WHILE an anesthesia resident at the University of California, Los Angeles—approximately 30 yr ago—I started thermoregulatory research under the guidance of the anesthesiologist and physiologist Eduardo Rubinstein, M.D., Ph.D. (Professor Emeritus of Anesthesiology). At that time hypothermia was considered a normal consequence of surgery, and not thought to be especially harmful. Few patients were warmed with anything other than relatively ineffective circulating-water mattresses; postoperative core temperatures after major abdominal surgery (all open in those days) was typically approximately 34.5°C.
Naturally, I started with a literature search. There were only approximately 30 articles about temperature in anesthesia at the time, including some about the problem of hyperthermia in unconditioned tropical operating rooms; approximately half were review articles—mostly by the same person. They all presented the same simple perspective: (1) anesthesia obliterates thermoregulatory defenses, making patients poikilothermic; (2) patients get cold during surgery because of excessive heat loss; and (3) postoperative reemergence of thermoregulatory control triggers shivering (“up to 400% increase in metabolic rate”!), which is the only serious complication of mild perioperative hypothermia. None of these conclusions were based on data.
Because I was a resident and did not have dedicated research time, I needed a project that was technically easy and could be done in the course of my routine clinical work. Eduardo told me to “go measure temperature; it’s easy.” Temperature monitoring is of course technically easy; but it turned out that the physiology of perioperative thermoregulation and heat balance was anything but simple.
We started our measurements in the recovery room (the term “post-anesthesia care unit” had yet to be invented). In those days, the American Board of Anesthesiology was less proscriptive about rotations, and the recovery room was hardly popular; I was thus able to spend 4 or 5 months collecting postoperative temperature and shivering data. The results were not encouraging: shivering was common enough, but inconsistently related to core temperature or thermoregulatory vasoconstriction. We finally concluded that the physiology described in review articles of the period was simply wrong. Roughly speaking, I spent the next 15 yr disproving nearly every concept and conclusion presented in these reviews.
Upon completion of my residency, the University of California at San Francisco hired me because they needed a pediatrician-anesthesiologist with a chemistry background to help start a magnetic resonance spectroscopy project. In fact, my chemistry background was trivial; I was a chemistry major at the University of California, Berkeley, but left for the Columbia University College of Physicians and Surgeons, New York, New York, after 3 yr. I think Ronald Miller, M.D. (Professor, Department of Anesthesia, University of California, San Francisco) confused me with my brother, Jonathan L. Sessler, Ph.D., who is a well-known chemist with whom I have since collaborated.1
I spent my first year in San Francisco doing magnetic resonance spectroscopy, but then returned to clinical thermoregulation. Some members of the department made no secret of the fact that they thought I was abandoning “real science.” However, George Gregory, M.D. (Professor, Department of Anesthesia, University of California, San Francisco) was encouraging, as was Henry Rosenberg, M.D. (Professor and Director of Medical Education, Department of Anesthesia, St. Barnabas Medical Center, Livingston, New Jersey) then at Hahnemann Hospital, Philadelphia, Pennsylvania. Professor Edmond Eger II, M.D., and Professor Emeritus John Severinghaus, M.D. (both from the Department of Anesthesia, University of California, San Francisco), from among others, were generous with their advice and mentorship. It would be dishonest to say that I had any idea at the time how the thermoregulation story would play out. In fact, at one point, I made a conscious decision to study thermoregulatory reflexes because they interested me even though I did not believe them to be clinically important . It seems I was wrong…
The first, and perhaps most real, thermoregulatory study (abstract above) was published in ANESTHESIOLOGY in 19882accompanied by an editorial by the great thermoregulatory physiologist Ted Hammel, Ph.D.3There are at least five interesting aspects of this study. Perhaps most importantly, it was the first study to show that anesthetized mammals are not poikilothermic; instead, humans who are given halothane do trigger thermoregulatory vasoconstriction in response to hypothermia, but not until core temperature reaches 34.4°± 0.2°C (normal approximately 36.5°C). In contrast, no patients randomized to normothermia vasoconstricted, demonstrating that vasoconstriction was a specific thermoregulatory defense, rather than resulting from hypovolemia or some irrelevant cause (fig. 1).
In subsequent studies, we quantified thermoregulatory vasoconstriction with nearly every volatile and intravenous anesthetic. Each drug has a unique dose–response curve, being linear for intravenous drugs,4–6but having disproportionate effect at high concentrations with volatile anesthetics.7–10Nonetheless, clinical doses of most drugs and drug combinations seem to trigger thermoregulatory vasoconstriction at core temperatures near 34.5°C (fig. 2).
The second interesting facet of our study was the demonstration that core temperature stabilized at the time of vasoconstriction. Thermoregulatory vasoconstriction is thus clinically important even during anesthesia, and prevents further hypothermia. How vasoconstriction constrains metabolic heat to the core thermal compartment is complicated and was subsequently quantified by Andrea Kurz et al .11(Kurz remains my longest-term and closest collaborator) using measurement techniques and models developed in collaboration with my physicist father, Andrew M. Sessler, Ph.D.12,13But suffice it to say that vasoconstriction is remarkably effective and explains why postoperative temperatures in unwarmed patients are rarely less than 34.5°C, no matter how large or long the operation might be.
In the absence of anesthesia, the thresholds (triggering core temperatures) for vasoconstriction and sweating are within a few tenths of a degree Celsius. In fact, the difference between these two thresholds defines normal body temperature, with small deviations in either direction triggering effective defenses. It is thus reasonable to use a “setpoint” model of thermoregulatory control in unanesthetized mammals, in which the thermoregulatory controller operates much like a home thermostat, being either fully “on” or fully “off.” But during anesthesia, the difference between these initial autonomic warm and cold defenses increases 10–20-fold depending on the type of anesthesia and its dose. A further complication is that even once triggered, each response has a gain (incremental increase per further temperature deviation)9,14–17and maximum response intensity,14–16which are also altered by general and neuraxial anesthetics. The third interesting aspect of our article is thus that it set the stage for a novel thermoregulatory model that accounts for individual and not-necessarily coordinated alterations in threshold, gain, and maximum response intensity. This model has become the standard for both clinical and physiologic studies.
The fourth interesting facet of our study is that it presented a novel methodology for evaluating vasoconstriction. Skin-temperature gradients, the difference between forearm and fingertip temperature, turns out to be a reliable measure of thermoregulatory vasoconstriction, which correlates extremely well with laser Doppler and the gold standard for flow, volume plethysmography.18
The basis for skin-temperature gradients is arterio-venous shunts in the fingers and toes, which are specialized high-capacity vessels that are designed to dissipate heat rather than provide local nutrition. (A given length of shunt conveys 10,000 times as much blood as a capillary.) When the central thermoregulatory controller opens shunts, the flow is so high that finger temperature increases to nearly core temperature. In contrast, finger flow is restricted to metatolic needs when the shunts are closed. Because fingers require little oxygen, finger temperature then decrease to nearly ambient temperature. The forearm is simply a skin-surface reference temperature that compensates for changes in ambient temperature. Because gradients are technically simple and resistant to artifact, they have since been used as a measure of thermoregulatory vasoconstriction (and finger flow, more generally) in hundreds of articles.
And finally, it is worth noting that our primary results are based on just 10 patients who were randomized to hypothermia (routine care at the time) or extra warming. Furthermore, the measurements were technically simple, consisting of just core temperature and two skin-surface temperatures. The study thus took only a couple of months to complete, and cost virtually nothing. But of course it was based on several years of (mostly) failed preliminary studies. So it is possible to make important advances with small, inexpensive, and technically simple studies—although considerable effort may be necessary to get to the point of knowing what to do.
This initial article on the vasoconstriction threshold during halothane anesthesia was the first of a few-hundred temperature-related studies. It led most immediately to the evaluation of dose-dependent effects of most anesthetics on the thresholds for sweating, vasoconstriction, and shivering—along with their effects on gain and maximum response intensity. The results were generally consistent with those for halothane: all anesthetics impair thermoregulatory control, but patients consistently respond to sufficient core and skin-temperature perturbations.4–10Many of these studies were conducted in volunteers, which permitted exquisite control of thermal conditions, and allowed us to use powerful crossover study designs that produced reliable results with relatively few participants (figs. 3and 4).
In the course of our threshold studies, we noticed that core temperature decreases precipitously during the initial hour of anesthesia, sometimes a full degree Celsius in just 30 min. The observation was not entirely new, but had always been attributed to heat loss from patients being undressed in a cool environment and to skin preparation with liquids that were allowed to evaporate. The difficulty is that to dissipate so much heat, mean skin temperature would have to increase to approximately 44°C—which clearly does not happen. That led to better understanding of thermal compartments and heat flow within the body. In fact, the initial hypothermia after induction of anesthesia is approximately 80% explained by core-to-peripheral redistribution of body heat rather than heat loss to the environment (fig. 5).12,19Subsequent studies demonstrated that internal flow of heat is a major determinant of core temperature under a wide variety of circumstances and is arguably as important as environmental heat balance.11,20,21A clinical consequence is that redistribution hypothermia can be largely prevented by prewarming patients before induction of anesthesia.22–24
A decade after starting to study thermoregulation, there was scant evidence that mild perioperative hypothermia caused any complications more serious than shivering and thermal discomfort. But there were increasingly compelling reasons to believe that it might. We thus conducted randomized trials evaluating the effects of hypothermia on surgical wound infections and coagulopathy, finding that each complication was markedly enhanced by just 1°–2°C of hypothermia.25,26A year later, in 1997, Frank et al. 27published the results of a trial showing that mild hypothermia causes morbid cardiac outcomes. Other studies followed in which we showed that mild hypothermia decreases drug metabolism28,29and prolongs recovery.30There are now more than 25 randomized trials demonstrating the adverse effects of mild perioperative hypothermia.
The clear causal link between hypothermia and serious outcomes involving numerous systems made the maintenance of normothermia a new standard-of-care. Maintaining normothermia is, in fact, now a “pay-for-performance” measure and routine in the United States. The need to maintain normothermia in turn led to dozens of studies evaluating heat transfer under various conditions and with various devices—work that continues. Other areas of interest included neuraxial anesthesia,31–33fever,34–37and blunting thermoregulatory responses to facilitate induction of therapeutic hypothermia.38–40Recognition that halothane impairs thermoregulatory defenses, but does not obliterate them, was thus the first step in understanding why and how surgical patients become hypothermic. That, in turn, led to an appreciation of just how harmful even mild hypothermia can be, and subsequently to good methods of preventing hypothermia.
For more than a decade after our initial thermoregulatory studies, we largely restricted our work to temperature-related investigations. However, we expanded to other topics when the Thermoregulation Group at the University of California in San Francisco became the Outcomes Research Institute at the University of Louisville, and subsequently the Department of Outcomes Research at the Cleveland Clinic. The Department of Outcomes Research—which has no clinical responsibility—has 40 full-time research personnel. The Department is the coordinating center for the international Outcomes Research Consortium,†the world’s largest clinical anesthesia research group. The Consortium coordinates approximately 100 studies at any given time, and publishes an average of a full paper per week—representing approximately 10% of the world’s clinical anesthesia research.41