Chapter 34. Approach to Fever in Critical Care

Introduction

Fever is an adaptation mechanism of the body in response to internal and external environmental stressors, and is a key indicator of immune system activation. Normal body temperature is maintained by peripheral nerves that transmit signals back to the hypothalamus. Fever occurs when cytokines cause an increase in body temperature in association with a rise in hypothalamic set point, and consists of three clinical phases: chill, fever, and flush. Elevated body temperatures can broadly be classified as hyperthermia syndromes and infectious and noninfectious fever (see Table 34-1). Hyperthermia occurs when thermoregulatory mechanisms fail and when heat production exceeds heat loss through either overproduction of heat or decrease in heat loss. Such examples of heat overproduction include thyrotoxicosis, pheochromocytoma, adrenal crisis, or salicylate toxicity via interruption of the citric acid cycle and uncoupled oxidative phosphorylation. Heatstroke or anticholinergic toxicity is mediated by a deficiency in mechanisms of heat dissipation. Some hyperthermia syndromes fall into both categories, such as postanesthesia neuroleptic malignant syndrome (NMS), and may cause profound hyperpyrexia. It is important to differentiate fever from hyperthermia because hyperthermia due to thermoregulatory failure is treated by a lowering of the body temperature by physical mechanisms (conduction, convection, evaporation); antipyretics are not effective. Both noninfectious and infectious causes of elevated body temperatures in the intensive care unit (ICU) will be discussed in detail in following sections.

Definition

The average body temperature set point is 37.0°C (98.6°F) and may vary by 0.5–1.0°C according to time of day or hormonal milieu: it is highest at 6:00 am and for women at the time of ovulation. Fever is defined in multiple ways. It is an isolated core body temperature >38.0°C (100.4°F) or two consecutive elevations of greater than 38.3°C. In neutropenic patients, fever may be defined as a single temperature greater than 38.3°F (101.0°F) or greater than 38.0°C (100.4°F) for 60 minutes. The American College of Critical Care Medicine and the Infectious Diseases Society of America (ACCM/IDSA) define a fever as a rise in body temperature greater than 38.3°C (101°F) and recommend any new fever be investigated.1 However, in immunocompromised or elderly individuals, a lower cutoff may be appropriate, as these patients may not be able to mount substantial febrile responses. Additionally, the fever response may be attenuated in patients with azotemia or congestive heart failure, or in patients receiving antipyretics or pain control with an antipyretic combination.

Epidemiology

Fever is common in critically ill patients and warrants clinical attention. Infections are the leading cause of temperature elevation in hospitalized patients, whereas hypothalamic disorders are less common. In patients admitted to an ICU with severe sepsis, the incidence of fever approached 90%.2 Prospective and retrospective studies have described wide ranges of fever prevalence in the ICU, ranging from 30% to 70%, with the highest incidence seen in noncardiac surgical patients in one study.3,4 In the neuro-ICU, the incidence of fever may approach 70%, only half of the fevers being due to infection, mainly nosocomial pulmonary infection.5 The incidence of hyperpyrexia due to anesthesia-induced malignant hyperthermia (MH) estimates ranges from 1:250 to 1:250,000, with a recent study evaluating 2001–2005 New York state discharge data establishing local incidence as 1 in 100,000.6

Pathophysiology

Basal metabolic activity in the liver and heart accounts for the majority of the body's heat production, while the skin accounts for the majority of heat dissipation. The lungs add a minor amount to basal metabolic heat dissipation through conduction and evaporation. Temperature is regulated not by one single neural area, but by feedback loops involving the hypothalamus, brainstem, and spinal cord that interact with the autonomic, somatic, and endocrine systems (see Figure 34-1). The earliest published observations regarding temperature regulation appeared in 1912 and described thermal sensitivity of the hypothalamic region.7 Further understanding was provided in the 1960s with three influential papers describing the role of the preoptic area of the anterior hypothalamus in thermoregulation.8–10 Stimulation by the anterior hypothalamus causes vasoconstriction and sweating, while posterior hypothalamic activation induces shivering. Vasoconstriction in response to an increased hypothalamic set point begins in the hands and feet as blood is shunted centrally. Shivering develops as a heat conservation mechanism to increase heat production from skeletal muscle. The nature of the response depends on ambient temperature; animal models injected with exogenous substances that raise temperature set point will increase heat production in a cold environment or decrease heat loss in a warm environment.

Fever is regulated at the level of the hypothalamus through pyrogen release by activated immune cells. Exogenous pyrogens, such as lipopolysaccharide (LPS) endotoxin in gram-negative bacteria or exotoxins such as toxic shock syndrome toxin (TSST) in Staphylococcus aureus, trigger febrile responses in the host. LPS complexes with a binding protein and attaches to the CD14 receptor of a macrophage resulting in cytokine release.11 Cytokines are soluble intracellular signal proteins that regulate local and systemic immune processes. Formulated as large-molecular-weight polypeptides, they are produced by monocytes, macrophages, and glial cells in response to inflammation, infection, or injury.12,13 Interleukin (IL) 1 and tumor necrosis factor (TNF) are structurally unrelated cytokines with a strikingly similar biologic function; both are secreted by antigen-presenting cells that augment binding and activation of T cells and promote growth and differentiation of B cells. TNF-α is produced by activated macrophages in response to LPS of gram-negative organisms, whereas TNF-β is a product of T lymphocytes. Together, IL-1, IL-6, and TNF are collectively referred to as proinflammatory cytokines (see Table 34-2).

Endogenous pyrogen, later reclassified as lymphocyte-activating factor and eventually found to be part of the IL-1 family, was the earliest isolated cellular product implicated in fever induction.14 Animal models of endogenous pyrogens indicate the febrile response is mediated by activation of calcium channels and can be attenuated by calcium channel blockers such as nifedipine or verapamil.15 The IL-1 gene family composed of IL-1α, IL-1β, and IL-1 receptor antagonist (IL-1ra) is encoded on the long arm of chromosome 2. Variable number tandem repeat polymorphisms in this receptor antagonist region are linked to autoimmune dysregulation syndromes such as psoriasis and inflammatory bowel disease.16 IL-1β is a potent inducer of IL-6 that is critical in the fever response as evidenced by the absence of fever production in IL-6-deficient knockout mice.17,18 Cytokines bind to and activate their own receptor, activating phospholipase A2 resulting in release of arachidonic acid, which is the substrate of cyclooxygenase and the rate-limiting enzyme in prostaglandin biosynthesis.19

Increased levels of IL-1, IL-6, interferon-γ, and TNF-α act on the hypothalamus to raise its inherent set point through the catecholamine cells of the medulla oblongata and the circumventricular organs.20 The organum vasculosum of the lamina terminalis (OVLT) is a vascular sensory organ of the brain that is unique in that it is composed of a capillary bed that lacks a blood–brain barrier and can therefore monitor the osmotic, ionic, and hormonal environment of the blood.21 When pyrogens are detected by the OVLT, prostaglandin E2 (PGE2) is released that triggers the PGE2 receptor on glial cells to release cyclic adenosine monophosphate (C-AMP). This activates the febrile response via the hypothalamic feedback loops involving vasoactive substances and neurotransmitters such as norepinepherine, dopamine, and serotonin.22 Additionally, these cytokines are also released during tissue trauma, especially IL-6.23

Endogenous antipyretics such as IL-10, a protein product of T-helper cells, have been shown in mouse models to inhibit the production of endogenous IL-1β, IL-6, and TNF during LPS-induced fevers.18,24 Additionally, arginine vasopressin, α-melanocyte-stimulating hormone, and glucocorticoids counter and limit the duration of fever.25

Temperature Measurement

In the ICU, body temperature may be measured peripherally or centrally. Temperature is most accurate when measured by thermistors on pulmonary artery catheters (the gold standard), bladder catheters, or esophageal probes. Rectal probes provide a close approximation of core temperature. Generally readings from rectal thermometers are a few tenths of a degree higher than core temperatures. Rectal thermometers are somewhat invasive for the awake and alert patients and are contraindicated in neutropenic patients. Oral temperature measurements are convenient, safe, and minimally invasive, although readings may be confounded by ingestion of hot or cold fluids, or mouth breathing. Additionally patients with decreased level of consciousness, arousal states, or altered mental status may not be able to comply with placement of a thermometer under the tongue. Readings may vary depending on the location of sublingual thermometer placement, and are generally 0.4°C (0.7°F) lower than rectal temperatures.26 Infrared tympanic membrane thermometers are less accurate than intravascular probes and rectal or oral thermometers. Temporal artery thermometers and axillary or femoral skinfold temperatures should not be used to record temperatures in an ICU.27

Effects on the Host

Conflicting evidence exists as to outcome associations with elevated temperature in ICU patients. One study suggested that crude mortality was higher in febrile patients (34.5% vs. 18.7%); however, when adjusted for patient severity, fever was no longer associated with mortality (P = 0.384).28 Low-grade fever was common in the ICU, and outcome varied by admission criteria.4 High fever was associated with an increased risk of death (20% vs. 12%) in a large literature review.

Numerous studies have found that after controlling for baseline predictors of poor outcome, fever in acute subarachnoid hemorrhage is independently associated with increased morbidity, including cognitive impairment, and mortality.29,30 In stroke patients, the earlier the fever onset, the greater the amount of cognitive dysfunction; hyperthermia appearing after 24 hours was not associated with poorer outcome.31

Controversy exists as to whether treating a fever is beneficial for the host. Elevated temperatures result in tachycardia, increased minute ventilation, resting energy expenditure, oxygen consumption, and sympathetic tone. Hyperthermia has been associated with rhabdomyolysis, disseminated intravascular coagulation, and multisystem organ failure.32 In animal models, fever has been shown to decrease the serum level of iron, which is a growth factor for many microbes.33 It may reduce the expression of virulence factors, enhance antibiotic susceptibility by lowering minimum inhibitory concentrations, and augment host responses.34,35 A recent study in a trauma ICU that randomized patients to permissive or aggressive treatment (650 mg acetaminophen q 6 hours) for fevers >38.5°C was stopped after the first interim analysis due to a statistically significant increase in deaths in the aggressively treated group (P = 0.06).36 Despite the lack of evidence-based outcomes data, pharmacologic and physical means of fever reduction are commonly employed.

Noninfectious Causes of Fever

Drug Fever

Drug fever is often a diagnosis of exclusion based on the background of administration of a new medication. A hallmark sign of drug fever is the disappearance of fever after the medication is stopped and reappearance after it is restarted. This is most often the result of a hypersensitivity reaction, and commonly occurs 7–10 days after the administration of an agent; it may be accompanied by rash, urticaria, or serum sickness. Although any medication can cause hypersensitivity reaction, antimicrobials (especially β-lactams), antimycobacterials, antiepileptics, antiarrhythmics (such as quinidine and procainamide), and antihypertensives (methyldopa and phenytoin) are common causes of fever.37 Certain classes of pharmacologic agents are associated with hyperthermia through disordered thermoregulatory mechanism. Sympathomimetics, anticholinergics, neurotransmitter active drugs, such as dopamine antagonists, serotonergic agents, and monoamine oxidase inhibitors, as well as inhaled anesthetics can disrupt the balance between heat production and dissipation. MH occurs in genetically predisposed individuals after exposure to certain pharmacologic agents. It is the result of a large efflux of calcium triggered by inhalation anesthetics or succinylcholine on a background of a genetic defect in the ryanodine, or calcium release channel in the sarcoplasmic reticulum of skeletal muscle.38 NMS can be seen with antipsychotic agents such as haloperidol, prochlorperazine, and metoclopramide or with the withdrawal of dopaminergic agents. It is characterized by muscular rigidity, autonomic dysregulation, extrapyramidal side effects, and hyperthermia, and is thought to be due to dopamine antagonism within the hypothalamus.39 Serotonin syndrome has similar clinical features, but also involves diarrhea, tremor, and myoclonus. It is related to excessive 5HT1A receptor stimulation, and may be exacerbated with the use of linezolid.40 Illicit drugs such as phencyclidine (PCP), ecstasy (methylenedioxymethamphetamine [MDMA]), lysergic acid diamide (LSD), and cocaine have been implicated in hyperthermia syndromes. MDMA ingestion leads to central deregulation of thermogenesis via activation of the sympathetic nervous system and an excessive release of norepinepherine with an uncoupling of adrenoreceptors and loss of heat dissipation.41

Head Injury

Elevated core temperature is common after many types of neurologic injury (ischemic, hemorrhagic, or traumatic) and is associated with increased risk of adverse outcome, even after controlling for confounders or modifiers, such severity of illness, diagnosis, age, and infection. Human and animal models have shown that fever exacerbates ischemic neuronal damage and is proportional to the degree of pyrexia.42 Hyperthermia postcardiac resuscitation was associated with an unfavorable neurologic recovery after cardiac pulmonary resuscitation.43 Fever was found to be strongly associated with an increased intensive care and overall length of stay, and higher overall mortality.44,45

Heatstroke

Heatstroke occurring in warm environments may be exertional or nonexertional in origin, and may be exacerbated by dehydration or antihistamines. Defined as a core temperature 40°C (>104°F), individuals at the extremes of age are at risk for nonexertional heatstroke during hot weather and heat waves. It is associated with upregulation of heat shock proteins in the brain that function as molecular chaperons and cellular repair proteins with cytoprotective effects.46

Neurologic Causes of Fever

Although fever may occur in up to one quarter of neuro-ICU patients, almost half are noninfectious.47 Stroke or subarachnoid hemorrhage can trigger febrile responses in noninfected patients, as can head trauma and neurosurgery involving the floor of the third ventricle.48

Miscellaneous

Vasculitis, hyperthyroidism, or mesenteric ischemia can trigger febrile responses in noninfected patients. A low-grade fever can also be seen in the cardiac care unit post–myocardial infarction, resulting from epicardial inflammation after a transmural infarct. Dressler's syndrome, likely mediated by antimyocardial antibodies, can also present with fever and a friction rub up to 2–3 months post–myocardial infarction.

Blood is an irritant, and when it accumulates or stagnates, it may induce fever. Hematomas and pulmonary embolism have been associated with fever.49 However, contrary to common dictum, deep vein thrombosis is not a common cause of isolated fevers as evidenced by a number of recent studies evaluating the rate of fever in patients with lower extremity deep venous thrombosis.50,51 Transfusion reactions are possible during or after receipt of blood products.

Noninfectious intra-abdominal processes such as pancreatitis, acalculous cholecystitis, and mesenteric ischemia are causes of fever in critically ill patients, and these entities frequently present with associated clinical signs and symptoms. Rheumatologic disorders such as systemic lupus erythematosus, and adult Still's disease, as well as occult malignancy, are uncommon, albeit possible, causes of fever in ICU patients.52

Infectious Causes of Fever

Central Nervous System Infection

Focal neurologic abnormalities generally occur with central nervous system (CNS) infection; however, in critically ill patients, a high index of suspicion is warranted even in the absence of focal findings, and appropriate imaging studies and culture data should be obtained.53 Fever is the most common acute presentation of bacterial meningitis in children; in adults and the elderly, confusion, nuchal rigidity, and headache are more common.48 Bacterial meningitis may occur after any neurosurgical procedure, but is most common with procedures related to open head trauma.54

Diarrhea

For evaluation of fever in the ICU, ACCM/IDSA defines diarrhea as more than two stools per day that conform to the container in which they are placed.53 Enteral feedings and medications are common causes of loose stool or diarrhea in the ICU patient. The most common enteric cause of fever in the ICU is Clostridium difficile and should be suspected in any patient with fever, elevated white blood cell count, and antimicrobial therapy or chemotherapy administration within 60 days of diarrhea onset. Other organisms that cause fever and diarrhea are generally community associated and rarely acquired after a patient is admitted to an ICU. Therefore, sending stools for routine culture or ova and parasites should be avoided unless the patient was admitted to the hospital with diarrhea or is human immunodeficiency virus (HIV) positive, or as part of an outbreak investigation.53 In patients with negative C. difficile toxin assays, it is important to consider increased gastrointestinal (GI) motility as a side effect of medications, enteral feedings, or hemorrhagic enterocolitis due to Klebsiella oxytoca.55,56

Intravascular Devices

Patients should be examined daily for signs of catheter entry site infections and phlebitis, and any expressed purulence should be sent for Gram stain and culture. Short-term peripheral and noncuffed central catheters should be removed if infection is suspected; with evidence of a tunnel infection or septic physiology, the catheter should be removed, cultured, and reinserted at a different site.53 It is not necessary to routinely culture all catheters removed from ICU patients, as catheters are frequently colonized within the lumen and this may not correlate with infection.53

Pneumonia

Pneumonia is a common cause of infection acquired in the ICU and a predominant cause of fever, especially in mechanically ventilated patients. For initial fever evaluation, a portable chest radiograph is sufficient. In the nonintubated patient, expectorated sputum or nasal/endotracheal aspirate is adequate to evaluate airway colonization or infection.53 Aspirates from the inner channel of the bronchoscope in intubated patients reflect upper airway colonization and may lead to overtreatment of colonizing organisms. Mini-bronchoalveolar lavage (BAL) or blind bronchoscopy, with a protected brush, is a reliable sampling method to obtain lower respiratory secretions.57 Respiratory cultures should be processed within 2 hours of collection.

Postoperative Fever

Fever is a common phenomenon in the first 48 hours after surgery. Initially, the etiology is noninfectious, but after 96 hours fevers can often be attributed to infectious processes.58 Wound infections are rare immediately postop, with the exception of S. pyogenes or clostridial infections that may present in the first 3 postoperative days. In the febrile postoperative patient, surgical sites should be examined daily for erythema, purulence, or tenderness, and incisions should be opened and cultured if infection is expected.53 New or persistent fevers after 96 hours warrant careful surgical site inspection as well as investigation into other etiologies of fever, including thromboembolic disease, drug reaction, MH, or catheter-related infection.

Sinusitis

Nosocomial maxillary sinusitis is a common entity in intubated patients and should be included in the differential diagnosis of fever in an ICU patient.59 Either two major criteria (cough, purulent nasal discharge) or one major plus two minor criteria (headache, earache, facial or tooth pain, malodorous breath, sore throat, or wheezing) suggest acute bacterial sinusitis in the outpatient setting; however, in critically ill patients, these signs may not be evident.60 Additionally, sinus films may be of limited value and sinus computed tomography (CT) or magnetic resonance imaging (MRI) scans may be difficult to obtain. For a definitive diagnosis, puncture and sampling of the involved sinus under aseptic technique should be performed.53 A prospective study of new-onset fever in surgical ICU patients after excluding bacteremia, catheter-related infections, or pneumonia found sinusitis diagnosed by three-view sinus films accounted for 24% of fevers, and the predominant microbiology was Klebsiella and Pseudomonas.61 Another study found the common pathogens by maxillary sinus aspirates were Acinetobacter (32%) and anaerobes (21%) and a combination of a nasal decongestant and topical nasal steroid was effective in decreasing the incidence of sinusitis in mechanically ventilated trauma patients.62

Urinary Tract Infection

Urinary tract infection is among the most frequent nosocomial infection in the ICU, and a common cause of fevers due to frequent bladder instrumentation. Not surprisingly, increased duration of catheter days is correlated with the risk of cystitis and pyelonephritis.63 The predominant pathogens involved in urinary tract infections in ICU patients include multidrug-resistant gram-negative rods. Cultures should be collected from the sampling port of the catheter and not the drainage bag, and processed by the microbiology laboratory within 1 hour. A colony count from a catheterized patient of >103 cfu/mL represents true infection.53

Immunocompromised Patients

Immunocompromised patients (i.e., HIV/acquired immune deficiency syndrome [AIDS], induced immune suppression from solid organ or bone marrow transplants, chemotherapy, or immune modulation therapy) are at risk for opportunistic bacterial, viral, and fungal infections. Special consideration must be given to immunodeficient patients with fevers. A broad range of infectious organisms can be seen, including cytomegalovirus, Pneumocystis jerovicii, Aspergillus, and endemic mycoses such as Histoplasma and Coccidioides.

Diagnostic Approach

Approach to the febrile patient begins with a proper diagnosis and management of the underlying disorder. Pseudosepsis, characterized by fever, elevated leukocyte counts with a left shift, and sepsis physiology with elevated heart rate and hypotension, can closely mimic infectious fevers, but may be attributed to rheumatologic, endocrine, or neurologic disturbances. Adrenal insufficiency and thyroid storm have been mistaken for sepsis.64 The magnitude of temperature elevation does not provide clues to the etiology, as fevers greater than 102°F may be present with both infectious and noninfectious causes.

Accurate history and physical exam, along with careful review of the hospital course, including previous in-patient or outpatient workup, is the first step in diagnosis and management of the febrile patient in the ICU. The 2008 ACCM/IDSA guidelines recommend that a new onset of temperature <36.0°C or >38.3°C warrants a clinical assessment.53 A thorough physical exam should be performed, including conjunctival and fundoscopic ocular exam, detailed oropharyngeal inspection, careful auscultative cardiopulmonary exam, and full skin exam including catheter insertion sites and dependent or posterior body surfaces when possible. Radiographic imaging should be employed if clinical signs warrant further investigation. Laboratory investigation based on results of a clinical assessment may include a complete blood count with a differential, complete metabolic panel, urine and sputum microscopy and culture, and additional Gram stain and culture specimens from any concerning site. Blood cultures are the only mandatory evaluation and should follow certain guidelines. According to the 2008 ACCM/IDSA guidelines for evaluation of new fever in critically ill adult patients, within the first 24 hours of a fever, and prior to initiation of antibiotics, three to four blood cultures from separate puncture sites should be drawn after decontamination with 2% chlorohexidine gluconate, preferentially, or 1–2% tincture of iodine.53 Patients with intravascular access should have one set drawn through the line and one set drawn peripherally. Access to intravascular device and stopper on the blood culture bottle should be swabbed with 70% alcohol and allowed to dry for 30 seconds prior to drawing 20–30 mL of blood for inoculation. Additional blood cultures should be drawn only for a suspicion of continuation or recurrent bacteremia or for a test of cure 48–96 hours after appropriate antimicrobial or antifungal therapy.

Antipyresis

Despite a lack of compelling evidence-based medicine, fever is commonly treated by pharmacologic and physical mechanisms to establish euthermia in the ICU. Rationale for treating hyperthermia includes therapeutic effect on metabolic consumption and patient comfort. In one study, external cooling was shown to decrease oxygen consumption by 20% in febrile critically ill patients if treated with paralytics to prevent shivering; however, if shivering was not inhibited, external cooling was shown to increase oxygen consumption.65 In a large clinical trial, treatment with intravenous ibuprofen reduced core temperature, heart rate, oxygen consumption, and lactic acid blood levels but did not decrease organ failure or 30-day mortality.66

The goal of treating fever is to reduce the hypothalamic set point and restore the balance of heat production and dissipation. Pharmacologic therapy includes nonsteroidal anti-inflammatory drugs (NSAIDs) and steroids, while physical mechanisms range from externally applied cooling blankets, fans, or ice packs to chilled intravenous fluids, gastric lavage, or intravascular catheter-based techniques. NSAIDs and acetaminophen/paracetamol inhibit the cyclooxegenase pathway and formation of PGEE2, and foster a return to a normothermic hypothalamic set point.67 Aspirin and NSAIDs effectively reduce fever, but can have an anticoagulant effect on platelets. Acetaminophen is the preferred antipyretic in adults, but increases the risk of Reye's syndrome in children and should be avoided. If bacteremia or infection is suspected, targeted antimicrobial therapy should be initiated and de-escalated as appropriate microbiologic information becomes available.

Drug hypersensitivity reactions should be treated with medication withdrawal. MH should be treated with immediate withdrawal of the anesthetic agent in conjunction with intravenous dantrolene and procainamide to prevent ventricular arrhythmia.

Fever is a well-preserved adaptive mechanism that may provide a survival benefit to the host. It is a common, nonspecific physical exam finding in the ICU, which warrants attention. An automatic and protocol drive toward normothermia should be avoided, as the etiology differs depending on underlying medical or surgical factors. Appropriate interventions range from careful observation to immediate and aggressive action, and should be decided on a case-by-case basis; no single “fever workup” should be implemented across all patient populations.

The author would like to thank Kathleen Casey, MD, for critical review of this manuscript.