Demographics and Injury Patterns
The mean age for the study group was 36.4 years. The means for the individual temperature groups ranged from 33 to 43, but the differences were not statistically significant (p > 0.72). Likewise there were no statistically significant differences between the temperature groups on gender or mechanism of injury. The vast majority of the included patients suffered from blunt injury (87%), which is typical of the population at the study institution.
Downtime for this study was calculated as the time from injury until the time blood was drawn. The mean downtime for the entire group was 54 minutes. The groups did not differ significantly, and the range for each of the groups was similar as well (Table 1).
Patients were classified as to the presence or absence of head, chest, abdominal, or extremity injury by examining ICD-9 code diagnosis. The percentage of patients in each temperature group with the injury present is given in Table 1. Whereas there are slightly more head injuries in the hypothermic groups, there were no statistically significant differences between any of the hypothermic groups and the nonhypothermic control group on the percentage of injuries for head, chest, abdomen, or extremities (p > 0.72).
The ISS was significantly higher in the temperature group 34.0 to 34.9[degree sign]C when compared with the nonhypothermic (temperature >or=to 37.0[degree sign]C) control group (p < 0.05). A progressive increase in ISS could also be seen across the temperature groups, with the colder patients tending to have a higher ISS.
The two coldest groups (34.0-34.9[degree sign]C and 33.0-33.9[degree sign]C) had significantly lower Glasgow Coma Scale scores than the nonhypothermic (<or=to37.0[degree sign]C) control group (p < 0.001). A linear decrease in Glasgow Coma Scale scores could also be seen as temperature fell.
The mean RTS for the two coldest groups [23,24] were significantly lower than the nonhypothermic control group (p < 0.001), and a linear decrease in RTS was noted as temperature fell. There were no significant differences in vitals signs (heart rate, blood pressure) between the groups on admission to the trauma receiving area (p > 0.38).
The coldest group (group 33) had a total hospital length of stay that was more than double the control group (11.3 vs. 4.8; p < 0.037), and their overall mortality rate was 43% compared with 5% in the nonhypothermic control group (p < 0.001).
There was no significant correlation between vital signs and fluid administered and there were no groupwise differences on amount of fluid administered. Even though groupwise differences were not found, there remained a strong correlation between temperature and the amount of fluid infused (r = -0.30; p = 0.002), and temperature and amount of blood infused (r = -0.26; p = 0.005). There was a similar correlation between injury severity and fluid infused (r = 0.31; p = 0.002). As suspected, there was also a correlation between downtime and the amount of fluid received (r = 0.32; p = 0.006). To attempt to sort out the differential contributions to coagulation, the correlations were repeated, controlling for (partialing out) the contribution of the intercorrelated factors (fluid and temperature). When the contribution of temperature is controlled for in the correlation of fluid and coagulation, the correlation becomes nonsignificant. However when fluid is controlled for in the correlation of temperature and coagulation, the r decreases slightly but remains significant. This finding indicates that for this sample, temperature has independent effects on coagulation not mediated by fluid infused, whereas the effects of fluid infused on coagulation seem to be significant only when mediated by temperature.
PT and aPTT did not differ among the temperature groups (p > 0.58), and all means were within normal limits. This finding is not unexpected, because on average it takes a 30 to 40% drop in factor activity for the aPTT/PT to become abnormal.  This finding is in contrast to the TEG coagulation index which is extremely sensitive to even small changes. Results of the TEG coagulation index test indicated that in fact, on average, all of the groups were hypercoagulable overall except for the coldest group (33.0-33.9[degree sign]C). The coldest group was on the hypocoagulable side of normal and was significantly different from the nonhypothermic control group (temperature >or=to 37.0[degree sign]C) (p < 0.001).
All groups had mean platelet counts within normal limits (reference range, 140-400 for the study institution) and none of the hypothermic groups differed significantly from the nonhypothermic control group (p > 0.18). A downward trend in number of platelets was noted as the mean group temperature decreased, however.
Mean hemoglobin and hematocrit levels were within normal limits for both men and women for all groups except the 34.0-34.9[degree sign]C group (group 34). Group 34 had a large number of patients with very low hemoglobin and hematocrit levels.
Systemic acidosis by itself may also alter coagulation. Serum carbon dioxide (CO2) correlates highly with arterial pH and is an accurate indication of acid-base balance. [25-27] The reference range for normal serum CO2 for the study institution is 21 to 30 mEq/L. In this study, the two coldest groups (34.0-34.9[degree sign]C and 33.0-33.9[degree sign]C) had below normal CO2 levels, and were significantly lower than the normothermic control group (p < 0.001). This finding infers that the two coldest groups had a serum pH that was more acidotic than the nonhypothermic (temperature >or=to 37.0[degree sign]C) control group. Acidosis, however, did not correlate significantly with injury severity (r = -0.15; p = 0.11) but did correlate significantly with temperature (r = 0.44; p < 0.001). When injury severity was controlled for (partialed out) in the correlation of CO2 and temperature, there was still an r of 0.39 of temperature with CO2 (p < 0.001). Temperature therefore accounted for six times more of the variation in acidosis as did injury severity. So for this sample, acidosis seems more likely to be a sequelae to hypothermia, compared with injury severity. Therefore in this study, acidosis most likely contributes to coagulopathy primarily as a result of hypothermia, not independent of it.
The results of this study demonstrate that only the very coldest trauma patients (33.0-33.9[degree sign]C) are significantly different from the nonhypothermic patients (>or=to37.0[degree sign]C) on any of the study variables. These patients had significant alterations in measures of platelet function (K time and MA) and in the enzyme function (angle). None of the groups differed significantly on either of the measures of fibrinolysis.
According to some sources, hemorrhage alone can lead to hypothermia.  In this study, however, there was no significant correlation between hemorrhage and hypothermia (r < 0.17; p > 0.07). This is similar to the findings of Luna et al. (1987), who found no difference in admission temperature between those patients who had severe hemorrhage requiring transfusion of 1 to 10 units of blood and patients who did not have severe hemorrhage.  Sori et al. (1987), using a rat model, also found no significant drop in temperature related to blood volume loss alone until there was a >50% blood loss accompanied by shock.  Multiple other studies confirm this finding that whereas hemorrhage may lead to hypothermia, hypothermia seems to be related to the resultant shock state rather than merely a volume loss. [8,28,30,31] In this study as well, hemorrhage did not seem to be an independent contributor to hypothermia.
Severe trauma is invariably accompanied by large amounts of soft-tissue damage, and the resulting release of tissue thromboplastin may lead to hypercoagulability and tissue hypoxia. Hypercoagulability is a common finding in trauma patients. [23,24,32-35] Reported rates range from 65% to 97% of severely injured patients presenting with some degree of hypercoagulability. [10,23,35,36] In this study, 68% of the patients had a TEG coagulation index indicating hypercoagulability (TEG coagulation index >2.0).and every temperature group except for the 33.0-33.9[degree sign]C group actually had a hypercoagulable group mean (>2.0) on the coagulation index (Table 1).
Resuscitation volume in this study was measured as amount of fluid (crystalloid, colloid, and blood) infused before the blood being drawn. Fluids may cause or exacerbate hypothermia if they are not sufficiently warmed, and may cause coagulopathy through dilution of clotting factors. [37-45] For this sample, temperature had independent effects on coagulation not mediated by fluid infused, whereas the effects of fluid infused on coagulation seem to be significant only when mediated by temperature.
The result obtained in this study is somewhat different from that commonly reported in the literature. It has generally been reported that the more fluid a patient receives, the colder they will become.  This finding is, however, dependent on many factors such as age, injury type, and shock state.  The difference found in this study may be attributable to several reasons. First of all, most studies of resuscitative fluid looked at patients in shock. [39,47-49] In this study, few patients (n = 4) met the criteria for shock upon hospital arrival and so were not aggressively resuscitated. Second, the study sample had very conservative fluid resuscitation in general, because patient temperature and fluid amount were assessed upon arrival to the trauma receiving area and fluid infused was only calculated to the point when blood was drawn. [50,51] At the point TEG bloods were drawn, most patients had received fluids only from the prehospital paramedics. Because the transport times were very short (mean, <20 minutes), the opportunity to receive fluids was very brief for patients in this study, unlike studies that examined patients through several hours of resuscitation. [47,51-55] Only 5% of the patients received over 1,000 mL of fluid. The finding of no independent effect of fluid, whereas correct for this study, are probably not representative of the effects of fluid resuscitation at higher volumes over longer time periods.
To summarize, although hemorrhage can affect hypothermia, in this sample, it seems to be a minor contributor. Mild intravascular coagulation in the form of hypercoagulability may be present in all but the coldest patients, yet its independent contribution to acidosis by means of tissue hypoxia is small compared with the contribution of hypothermia alone. Acidosis in the sample contributes to coagulopathy mainly as a result of hypothermia, not independent of it. The amount of fluid transfused is significantly related to coagulation, but primarily as mediated by temperature, not independent of it. It is clear for the sample that, although many factors may affect coagulation, temperature is clearly the major independent contributor to coagulation alteration.
The last area of discussion then, is the relationship of hypothermia to coagulopathy, which was the target of this study. On the first research question, there were no significant differences between temperature groups on the r time. This is despite that the coldest group (33.0-33.9[degree sign]C) had a mean r time that was nearly 2 mm (60 seconds) different from the control group. In fact, all groups had mean r times that were shorter than the stated reference range of 10 to 14 mm (5-7 minutes), indicating a hypercoagulable status in general. This is consistent with the general findings of hypercoagulability in trauma patients discussed earlier. [23,35,36]
Although the r time was shortened, once clotting had been initiated, the clots of the coldest group of patients (33.0-33.9[degree sign]C) grew at a significantly slower rate than the clots of the nonhypothermic control group (>or=to37[degree sign]C), as evidenced by a much smaller mean angle (55.5 degrees vs. 69.4 degrees). Although these findings support in part the theory of enzyme inhibition, but one must wonder why the angle was significantly different, while the r time was not. Part of the reason may be simply due to a lack of power for the comparison. With seven patients in the group, sample size is small, and combined with a small effect size (0.048), power is low at 0.41. Previous studies who found that r time was decreased used a paired t test using the patients as their own controls. [9,10,12,13] In this way they were able to control more of the variance. This idea is supported by the fact that the means for this study are consistent with those obtained by the other research groups. In addition, even though the difference is not statistically significant, there is some effect for temperature, as the coldest group (33.0-33.9[degree sign]C) had the longest r time by far, consistent with the literature. An alternative explanation as to why angle was significant and r time was not may be that because r time measures the time to initial clot formation, the large amount of tissue thromboplastin available secondary to the trauma ensures prompt initiation of clotting in these patients despite depression of enzyme function. However, because there is no surplus of the other factors in the cascade, the enzymatic reactions necessary for continued growth are slowed by the colder temperatures. It is clear that once patient temperature drops below 34[degree sign]C that the blood clot grows at a much slower rate than for patients whose temperature is normal.
The second research question examined platelet function, and both measures of platelet function (K and MA) were significantly lower for the patients whose temperature was below 34[degree sign]C (group 33). They had decreased platelet function in the polymerization phase (K time) and also had decreased MA or strength of their clots. The theory of platelet alteration and sequestration as originally proposed, and later replicated, is that platelets are sequestered in the liver and spleen during hypothermia. [17,56] Others found that along with sequestration, multiple morphologic changes in the platelets were present.  These theories indicate that, for the platelet variance component to be altered fully, there needs to be "in vivo" hypothermia, not just the cooling of blood from a healthy sample. This theory would explain the mixed results obtained from those who used in vitro cooling or warming methods. [9,11-13,57,58]
It is difficult to determine from this study whether hypothermia produces a (relative) quantitative platelet deficiency or whether there is a qualitative deficiency as well. Although the mean platelet count for all of the groups was quantitatively adequate by the reference range for the study institution (range, 140-400 cells), there was a strong correlation between the patients absolute platelet count and both the K time (r > 0.47; p < 0.001) and MA (r > -0.42; p < 0.001). This finding implies that although the platelets were within "normal limits," their absolute numbers might still play a role in clotting. Indeed, platelet count was trending downward in the colder groups, even though the difference was not significant (Table 1). Although there is clearly an alteration in platelets for the coldest group, there is no way to determine definitively from the thrombelastogram whether it is a quantitative or qualitative deficiency, or if it is caused by platelet alteration or sequestration.
The third and final research question dealt with fibrinolysis. No significant differences were found between any of the temperature groups on either of the fibrinolytic measures (ly30, ly60). Again, part of the reason for no significance might be attributed to a lack of statistical power for the multivariate test. There is also a great deal of within-group variation, particularly in the coldest group, which would make it more difficult to obtain significant statistical differences. The most simple alternative explanation may be that because all of the measures of fibrinolysis were well within normal limits (ly30 < 7.5; ly60 < 15.0), and because the thrombelastogram studies agree, that fibrinolysis is simply not an important variable in coagulation at these temperatures for these patients. Indeed this theory is supported by the fact that Kearney et al. (1992) found fibrinolysis only in the patient with severe head injury without regard to temperature, and Gando et al. (1992) found no evidence for a fibrinolytic activity in any trauma patients, including those patients with head injuries. [32,33] There remains some evidence in the literature that fibrinolysis plays a role, and more study is warranted. [59-61]
It is easy to try to attribute coagulation alterations to the magnitude of injury, because there is somewhat of a reciprocal relationship between temperature and injury severity. However, that temperature group 34.0 to 34.9[degree sign]C had the highest injury severity, yet did not differ significantly from the nonhypothermic control group is evidence that injury severity alone is not the answer. Several other studies have also found that coagulation alterations are indeed independent of injury severity. [4,32,41]
Enzyme activity slowing and decreased platelet function individually contribute to hypothermic coagulopathy in trauma patients, particularly below 34.0[degree sign]C. All the coagulation measures affected are part of the polymerization process of platelets and fibrin, and this may be the mechanism by which the alteration in coagulation occurs.
In the sample, while there were significant differences on measures of coagulation for the coldest group (33.0-33.9[degree sign]C), only one measure (MA) was outside of the "normal" reference range. All of the subjects in the other temperature group exhibited hypercoagulable test results. So while hypothermia does affect coagulation, the hypercoagulability that is the result of severe trauma seems to allow the body to compensate for some of this hypothermic alteration.
Hypothermia and coagulation are complicated variables, potentially affected by injury severity, hemorrhage, and environment. Further research is required to look at a broader range of temperatures in a multicenter trial to obtain larger numbers in the coldest groups and to more completely evaluate this vital area of trauma care.
1. Sauaia A, Moore F, Moore E, et al. Epidemiology of trauma deaths: a reassessment. J Trauma. 1995;38:185-193.
2. American College of Surgeons. Advanced Trauma Life Support. Chicago, Ill: First Impression; 1993.
3. Patt A, McCroskey B, Moore E. Hypothermia induced coagulopathies in trauma. Surg Clin North Am. 1988;68:775-785.
4. Ferrara A, MacArthur J, Wright H, Modlin I, McMillen M. Hypothermia and acidosis worsen coagulopathy in the patients requiring massive transfusion. Am J Surg. 1990;160:515-518.
5. Britt L, Dascombe W, Rodriguez A. New horizons in management of hypothermia and frostbite injury. Surg Clin North Am. 1991;71:345-370.
6. Reed R, Bracey A, Hudson J, et al. Hypothermia and blood coagulation: dissociation between enzyme activity and clotting factor levels. Circ Shock. 1990;32:141.
7. Luna G, Maier R, Pavlin E, Anardi D, Copass M, Oreskovich M. Incidence and effect of hypothermia in seriously injured patients. J Trauma. 1987;27:1014-1018.
8. Jurkovich G, Greiser W, Luterman A, Curreri P. Hypothermia in trauma victims: an ominous predictor of survival. J Trauma. 1987;27:1019-1024.
9. Douning L, Ramsay M, Swygert T, et al. Temperature corrected thrombelastography in hypothermic patients. Anesth Analg. 1995;81:608-611.
10. Dwyer K, Watts D, Kaufmann C, Dols S, Crews J, Trask A. Detecting coagulation alterations in adult trauma victims with mild hypothermia using thrombelastography: a pilot study. Paper presented at the 56th Annual Meeting of the American Association for the Surgery of Trauma, September 19-21, 1996, Houston, Texas.
11. Stapelfeldt W, Bjerke R, Labuda M, et al. Effect of hypothermia on the thrombelastogram. Anesthesiology. 1994;81:A192. Abstract.
12. Eldering S, Aragon G, Moore E. Fatal hepatic hemorrhage after trauma. Am J Surg. 1979;138:883-887.
14. Ellis P, Klunsasser L, Speer R. Changes in coagulation occurring in dogs during hypothermia and cardiac surgery. Surgery. 1957;41:198-206.
15. Mahajan S, Myers J, Baldini M. Disseminated intravascular coagulation following rewarming during hypothermia. JAMA. 1981;245:2517-2518.
16. Sincar P. Plasma volume, bleeding, and clotting time in hypothermic dogs. Proc Soc Exp Biol Med. 1954;87:192-196.
17. Villalobos T, Adelson E, Barilia T. Hematolytic changes in hypothermic dogs. Proc Soc Exp Biol Med. 1955;89:192-196.
18. White T, Krivit W. An ultrastructural basis for the shape changes induced in platelets by chilling blood. Blood. 1967;30:635-675.
19. Danzl D, Hedges J, Pozos R. Hypothermia outcome score: development and Implications. Crit Care Med. 1989;17:227-231.
20. Haemoscope Corporation. Thrombelastograph Coagulation Analyzer: The Global Coagulation Test for Functional Analysis of Haemostasis Using Whole Blood. Skokie, Ill: Haemoscope Corporation; 1991.
21. Mallett S, Cox D. Thrombelastography. Br J Anaesth. 1992;69:307-331.
22. Speiss B, Gillies B, Chandler W, Verrier E. Changes in transfusion therapy and re-exploration rate after institution of a blood management program in cardiac surgical patients. J Cardiothorac Vasc Anesth. 1995;9:168-173.
23. Attar S, Boyd D, Layne E, McLaughlin J, Mansberger A, Cowley R. Alterations in coagulation and fibrinolytic mechanisms in acute trauma. J Trauma. 1969;9:939-965.
24. Enderson B, Chen J, Robinson R, Maull K. Fibrinolysis in multisystem trauma patients. J Trauma. 1991;31:1240-1246.
25. Powers L. Diagnostic Hematology. St. Louis, Mo: CV Mosby Company; 1989.
26. Sacher R, McPherson R, Campos J. Widmann's Clinical Interpretation of Laboratory Tests. 10th ed. Philadelphia: FA Davis Company; 1991.
27. Wallach J. Interpretation of Diagnostic Tests. 5th ed. Boston: Little, Brown and Company; 1992.
28. Bergstein J, Slakey D, Wallace J, Gottleib M. Traumatic hypothermia is related to hypotension, not resuscitation. Ann Emerg Med. 1996;27:39-42.
29. Sori S, El-Assuooty A, Rush B, et al. The effect of temperature on survival in hemorrhagic shock. Am Surg. 1987;53:706.
30. Bernabei A, Levinson M, Bender J. The effects of hypothermia and injury severity on blood loss during trauma laparotomy. J Trauma. 1992;33:835-839.
31. Steineman S, Shackford S, Davis J. Implications of admission hypothermia in trauma patients. J Trauma. 1990;30:200-202.
32. Kearney T, Bentt L, Grode M, Lee S, Hiatt J, Shabot M. Coagulopathy and catecholamines in severe head injury. J Trauma. 1992;32:608-612.
33. Gando S, Tedo I, Kubota M. Posttrauma coagulation and fibrinolysis. Crit Care Med. 1992;20:594-600.
34. Olson J, Kaufman H, Moake J, et al. The incidence and significance of hemostatic abnormalities in patients with head injuries. Neurosurgery. 1989;24:825-831.
35. Engelman D, Gabram S, Allen L, Ens G, Jacobs L. Hypercoagulability following multiple trauma. World J Surg. 1996;20:5-10.
36. Kaufmann C, Dwyer K, Crews J, Dols S, Trask A. Usefulness of thromboelastography in assessment of trauma patient coagulation. J Trauma. 1997;42:716-720.
37. Stoner H. Studies of the mechanism of shock: the impairment of thermoregulation by trauma. Br J Exp Pathol. 1969;50:138-145.
38. West H, Jurkovich G, Donnell C, Luterman A. Immediate prediction of blood requirements in trauma victims. South Med J. 1989;82:186-189.
39. Gubler K, Gentilello L, Hassantash S, Maier R. The impact of hypothermia on dilutional coagulopathy. J Trauma. 1994;36:847-852.
40. Keatinge W. Survival in Cold Water. Oxford: Blackwell Press; 1969.
41. Leslie S, Toy P. Laboratory hemostatic abnormalities in massively transfused patients given red blood cells and crystalloid. Transfus Med. 1991;96:770-773.
42. Murray B. Severe lactic acidosis and hypothermia. West J Med. 1981;134:162-166.
43. Popovic V, Popovic R. Hypothermia in Biology, and Medicine. New York: Grune & Stratton; 1974.
44. Rutlidge R, Sheldon G, Collins M. Massive transfusion. Crit Care Clin. 1986;2:791-805.
45. West H, Jurkovich G, Donnell C, Luterman A. Immediate prediction of blood requirements in trauma victims. South Med J. 1989;82:186-189.
46. Von Reuden K, Dunham C. Sequelae of massive fluid resuscitation in trauma patients. Crit Care Nurs Clin North Am. 1994;6:463-472.
47. Bickell W, Wall M, Pepe P, et al. Immediate versus delayed fluid resuscitation for hypotensive patients with penetrating torso injuries. N Engl J Med. 1994;331:1105-1109.
48. Dennis J. Blood replacement, massive transfusion, and hemostasis in hemorrhagic shock. Trauma Q. 1992;8:2-68.
49. Harnett R, Pruit J, Sias F. A review of the literature concerning resuscitation from hypothermia. Part I. The problem and general approaches. Aviat Space Environ Med. 1983;54:425-434.
50. Silva R, Moore E, Galloway W. Reducing coagulopathy after trauma. Infect Surg. 1985;May:329-334.
51. Rutledge R, Sheldon G. Bleeding, and coagulation problems. In: Feliciano D, Moore E, Mattox K, eds. Trauma. 3rd ed. Stamford, Conn: Appelton & Lange; 1996.
52. Smith J, Bodai B, Hill A, Frey C. Prehospital stabilization of critically injured patients: a failed concept. J Trauma. 1985;25:65-70.
53. Johnson T, Chen Y, Reed R. Relative sensitivity of the clotting cascade to hypothermia. Surg Forum. 1989;40:199.
54. Kurrek M, Reed R. Effect of hypothermia on enzymatic activity of thrombin and plasmin. Surg Forum. 1987;38:1987.
55. Rohrer M, Natale B. Effect of hypothermia on the coagulation cascade. Crit Care Med. 1992;20:1402-1405.
56. Hesssel E, Schmer G, Dillard D. Platelet kinetics during deep hypothermia. J Surg Res. 1980;28:23-27.
57. Paul J, Cornillon B, Baguet J, et al. In vivo release of a heparin-like factor in dogs during profound hypothermia. J Cardiovasc Surg (Torino). 1981;89:519-527.
58. Brunker J, Goldstein R. Coagulation during hypothermia in man. Proc Soc Exp Biol Med. 1958;97:199-202.
59. Carden D, Novak R. Disseminated intravascular coagulation in hypothermia. JAMA. 1982;247:2099-3006.
60. Chadd M, Gray O. Hypothermia, and coagulation defects in the newborn. Arch Dis Child. 1972;47:819-821.
© Williams & Wilkins 1998. All Rights Reserved.
61. Tolman K, Cohen A. Accidental hypothermia. Lancet. 1961;2:1213-1219.