December 29, 2007
CLEVELAND, Dec. 27 — Less than a degree of hypothermia is enough to significantly increase blood loss during surgery (P<0.009), researchers here said.
A meta-analysis of published studies shows that the same degree of hypothermia is also enough to increase the need for transfusions significantly (P<0.027), according to Daniel Sessler, M.D., and colleagues at the Cleveland Clinic.
The findings clarify an area of controversy, since studies of the issue had given conflicting results, the researchers reported in the January issue of Anesthesiology.
“When all the studies were evaluated together, the results clearly show that even very mild hypothermia increases blood loss and transfusion requirements by clinically important amounts,” Dr. Sessler said.
The data imply that — except for special cases — patients should be maintained at a normal temperature during surgery, the researchers concluded.
In a literature search, the researchers narrowed 1,800 studies down to 18 that met initial criteria; four were excluded for reporting problems. The meta-analysis was thus based on 14 studies that met preset criteria for an analysis of blood loss and 10 that met criteria for analysis of transfusion.
Studies were excluded if they were retrospective, if the core temperature was reduced to less than 34° C, if local cooling methods had been used, or if the sample size was smaller than 15.
All told, there were 1,219 patients in the blood loss studies and 985 in the transfusion studies. The median of the mean temperatures reported for normothermic patients in the 14 blood loss studies was 36.6° C, compared to 35.6° C in the hypothermic patients.
The median temperature difference between normothermic and hypothermic groups in the studies was 0.85° C.
Normothermia was associated with significantly lower blood loss. The estimated ratio of geometric means of total blood loss was 0.84, favoring a normal temperature, with a 95% confidence interval from 0.74 to 0.96, which was significant at P<0.009.
Normothermia was also associated with a reduced need for transfusion. The overall estimated relative risk was 0.78, with a 95% confidence interval from 0.63 to 0.97, which was significant at P<0.027.
In other words, the researchers said, even mild hypothermia of less than 1° C increases blood loss by about 16% and the relative risk for transfusion by about 22%.
Dr. Sessler and colleagues noted that hypothermia in the evaluated studies was “of a magnitude that is typical for unwarmed surgical patients.” They added that therapeutic hypothermia would probably produce greater impairment of coagulation.
But Dr. Sessler said there are still situations where induced hypothermia is necessary.
“In occasional patients, mostly those at risk for brain injury, hypothermia is perfectly appropriate,” Dr. Sessler said. “In these patients, physicians need to trade off the potential benefits and risks and choose the optimal approach for each individual.”
The researchers cautioned that the study — like all meta-analyses — is subject to any flaws in the underlying studies. It may also be confounded, they said, by publication bias.
They also cautioned that there are several instances of meta-analyses being contradicted by large prospective randomized trials. In this case, however, it is unlikely that a prospective randomized trial could be conducted ethically, since keeping surgical patients at normal temperatures has become the standard of care.