Chapter 19: Fever and Temperature Modulation

Fever after brain injury independently worsens outcome. One of the important mechanisms is by exacerbating inflammatory cascades. Postinjury elevations in temperature have been shown to increase inflammatory processes, including elevations in proinflammatory cytokines, the increased accumulation of polymorphonuclear leukocytes in injured tissue.¹ The hypothalamus is a key center for thermoregulation and damage to this structure can result in hyperthermia. Vascular and inflammatory cascades appear to be extremely sensitive to mild elevations in temperature following CNS injury.

Several studies also suggest that one of the key impacts of fever is an increase in neuronal excitotoxicity. Elevations in temperature have been reported to increase neurotransmitter release, accelerate free-radical production, increase intracellular glutamate concentrations, and potentiate the sensitivity of neurons to excitotoxic injury.¹ In an experimental microdialysis study of focal ischemia, glutamate release was significantly higher in hyperthermic than normothermic rats, indicating the importance of focal brain temperature on neurotransmitter release.² Other investigators have observed increases in cellular depolarization in the ischemic penumbra surrounding damaged neuronal tissue, increased neural intracellular acidosis, and inhibition of enzymatic protein kinases, which are responsible for synaptic transmission and cytoskeletal function in relation to elevations in core body temperature.³ At the molecular level, elevations in temperature have been shown to enhance the expression of heat-shock proteins, as well as receptor expression associated with glutamate neurotransmission.⁴

Experimental and clinical studies have suggested that fever can also directly induce nervous system injury.⁵ In these circumstances, it is usually a temperature higher than 40°C for extended periods that produces abnormalities in the blood-brain barrier, as well as profound cardiovascular, metabolic, and hemodynamic dysfunction. Thus, mechanisms underlying hyperthermia-induced pathophysiology and secondary hyperthermia following central nervous system (CNS) injury are similar.

A recent meta-analysis of all brain injury types found that fever is related to morbidity and mortality.⁶ However, several important questions regarding the impact timing and duration of fever on outcome remain unanswered. Is it the fever seen within the first 24 hours or within the first 10 days the most influential on outcome? Which is more important in influencing outcome: the number of febrile episodes or the overall burden of temperatures above 37°C in the first week? Both clinical and experimental evidence indicate that each type of injury may have an optimal time window for fever control.

The new onset of fever should trigger a careful diagnostic evaluation to find a source of infection. It is important to be familiar with the patient's history, with particular attention being paid to predisposing causes of fever. For example, patients presenting in coma who require emergent endotracheal intubation are likely at a higher risk for developing pneumonia from aspiration. In addition to auscultation of the lung fields and careful abdominal examination, particular attention should be paid to any surgical wounds or skin ulcerations as well as for cerebrospinal fluid (CSF) leaks (otorrhea or rhinorrhea) in patients who underwent a craniotomy. Review of chest radiographs should focus on any evidence of new infiltrates or effusions. Initial laboratory tests should focus on peripheral white blood cells (WBCs) and cultures of blood, urine, sputum, and CSF in patients with an external ventricular or lumbar drain. If the patient is endotracheally intubated or has a tracheotomy, obtain a sample of sputum for Gram stain either by blind suctioning or bronchoalveolar lavage. Central venous catheters that have been in place for longer than 96 hours should be removed, and the tip should be submitted for semiquantitative microbiology. In patients receiving antibiotics for more than 3 days, a stool sample should be analyzed for the presence of Clostridium difficile toxin (Figure 19-1).

Pharmacologic Interventions

Endogenous pyrogens released by leukocytes in response to infection, drugs, blood products, or other stimuli cause fever by stimulating cerebral prostaglandin E synthesis and as a result raise the hypothalamic temperature set point.⁷ Antipyretic agents including acetaminophen, aspirin, and other nonsteroidal anti-inflammatory drugs (NSAIDs) are believed to block this process by inhibiting cyclooxygenase-mediated prostaglandin synthesis in the brain, resulting in a lowering of the hypothalamic set point. This activates the body's two principal mechanisms for heat dissipation: vasodilation and sweating.⁷ The effectiveness of antipyretic agents is tightly linked to conditions where thermoregulation is intact. Therefore, they are more likely to be ineffective in brain-injured patients with impaired thermoregulatory mechanisms. Corticosteroids also have antipyretic properties but are not used clinically to treat fever because of their side effects.

Whether acetaminophen alone is more effective than placebo for treating fever in adult ICU patients is still unclear. The majority of studies have been conducted in the pediatric population, where weight-adjusted doses have been shown to be effective in reducing fever. In the adult neurocritical care population, acetaminophen has been most widely studied in an attempt to maintain normothermia in patients with acute stroke. Koennecke and Leistner⁸ showed that treatment with acetaminophen in a daily dose of 4000 mg resulted in a substantial reduction of the proportion of patients with body temperatures over 37.5°C (the amount of temperature reduction was not reported). Kasner et al⁹ observed a difference of 0.2°C in body temperature in favor of treatment with acetaminophen (approximately 4 g per day) as compared to placebo in patients with hemorrhagic or ischemic stroke, although not statistically significant. Two recent phase II studies have demonstrated that perhaps a higher dose of acetaminophen (6000 mg per day) is more effective in maintaining normothermia/preventing fever. Based on the results from these studies, a large phase III study assessing the ability to maintain normothermia after acute ischemic stroke is under way.¹⁰

Ibuprofen has been widely studied in the pediatric population, with equivalent or superior efficacy as compared to acetaminophen. However, only one randomized, controlled study has been conducted in adult patients with brain injury, which demonstrated that ibuprofen (2400 mg per day) was not shown to be better than acetaminophen or placebo in maintaining normothermia after ischemic stroke.¹⁰ A recent small randomized study of a continuous infusion of diclofenac sodium (0.04 mg/kg per hour) was found to be effective in reducing the burden of fever and number of febrile events in critically injured traumatic brain injury (TBI) and subarachnoid hemorrhage (SAH) patients.¹¹ Although no adverse effects related to increasing hemorrhage rates were reported, larger studies are needed before this therapy can be widely utilized.

Nonpharmacologic Interventions

External cooling

External cooling reduces body temperature by promoting heat loss without affecting the hypothalamic set point. Four modes of heat transfer constitute the basis of interventions to promote heat loss: (1) evaporation (eg, water sprays or sponge baths); (2) conduction (eg, ice packs, water-circulating cooling blankets, immersion); (3) convection (eg, fans, air-circulating cooling blankets); and (4) radiation (ie, exposure of the skin).¹² In patients with temperature elevations caused by impaired thermoregulation, such as what occurs after brain injury, antipyretic agents are usually ineffective, and temperature reduction may only be achieved by external cooling. However, external cooling can result in reflex shivering and vasoconstriction as the body attempts to generate heat and counteract the cooling process.

Few controlled studies have evaluated the efficacy of external cooling interventions for lowering body temperature in humans. Previous experimental studies have shown that the combination of evaporative and convection cooling, with water sprays or sponging and forced airflow, is more effective than conduction cooling or either method alone for reducing temperature in non–brain-injured patients with hyperthermia.¹³ In a study of febrile neurocritical care patients treated with acetaminophen, air-blanket cooling had a small benefit that did not reach statistical significance.¹⁴

Water-circulating cooling blankets, a form of conductive cooling, are the most commonly used treatment for acetaminophen-refractory fever in critically ill adults. However, there are few data regarding their efficacy. Two small controlled studies have evaluated external cooling in adult ICU patients. One study compared the use of acetaminophen alone with tepid water sponging or with a water-circulating cooling blanket in febrile neurologic patients and found no difference between the three treatments.¹⁵ Another study of febrile patients under sedation, analgesia, and mechanical ventilation found that ice-water sponging was superior to two IV NSAIDs¹⁶ in a nonrandomized crossover study. A large observational study found no difference in the mean cooling rate in febrile ICU patients treated with or without water-circulating cooling blankets.¹⁷ A feature commonly found with water circulating blankets is the wide fluctuations in temperature that occur, with temperature overshoot being very common.

Recent engineering advancements have introduced a new set of surface-cooling blankets that are much more efficient at achieving and maintaining normothermia (Figure 19-2). Each system works by utilizing tightly wrapped pads that circulate cold water to promote conductive heat loss. In a randomized controlled trial of 53 neurologically injured patients that had a fever (temperature ≥ 38.3°C) for at least 2 hours after the administration of 650 mg of acetaminophen were randomized to treatment with an advanced cooling blanket system or a conventional water-circulating cooling blanket. Despite having a slightly higher baseline mean temperature, patients treated with the advanced system experienced a 75% reduction in fever burden and became normothermic faster than the control group, despite a significantly higher rate of shivering.¹⁸

Intravascular cooling

Over the past few years, a number of intravascular devices have become available to lower body temperature (Figure 19-2). They all work to control fever by directly lowering the blood temperature with cooled saline, which circulates through balloons or channels around an intravascular catheter. While there is no direct contact with the blood, the cold saline solution extracts heat from the blood, thus lowering body temperature.¹³

Infusion of 4°C normal saline is an appealing option because it is inexpensive and easy to administer in the critical care setting. The use of cold saline boluses has recently been studied and advocated for the induction of hypothermia in cardiac arrest patients.¹⁹ A small case series also found the rapid infusion of large-volume cold saline to be a safe and effective method to achieve normothermia in select brain-injured patients.²⁰ In addition to its rapid onset, the large volume of infusion can help offset the fluid imbalance that may be observed during the induction of hypothermia. Therefore, the administration of large-volume cold saline could be considered during the induction phase of fever control.

Clinical Evidence

Subarachnoid hemorrhage

Fever after SAH occurs in up to 70% of all patients in the first 10 days after SAH. In addition to disease severity and the amount of SAH present, the presence of intraventricular hemorrhage is a strong risk factor for the development of fever.²¹ Experimental models have demonstrated that the presence of even the smallest amount of blood within the CSF can induce fever.²²

Clinically, the febrile response has been implicated in the development of cerebrovascular vaso-spasm after SAH.²³ Numerous recent studies have found that after controlling for baseline predictors of poor outcome, fever is associated with an increased risk of death or severe disability, loss of independence in instrumental activities of daily living (IADL), and cognitive impairment.⁶ A recent case-control study that matched SAH patients by age, severity of injury, and amount of blood on initial computed tomographic (CT) scan found that the application of therapeutic normothermia for the first 2 weeks after hemorrhage was associated with an improvement in 12-month outcomes. This study also found a higher rate of pneumonia and longer ICU length of stay in the therapeutic normothermia group.²⁴ These preliminary findings point to the fine balance between infectious risks and benefit of prolonged fever control.

Cardiac arrest

Recent reports have demonstrated the benefit of therapeutic hypothermia for the first 12 to 24 hours after cardiac arrest,^25,26 and hypothermia has been adopted as the standard of care for survivors of cardiac arrest by the American Heart Association.²⁷ However, the ongoing importance of controlling temperature and preventing fever after the initial 24 hours has not been well studied. Previously published studies indicated that the occurrence of fever in the first 72 hours after cardiac arrest was very common and was independently associated with poor outcome.^28,29 It is not known whether there is still a high incidence of fever in the 48 hours after hypothermia, nor what the influence of applying hypothermia in the first 24 hours may have on the impact of subsequent fever. It may be reasonable to conclude, however, that ongoing maintenance of fever control (therapeutic normothermia) is beneficial for at least 48 hours after 24 hours of therapeutic hypothermia.

Spinal cord injury

Fever is a common complication in patients with spinal cord injury (SCI), with the most common cause being infections, especially pneumonia and urinary tract infections.¹ However, similar to brain-injured patients, this population also has a high incidence of thermoregulatory problems, deep venous thrombosis, and fever of unknown etiology.³⁰ Experimental studies have shown that inducing hyperthermia immediately after cord injury is associated with increased tissue damage and worse outcomes in behavioral and histopathologic measures as compared to normothermic conditions.¹ No clinical studies to date have looked at fever after spinal cord injury, limiting the support for aggressive maintenance of normothermia in this patient population beyond the initial resuscitation phase and into the ICU setting.

Stroke

Decades of experimental research have consistently pointed to the negative influence of fever at the time of ischemia. Maintaining normothermia acutely after ischemia stabilizes the blood-brain barrier, reduces cerebral metabolism, ischemic depolarizations, and free-radical production.³ Clinical studies have demonstrated the strong independent association between fever and stroke severity as well as outcome both at admission and in the first 24 hours after ischemic stroke⁶; however, the impact fever has beyond 24 hours has not been well studied. Mechanisms that influence secondary injury in ischemic stroke remain active for several days, and, clinically, the period for peak cerebral edema occurs within the first 3 to 5 days postinjury. Aggressive control of temperature during this time period may be warranted.

The clinical and experimental evidence for the impact of fever after intracerebral hemorrhage (ICH) is much more limited but comes to the same essential conclusion that fever in the acute stage after ICH is detrimental. Schwarz et al demonstrated that the development of fever within the first 72 hours after injury was associated with outcome on discharge, even after accounting for measures of injury severity.³¹ While clinical data are lacking with regard to the importance of fever beyond 72 hours, cerebral edema in ICH is known to persist for longer periods than seen in ischemic stroke and, therefore, these patients may require a longer period of fever control.

Traumatic brain injury

Fever early after TBI is associated with a poor Glasgow Coma Scale on presentation, the presence of diffuse axonal injury, cerebral edema on the initial head CT scan, systolic hypotension, hyperglycemia, and leukocytosis.³² Like all other brain injuries, fever after TBI is likely related to both the development of infection and degree of hypothalamic dysfunction from injury. Observational studies have found that the occurrence of fever in the first week after injury is associated with increased intracranial pressure (ICP), neurologic impairment, and prolonged ICU stay.³³ The largest cohort study is by Jiang et al, who reported a strong relationship between fever and outcome in a study of 846 TBI patients.³⁴ Taken together, these studies indicate that fever at any time in the first week after injury is associated with intermediate decline and long-term poor outcome and should be treated aggressively.

While most of the literature reports fever in relation to core body temperature, it is important to note that brain temperature is often higher than core body temperature. Rossi et al found that the incidence of temperature measurements exceeding 38°C in the brain was 15% higher as compared to simultaneously measured core body temperature from the pulmonary artery.³⁵ The difference between brain and core temperatures has been found to range up to 2°C depending on the characteristics of the patient, the probe placement, and interactions with other physiologic variables.^35,36 As patients become febrile, the gap between brain and core temperatures increases, which may indicate that the true incidence of febrile temperatures in the brain may be even higher than reported in large observational studies that have only measured core body temperature.

However, brain temperature monitoring requires specialized invasive intracranial monitoring, which is not indicated in the majority of brain-injured patients, and current practice is to target core body temperature. The most accurate measure of core body temperature is obtained from a pulmonary artery catheter, which faces the same limitation of the invasive intracranial monitoring. As a result, the two most common modes of obtaining a continuous core temperature reading is via either a bladder temperature probe or an esophageal temperature probe. A bladder temperature probe has the advantage of being attached to a Foley catheter and, therefore, can serve dual purposes. Accurate bladder temperatures rely on normal urinary flow such that both polyuria and oliguria can result in wide fluctuations in temperature readings and render a bladder monitor inaccurate. This is a significant and common limitation that should be considered when using bladder temperatures during normothermia or hypothermia therapy. Esophageal temperature monitors are often isolated probes that serve only to measure temperature, although there are some probes available that are intertwined into an existing nasogastric tube. The great advantage of esophageal temperature as the measure of body temperature is its close correspondence to temperature in the pulmonary artery. However, some means of ensuring that the temperature probe is placed in the lower esophagus is necessary (eg, verification by chest radiography) since catheters may tend to form into a "U" shape in swallowing so that the tip is high in the esophagus even though a sufficient length of catheter has been swallowed.

Fever is not only a cardinal sign of infection, it is also an adaptive response that enhances the ability to fight infection, and by inducing normothermia, this adaptive response may be impaired. A prospective, randomized clinical trial demonstrated that treatment of septic patients with intravenous ibuprofen for 48 hours improved physiologic end points but did not decrease the incidence of organ failure or 30-day mortality.³⁷ These studies suggest that the febrile response may have beneficial effects. This should always be weighed against the lack of evidence for the routine use of therapeutic normothermia in brain-injured patients.

Should the risk of ongoing secondary injury of fever be deemed to be greater than the risk of eliminating fever, routine infection surveillance is important. Unfortunately, no standard approach on the best way to do this surveillance currently exists. One large randomized controlled study of therapeutic normothermia using an intravascular cooling device obtained cultures and a chest radiograph if the WBC count rose by 20% or temperature exceeded 38°C. Others have advocated obtaining cultures when the water temperature in water-circulating blankets dips below 10°C even though this may not be directly related to the development of infection. The inability to accurately detect infection during therapeutic normothermia can result in antibiotic misuse and selection of resistant organisms.

More extensive diagnostic evaluation should be considered in a graded fashion based on history, physical examination findings, laboratory results, persistence of fever despite presumably appropriate antimicrobial chemotherapy, or clinical instability. These additional tests and procedures include diagnostic thoracentesis, paracentesis, and lumbar puncture. Imaging studies should be considered, including abdominal or cardiac ultrasonography and head, chest, or abdominal CT.

Although fever in the ICU is most commonly due to infection, myriad noninfectious causes of systemic inflammation can also result in hyperthermia. Important noninfectious causes of fever in ICU patients are listed in Table 19-1. Although it is believed that fever from noninfectious causes rarely results in a high core temperature, data in support of this view are lacking. As well, infections are rarely, if ever, associated with core temperatures greater than 40°C. When the core temperature is this high, the clinician should also suspect malignant hyperthermia or neuroleptic malignant syndrome.

The shivering/vasoconstriction response is dependent on a temperature set point mediated via the preoptic nucleus of the anterior hypothalamus. The thermoregulatory system utilizes a series of positive- and negative-feedback loops to minimize fluctuations, maintaining core body temperature within 0.1°C to 0.2°C. The overall goal of such tight control is to reduce oxygen utilization and caloric expenditure to maximize metabolic efficiency as well as protect crucial enzymatic function.

In normal conditions, the hypothalamus coordinates a response to maintain core body temperature at normothermia (37°C), and the shivering/vasoconstriction response begins once temperatures fall below 36°C. In brain-injured patients, the set point is believed to be elevated, and as a result, the thermoregulatory response can be seen when lowering body temperatures to normothermic levels. In fact, the incidence of shivering has been reported to occur in up to 40% of patients undergoing therapeutic normothermia.

The metabolic consequences of this response can be extensive. An important and consistent consequence is a dramatic increase in resting energy expenditure (REE), carbon dioxide production (Vco₂), and oxygen consumption (Vo₂). By affecting several muscular groups for prolonged periods, shivering triggers an increase in metabolic demand, which translates into higher Vo₂ combined with increased respiration.³⁸ The clinical consequences include increased tissue ischemia, which has been associated with increased morbidity rate in postoperative cardiac surgery patients.³⁹ The increase of Vo₂ linked to shivering is proportional to the affected muscular mass and can increase the basal rate of oxygen consumption by two- to threefold.³⁸ Factors associated with an increased response include young age, higher muscle mass, and low serum magnesium.⁴⁰ The increased metabolic demand of uncontrolled shivering can lead to a dramatic change in the utilization of carbohydrates, lipids, and protein, promoting catabolism in critically ill patients.⁴¹ For all these reasons, uncontrolled shivering can eliminate any benefit of fever control, making combating and preventing shivering crucial when inducing and maintaining normothermia.

The ability to differentiate the graded metabolic response to shivering is important during therapeutic normothermia, particularly as an end point for antishiver interventions. Previous measures of shivering in the postoperative setting provided qualitative assessments that may not have translated to the brain-injured patient.⁴² More recently, a simple grading scale, the Bedside Shivering Assessment Scale (BSAS),⁴³ was developed by assessing the correlation of bedside shivering assessments with systemic metabolic stress quantified by indirect calorimetry (Table 19-2).

By clinically assessing the muscular involvement in the trunk and limbs, the BSAS provides an accurate representation of the metabolic impact of shivering. The ability to accurately identify the intensity of shivering has distinct advantages when developing a rationale approach toward treating shivering without oversedation.

When adopting a stepwise approach toward shiver control that coincides with the initiation of therapeutic normothermia, less sedating options (eg, surface counterwarming) are preferred as the initial interventions (Table 19-3). Cheng et al⁴⁴ have shown that a linear relationship exists between core temperature and the average skin temperature for the appearance of shivering in the nonanesthetized patient. The threshold temperature for shivering is equal to the sum of 20% of the mean skin temperature and 80% of the core temperature. Therefore, to inhibit shivering, the average skin temperature must be raised by at least 4°C to be as efficient as a 1°C increase in core temperature. Radiant heat systems first applied in the recovery room proved to be an efficient way of preventing postanaesthetic shivering⁴⁵ or rapidly inhibiting it when it occurs. A recent prospective study demonstrated that the application of a forced-air warming blanket set at 43°C can effectively limit the metabolic impact of shivering in brain-injured patients. Surface counterwarming is a safe, effective, cheap, and nonsedating antishivering intervention that should be applied in all patients.

Pharmacologic interventions without significant sedating effects include buspirone and magnesium. Buspirone is a serotonin 1A (5-HT_1A) partial agonist that has shown special antishivering activity by activating hypothalamic heat-loss mechanisms. It is only mildly sedating and provides a good synergistic effect when combined with other antishivering interventions. The main disadvantage of buspirone is that it is administered orally and, therefore, may not be reliably absorbed in critically ill patients.⁴⁶ The intravenous administration of magnesium promotes both cutaneous vasodilation and muscle mild relaxation. Magnesium may also confer some protection from tissue ischemia, although recent studies of its neuroprotective properties have not been conclusive. Regardless, low serum magnesium levels have been shown to be a risk factor in the development of shivering, and all efforts should be made to maintain serum levels between 3 and 4 mg/dL. Levels higher than this may be associated with depressed sensorium and respiratory effort.

When these initial measures are not effective, pharmacologic agents that are more effective, but also more sedating, medications are utilized. Dexmedetomidine is a centrally acting α agonist that has distinct advantages of being an infusion with a short half-life. In tests with healthy volunteers, it is very effective in lowering the shivering threshold, and it has a synergistic impact when combined with buspirone.⁴⁶ The main limitations of the use of dexmedetomidine are bradycardia and hypotension which can be encountered at the higher doses necessary to achieve shiver control.

The most effective antishivering pharmacologic agent is meperidine.⁴⁶ It is the only opiate to have special "antishiver" properties owing to its κ-receptor activity (in addition to μ-receptor activity). It may also have central α₂-agonist activity. Given the short half-life of its antishivering properties, multiple dosing is required to achieve sustained shiver control, which often results in prolonged sedation. Lowering of the seizure threshold, especially in the setting of renal insufficiency, also limits its long-term utilization.

There is currently no clinical evidence for the duration that fever impacts outcome after each form of brain injury. Without such information it is impossible to understand what the optimal duration should be for therapeutic normothermia. Most studies have observed that the fever burden in the first week after injury is important in determining outcome. SAH studies have extended the duration of importance out to the first 14 days after injury. Therefore, it would be reasonable to maintain normothermia for at least the first week after injury (2 weeks for SAH patients). Additional issues, such as the impact of ongoing fever on ICP, may be factored into a continued need for fever control beyond this time frame. As well, a decline in mental status is occasionally observed with the rebound fever that occurs with discontinuation of a temperature-modulating device. It is reasonable to reinstitute normothermia and attempt discontinuation at a later time. The phenomenon of rebound fever with discontinuation of a device is nearly universal and may have to do with the impact of these powerful devices on the hypothalamic thermoregulatory feedback loops, although the exact mechanisms are not known.