Thromboelastography (TEG) is a relatively fast and accurate tool for bedside monitoring of coagulation disorders. It represents a global coagulation test of factors, interaction between thrombocytes and fibrin fibres, and fibrinolysis.1 Both TEG and standard coagulation tests are usually carried out at a laboratory temperature of 37°C, disregarding the effects of real blood temperature. Hypothermia significantly interferes with coagulation2,3 through enzyme inhibition, thrombocyte dysfunction and changes in fibrinolysis.4 At a temperature of 33°C, the first impairment of primary adhesion and aggregation of thrombocytes becomes evident.5 The difference between the in-vivo and laboratory measurement temperature may give rise to misinterpretation of the results obtained by both TEG6,7 and also standard coagulation tests.8,9 The sources of these investigations fail to advise on whether TEG should be adjusted for temperature differences.
Data from Venema et al.10 have recently reported that reproducibility of TEG coagulation variables is poor. We decided to compare temperature adjusted and non-adjusted TEG measurements in clinical practice to find out whether it is necessary to correct TEG variables for in-vivo temperature, considering its relatively low reproducibility.
The aim of this study was to evaluate the relevance of temperature adjustment in patients during hypothermia. We hypothesised that the temperature adjustment would not be important because of low TEG reproducibility.
The study was approved by the local Ethics Committee of the St Anne's University Hospital Brno, Czech Republic (No. NS-10097–3; Chairman Vladimir Soska, MD, PhD; 01/02/2008) and written consent was obtained from all patients or next of kin. It was performed between January 2009 and July 2010 at the Department of Anaesthesiology and Intensive Care, St Anne's University Hospital Brno, Czech Republic. All TEG measurements were performed using Haemoscope (Haemoscope Corp., Sokie, Illinois, USA)1 by the first author with the support of a trained research nurse.
This prospective observational study was performed on survivors of cardiopulmonary resuscitation (CPR) with known cardiac disease in whom therapeutic hypothermia (32 to 34°C) was indicated for 24 h. No specific exclusion criteria were defined.
Therapeutic hypothermia was performed according to a local ICU protocol independent of the study. Continuous temperature measurements were obtained from a Swan-Ganz catheter, a continuous pulse contour cardiac analysis catheter or thermosensor from the nasopharynx. During the study, all patients were under deep sedation using a combination of propofol, sufentanil and midazolam and were observed for 36 h. Each patient had arterial and central venous catheters. Every TEG measurement for a given individual taken during the study used the same invasive access – arterial catheter in the majority of cases – because of known arteriovenous coagulation differences.11
TEG was analysed in 12-h intervals as follows: before hypothermia (T0), at 12 h of hypothermia (T1), at 24 h of hypothermia (T2) and after 12 h of rewarming (T3). A freshly drawn blood sample using a 22-gauge needle and ml syringe from the invasive access as mentioned above was divided into four aliquots and analysed immediately and simultaneously on two thromboelastographs. Two samples were activated with kaolin for 2 min and then analysed with additional heparinase. Heparinase type 1 and 2 were used to eliminate the effects of unfractioned heparin, low molecular weight heparin (LMWH) and heparin-like substances which are released during hypothermia.12 One sample was analysed at the actual temperature (adjusted) of the patient and the other one at 37°C (non-adjusted). The other two samples were analysed using Rapid-TEG analysis, a novel method that shortens analysis time significantly by using kaolin to activate the ‘internal pathway’ and recombinant tissue factor to activate the ‘external pathway’ of the coagulation cascade also. These samples were also analysed at the actual body temperature of the patient (adjusted) and at 37°C (non-adjusted). Table 1 shows TEG variables, a brief description of their characteristics and normal ranges from the official TEG manual. Rapid-TEG variables are similar to the standard TEG variables except for the activated clotting time (ACT) which replaces the R value. The ACT is the time in seconds from the start of measurement to the first fibrin build-up.
Data are presented as median (lower quartile, upper quartile). Bland–Altman plots were used to analyse the agreement between temperature adjusted and non-adjusted TEG measurements. For better understanding of the significance of the changes, bias is presented as a relative mean of differences between temperature adjusted and non-adjusted samples ((variableadjusted−variablenon-adjusted)/((variableadjusted + variablenon-adjusted)/2)×100. Bias is presented together with limits of agreement, which are defined as bias +/−1.96 standard deviations (SD). Statistical analysis was performed with Statistica CZ 9.1. (StatSoft s.r.o., Prague, Czech Republic).
We recruited 30 survivors of CPR (22 men and eight women) who provided 400 samples. Three patients died during the study. Median age was 65.5 (52 to 74) years and the length of CPR was 15 (10 to 20) min. The primary cause of cardiac arrest was malignant arrhythmia, which in 12 patients complicated acute myocardial infarction (AMI). Body temperatures were 35.9°C (35.4 to 36.0°C) at T0; 33°C (32.6 to 33.9°C) at T1; 33.1°C (32.8 to 33.9°C) at T2 and 36.4°C (36.0 to 36.9°C) at T3.
Table 2 shows results of Bland–Altman analysis of temperature adjusted and non-adjusted samples in Kaolin–Heparinase-TEG and Rapid-TEG.
Whole results of Kaolin–Heparinase-TEG are presented in Supplement Table 1, http://links.lww.com/EJA/A29. Whole results of Rapid-TEG are presented in Supplement Table 2, http://links.lww.com/EJA/A29.
In hypothermia, biases for clot formation variables were −15.1% for R, −16% for K, −19% for TMA (time to reach maximum amplitude) and 11% for Angle, respectively. Bias for clot strength variables was low for both MA (0%) and G (1%). Bias for fibrinolysis variable Ly30 was also low (0.7%). Biases at temperatures close to 37°C (time points T0, T3) were low for clot formation, fibrinolysis and strength Kaolin–Heparinase variables (ranging from −3 to +3%).
The limits of agreement between the two methods (temperature adjusted and non-adjusted) were wide for clot formation variables in normothermia (ranging from −48 to 42%) and also in hypothermia (ranging from −55 to 37%). Limits of agreement for clot strength variables were as follows: MA (−16 to 13%), G (−39 to 33%) in normothermia and MA (−9 to 10%) and G (−29 to 30%) in hypothermia, respectively. Limits of agreement for fibrinolysis variable Ly30 were (−2.24 to 3.78%) during hypothermia and (−2.59 to 2.16%) during normothermia. For further details, see Fig. 1.
In hypothermia, bias was the highest for clot formation variables R (−25%) and ACT (−16%). Other clot formation biases were −13% for K, 8% for Angle and −9% for TMA. Biases for fibrinolysis and clot strength variables were low (0.6% for Ly30, 1% for MA and 2% for G). During normothermia, biases for all measured variables (clot formation, clot strength and fibrinolysis) ranged from −1 to −10%.
The limits of agreement for Rapid-TEG clot formation variables were wide ranging: from −114 to 65% in hypothermia and from −117 to 97% in normothermia. Limits of agreement for clot strength variables were −27 to 24% for MA and −64 to 56% for G in normothermia and −18 to 20% for MA and −50 to 59% for G in hypothermia. Limits of agreement for fibrinolysis variable Ly30 were (−2.37 to 3.65%) during hypothermia and (−2.32 to 3.18%) during normothermia. For further details, see Fig. 2.
The clot strength and fibrinolysis variables had low ranges of bias both in normothermia and hypothermia. The clot formation variables were influenced by temperature adjustment during hypothermia (presence of systematic bias around −17%), but the wide limits of agreement make this finding difficult to interpret. Differences between temperature adjusted and non-adjusted samples in TEG measurements were, in the setting of clinical routine, not important.
Differences in Kaolin–Heparinase-thromboelastography and Rapid-thromboelastography caused by temperature adjustment
During hypothermia in Kaolin–Heparinase-TEG and also partially in Rapid-TEG, the clot formation variables showed relatively small systematic bias when adjusted to in-vivo temperature. Bias for Kaolin–Heparinase-TEG ranged from −15 to −19% and for Rapid-TEG from −9 to −25% compared to normothermia in which bias ranged from −3 to 3% for Kaolin–Heparinase-TEG and from −10 to 2% for Rapid-TEG. The biases for clot strength and fibrinolysis variables were low and were not influenced by temperature adjustment.
The presence of systematic bias could be explained by the ability of temperature adjusted TEG to measure the hypothermia-induced coagulation changes. To explain the presence of systematic bias in clot formation variables, we compared paired data using the Wilcoxon matched pair test during different temperature adjustments. In Kaolin–Heparinase-TEG, we found diminished clot formation during hypothermia, which is in agreement with the experimental study in rabbits by Shimokawa et al.,13 clinical studies in liver transplant patients by Douning et al.7 and in patients under general anaesthesia by Kettner et al.14 Similarly, variable MA was without significant change. In Rapid-TEG, which has not been tested in hypothermia so far, the plasma coagulation variables (R, TMA and ACT) were slightly different when temperature adjustment was done. These differences were more pronounced for ACT (for these results, see supplement Table 1 and supplement Table 2, http://links.lww.com/EJA/A29).
Nevertheless, the presence of systematic bias was hampered by wide limits of agreement that were independent of temperature adjustment. Limits of agreement for clot strength and fibrinolysis variables were narrower and independent of temperature adjustment also.
Risk of bleeding in patients following cardiopulmonary resuscitation
We decided to study patients after CPR because of the high risk of complications arising from haemorrhage in this group. Bleeding can develop as a result of CPR trauma, anticoagulation and antiaggregation therapy, and can be exacerbated by deteriorating coagulopathy. Implicated in the latter are several states that occur following CPR, such as hypoperfusion,15 acidosis16 and therapeutic hypothermia, which is a part of routine care.17 Therefore, we believe that monitoring coagulopathy and identifying its causes is important in this group of patients and can influence the clinical decisions of pharmacotherapy such as treatment with LMWH.
Comment on dropouts, limitations
Some samples at separate time points were missed for a number of reasons that included the death of three patients, insufficient time to perform TEG before therapeutic hypothermia was commenced and technical problems. We believe we obtained an adequate number of measurements to justify our conclusion.
Choice of statistical method
We decided to use the Bland–Altman method for analysis of the agreement between temperature adjusted and non-adjusted TEG measurements. The purpose of this study was not to answer whether hypothermia induced coagulation changes, but to compare two methods of TEG analysis. All the measurements were done simultaneously on two TEG machines (one temperature adjusted and the other one non-adjusted) and we think that for this kind of analysis, the Bland–Altman method is appropriate.
Reproducibility of thromboelastography measurements and other subgroups
One of the advantages of TEG is the ability to perform measurements at different temperatures at the bedside, which is not true for standard laboratory tests. Our findings of the wide limits of agreement and relatively small systematic bias suggest that the importance of temperature adjustment in patients after CPR is low. Data about low reproducibility of clot formation variables in TEG measurements have recently been published,10 but this might be less relevant in bleeding trauma patients, in whom the pathophysiology might be more disturbed, giving a wider range of TEG measurements. It would be interesting to find out how wide the limits of agreement are in these cases. We hypothesise that even with wide limits of agreement, pronounced changes in coagulation would not go undetected.
Although analysis of TEG adjusted to actual temperature during hypothermia yields results with small systematic bias, the relevance of temperature adjustment in clinical routine is low because of the precision limits of TEG measurement itself. Therefore, we see no need to perform TEG analysis at the in-vivo temperature.
Assistance with the study: The authors would like to thank Maria Majkus for language correction of the article.
Financial support and sponsorship: The study was supported by Grant IGA MZCR NS 10097–3.
Conflicts of interest: none declared.
1. Luddington RJ. Thrombelastography/thromboelastometry. Clin Lab Haematol
2. Reed RL 2nd, Johnson TD, Hudson JD, Fischer RP. The disparity between hypothermic coagulopathy and clotting studies. J Trauma
3. Rohrer MJ, Natale AM. Effect of hypothermia on the coagulation cascade. Crit Care Med
4. Britt LD, Dascombe WH, Rodriguez A. New horizons in management of hypothermia and frostbite injury. Surg Clin North Am
5. Lier H, Krep H, Schöchl H. Coagulation management in the treatment of multiple trauma. Anaesthesist
6. Ramaker AJ, Meyer P, van der Meer J, et al. Effects of acidosis, alkalosis, hyperthermia and hypothermia on haemostasis: results of point of care testing with the thromboelastography
analyser. Blood Coagul Fibrinolysis
7. Douning LK, Ramsay MA, Swygert TH, et al. Temperature corrected thrombelastography in hypothermic patients. Anesth Analg
8. Wolberg AS, Meng ZH, Monroe DM 3rd, Hoffman M. A systematic evaluation of the effect of temperature on coagulation enzyme activity and platelet function. J Trauma
9. Fukuda A, Ishida H, Kubota M, et al. Clinical data obtained through coagulation testing suggests that hypothermia exerts influence on a patient's blood coagulation reaction. Rinsho Byori
10. Venema LF, Post WJ, Hendriks HGD, et al. An assessment of clinical interchangeability of TEG and RoTEM thromboelastographic variables in cardiac surgical patients. Anesth Analg
11. Naimi S, Goldstein R, Proger S. Studies of coagulation and fibrinolysis of the arterial and venous blood in normal subjects and patients with atherosclerosis. Circulation
12. Paul J, Cornillon B, Baguet J, et al. In vivo release of a heparin-like factor in dogs during profound hypothermia. J Thorac Cardiovasc Surg
13. Shimokawa M, Kitaguchi K, Kawaguchi M, et al. The influence of induced hypothermia for hemostatic function on temperature-adjusted measurements in rabbits. Anesth Analg
14. Kettner SC, Sitzwohl C, Zimpfer M, et al. The effect of graded hypothermia (36 degrees C-32 degrees C) on hemostasis in anesthetized patients without surgical trauma. Anesth Analg
15. Brohi K, Cohen MJ, Ganter MT, et al. Acute traumatic coagulopathy: initiated by hypoperfusion. Ann Surg
16. Lier H, Krep H, Schroeder S, Stuber F. Preconditions of hemostasis in trauma: a review. The influence of acidosis, hypocalcemia, anemia, and hypothermia on functional hemostasis in trauma. J Trauma
17. Nolan JP, Morley PT, Vanden Hoek TL, et al
. Therapeutic hypothermia after cardiac arrest. Circulation