Modelling hypothermia in patients undergoing surgery

Abstract

Anesthesia causes substantial perturbation in the human heat balance. Nearly all patients administered anesthesia become hypothermic. Under normal physiological conditions, the core-to-peripheral temperature gradient is maintained by tonic vasoconstriction. By the induction of anesthesia, vasoconstriction is impaired. Hence, heat redistribution takes place from the warm core to the colder periphery, leading to hypothermia. The heat balance during cardiac surgery differs from most other surgery types in that the body is also actively cooled by means of a heart lung machine to provide extra protection to the heart and the brain. A drawback of rewarming with help of the heart lung machine is that heat is transferred to the core compartment more quickly than to the peripheral tissues, leading to large core-to-periphery gradients. After decoupling the heat lung machine, internal redistribution of heat causes afterdrop: a decrease in temperature of the core. Afterdrop slows down the patient’s recuperation process. Therefore, more knowledge is needed about the impaired thermoregulatory system during anesthesia and the effect of different protocols on temperature distribution. This thesis focused on the development of a computer model that is able to describe heat transfer during anesthesia with the emphasis on cardiac surgery. The model that was developed consists of three parts: 1) a passive part, which gives a simplified description of the human geometry by means of a multi-segmental, multi-layered representation of the body, and that takes into account all passive heat transfer processes, 2) an active part that takes into account the thermoregulatory system as function of the amount of anesthesia and 3) submodels, through which it is possible to adjust the surgery and patient specific boundary conditions. Heat transfer in the passive part was modelled with help of the Pennes’ bioheat equation. This equation was solved using spatial and temporal discretization schemes. Boundary con168 Summary ditions were formulated to account for conductive, convective, radiative and respiratory heat losses. Specific submodels were designed to model the thermal influences of the heart lung machine, forced-air heaters, heating mattress and the heat loss through the wound. For the development of the thermoregulatory model, patient data was required. To that end, a clinical experiment was conducted. Two groups of aortic valve patients were studied: one group was rewarmed with and one group was rewarmed without using forced-air warmers. A significant reduction of afterdrop was observed in the group that was rewarmed with forced-air heating. The active model was derived combining a pharmacological model and the data of the aortic valve patients. The pharmacological model was used to calculate the propofol (the most often used anesthetic agent) concentration in the blood. Anesthetic drugs lower the threshold for vasoconstriction in linear proportion to increased plasma concentration. A relation was derived between the anesthesia concentration calculated with help of the pharmacological model and the vasoconstriction threshold found in the aortic valve patients. As a first approach, a stepwise response was used to model the gain and intensity of the vasoconstriction response. The model was validated by comparing temperatures predicted by the computer model to experimental data. A method was developed to refine the vasoconstriction relations of the thermoregulatory model. It was possible to determine the intensity of the centrally mediated sympathetic vasoconstrictor tone and the proportional distribution coefficients for vasoconstriction on different body parts. The method was used in a study protocol involving healthy volunteers for three body parts. In addition, detailed measurements were performed on volunteers to obtain proportionality values for the other body parts. The refined vasoconstriction model was added in the whole body thermal model. The complete model was validated against experimental data of healthy subjects and cardiac patients and showed in general good agreement. The validity of the model was tested for other types of surgery, i.e. orthopedic back surgery and deep hypothermic surgery with circulatory arrest. Finally, the model was used to study the effect of different temperature protocols like the use of forced-air heaters, increasing the environmental temperature, using heating mattresses or using a mild hypothermia protocol instead of a moderate hypothermia protocol. Overall, the model is able to predict temperature responses of healthy persons and patients undergoing surgery at temperatures between moderate hypothermia and normothermia, with skin temperatures ranging between 30 and 34oC. If the boundary conditions and initial conditions are accurately known, the model predicts core temperature within typically 0.5oC and skin temperature within typically 1oC.