.M. (2006) Abnormal haemostasis tests and bleeding in chronic liver disease: are they related? No. Journal of Thrombosis and Haemostasis, 4, 721723.
  • Mavrommatis, A.C., Theodoridis, T., Orfanidou, A., Roussos, C., Christopoulou-Kokkinou, V. & Zakynthinos, S. (2000) Coagulation system and platelets are fully activated in uncomplicated sepsis. Critical Care Medicine, 28, 451457.
  • Niessen, R.W., Lamping, R.J., Jansen, P.M., Prins, M.H., Peters, M., Taylor, F.B. Jr, De Vijlder, J.J., Ten Cate, J.W., Hack, C.E. & Sturk, A. (1997) Antithrombin acts as a negative acute phase protein as established with studies on HepG2 cells and in baboons. Thrombosis and Haemostasis, 78, 10881092.
  • Reverter, J.C. (2006) Abnormal haemostasis tests and bleeding in chronic liver disease: are they related? Yes. Journal of Thrombosis and Haemostasis, 4, 717720.
  • Roberts, H.R., Monroe, D.M. & Escobar, M.A. (2004) Current concepts of hemostasis: implications for therapy. Anesthesiology, 100, 722730.
  • Segal, J.B. & Dzik, W.H. (2005) Paucity of studies to support that abnormal coagulation test results predict bleeding in the setting of invasive procedures: an evidence-based review. Transfusion, 45, 14131425.
  • Sorensen, B., Johansen, P., Christiansen, K., Woelke, M. & Ingerslev, J. (2003) Whole blood coagulation thrombelastographic profiles employing minimal tissue factor activation. Journal of Thrombosis and Haemostasis, 1, 551558.
  • Taylor, F.B. Jr, Chang, A., Ruf, W., Morrissey, J.H., Hinshaw, L., Catlett, R., Blick, K. & Edgington, T.S. (1991) Lethal E. coli septic shock is prevented by blocking tissue factor with monoclonal antibody. Circulatory Shock, 33, 127134.
  • Toh, C.H., Ticknor, L.O., Downey, C., Giles, A.R., Paton, R.C. & Wenstone, R. (2003) Early identification of sepsis and mortality risks through simple, rapid clot-waveform analysis. Implications of lipoprotein-complexed C reactive protein formation. Intensive Care Medicine, 29, 5561.