The usual knee-jerk response to fever is to reduce it (fever here is defined as rectal temperatures >38.3ºC / 101ºF). Although clinicians know that typical fevers below approximately 40-41ºC (104-105.8ºF) are generally harmless for most patients, fevers below this level are still often treated in hospital and intensive care settings. The general perception is that while most fever may not itself be harmful, neither is lowering it.
Acetaminophen (paracetamol) is typically the antipyretic agent of choice, but NSAIDs are also commonly used for this purpose. Although not designed to address the specific question of acetaminophen use for fever lowering, but rather to detect dangerous overdoses of the drug, two recent studies, out of Thomas Jefferson University Hospital in Philadelphia and Partners Healthcare in Boston, have shown that acetaminophen was commonly over-prescribed to supratherapeutic levels in many inpatients. Acetaminophen is in fact one of the most commonly used drugs in inpatient and outpatient settings, often for the purpose of fever reduction.
An increasing amount of evidence is now suggesting that fever lowering in infections, and particularly sepsis, may not be as harmless as previously thought. A number of studies that will be discussed here have shown that fever improves the immune response and leads to better outcomes in infections. Specifically, mortality in sepsis may actually be reduced if the fever is allowed to take its course untreated even to what is considered relatively high temperatures of 40-41ºC/104-105.8ºF).
First, it should be noted that this applies only to fever of infectious origin. Temperature increases may also be caused by non-infectious fever or hyperthermia. Fever occurs when the temperature rise is caused by a resetting of the hypothalamic set point, whereas hyperthermia is caused by an inability to lose sufficient heat at a normal hypothalamic set point. Treatment of these conditions is not discussed here, but the differential diagnosis list is extensive and should be considered early in the evaluation of the febrile patient. Non-infectious causes of fever include thrombosis (eg, DVT, PE), inflammatory and autoimmune diseases (eg, SLE, vasculitis), CNS injury (eg, brain or spinal cord injury or encephalitis causing damage to the body’s thermostat in the hypothalamus or temperature control pathways in the spinal cord), infusion reaction, aspiration, malignancy, and intra-abdominal causes (eg, acalculous cholecystitis, pancreatitis), among others. Hyperthermia may be caused by heat overexposure (eg, heat stroke), endocrine disease (eg, thyrotoxicosis, pheochromocytoma), and malignant hyperthermia induced by certain drugs (eg, anasthetics, neuroleptics, cocaine). [2, 5]
Secondly, fevers exceeding 41ºC (105.8ºF) should typically be treated with antipyretics and optionally cooling blankets. Fevers exceeding this temperature are often associated with non-infectious causes like malignant hyperthermia and may be harmful. Core temperatures that exceed 42ºC (107.6ºF) can cause brain injury and rhabdomyolysis. Infections on the other hand rarely cause fevers higher than 41ºC, and fevers in the range of 38.9ºC (102ºF) to 41ºC (105.8ºF) usually have an infectious source.[2, 3, 5]
The increase in core body temperature during infection affects the immune system in complex and only partially understood ways. Although fever may improve the immune response through enhancement of cytotoxic cell function and modulation of certain cytokines leading to increased pathogen clearance, it also increases metabolic rate, heart rate and oxygen consumption and may overdrive cytotoxic effectors potentially leading to tissue ischemia and injury. Oxygen consumption increases approx. 10% per degree Celsius and heart rate increases approx. 10 bpm per degree Fahrenheit (18 bpm per degree Celsius since 1ºC = 1.8ºF) up to a temperature of approx. 40.5ºC.[4, 5] Patients with cardiopulmonary insufficiency (eg, pneumonia, ARDS, CAD, history of MI) may not tolerate these physiological changes well, and therefore a lower threshold for fever reduction treatment may be necessary in these patients. Very high fevers may also be harmful to fetal development, so the fever treating threshold should also be lower in pregnant women.[3, 5]
Despite these caveats, there is increasing evidence suggesting that allowing higher infectious fevers to go untreated may be beneficial and lead to better outcomes in sepsis management.
In 2005, a study out of the University of Miami examined the effects of antipyretic medicine administration in critically ill febrile patients admitted to the institution’s trauma ICU. The study compared an aggressive treatment group of patients (n=44) treated with 650 mg acetaminophen every 6 hours for fever >38.5ºC (>101.3ºF) and cooling blankets added if it rose over 39.5ºC (103.1ºF), with a permissive treatment group (n=38) receiving acetaminophen and cooling blankets only when the fever rose over 40ºC (104ºF). There were 131 infections in the aggressive group vs. 85 infections in the permissive group (4 +/- 6 vs. 3 +/- 2 infections per patient, p = 0.26), and 7 deaths in the aggressive group vs. 1 death in the permissive group (p=0.06, Fisher Exact Test). Although statistical significance was not reached due to the small sample size, the mortality difference between the groups was so high on interim analysis that the study had to be prematurely stopped due to ethical concerns.
Similarly, a 2006 study out of the University of Virginia looking at fever in surgical patients with blood stream infection (BSI) found that the maximal temperature reached by BSI patients was highly correlated with decreased mortality (Odds ratio 0.596, 95%CI 0.470-0.756, p<0.0001). Based on these results, the authors concluded that reasonable temperature elevations should be allowed to go untreated in patients who don’t have substantial discomfort related to fever.
Reports of the benefits of fever in sepsis and infection have appeared in many other articles and studies. One report published in a 2000 issue of the Journal of Clinical Infectious Diseases looked at the physiology of fever in infection and its possible effects on infection fighting and concluded that antipyretic therapy should be withheld in early sepsis and SIRS until the temperature exceeds 41ºC (105.8ºF) unless the patient has cardiac or pulmonary dysfunction. Another report published in a 2005 issue of the Journal of the Association of Indian Physicians examined the effects of fever on infection in critically ill ICU patients, and similarly recommended that fever <41.1ºC (106ºF) should not be aggressively treated unless patients were pregnant or had limited cardiorespiratory reserve, a recent stroke, or traumatic brain injury.
Although the evidence is compelling, the problem to date has been that most of it is based on small, observational or retrospective studies. What are clearly needed are large prospective, randomized, placebo-controlled trials. Researchers in New Zealand and Australia are now planning to tackle this very challenge.
The Permissive HyperthErmiA Through Avoidance of Paracetamol in Known or Suspected Infection in ICU (HEAT study) is a prospective, randomized, double-blind, placebo-controlled trial that will attempt to determine whether paracetamol (acetaminophen) influences the outcomes of critically ill patients with known or suspected infection. This is a phase 2b trial that will be undertaken in 700 patients with sepsis in 11 ICUs in New Zealand and 11 ICUs in Australia. The study is funded by the Health Research Council of New Zealand. A podcast by Dr. Paul Young, a primary investigator in the study, discussing the role of fever in infection and describing the study in more detail, is available here.)
Although more definitive evidence-based answers regarding fever’s benefits in infection will have to wait for the results of larger randomized studies such as HEAT, a few things are already fairly clear. Fever is the body’s natural response to infection, and the ability to mount a fever is a sign of a functioning immune system and a good prognostic indicator in septic patients. Conversely, a poor fever response is a sign of a weakened immune system that leads to worse outcomes. A number of studies have shown that septic patients who fail to develop a fever have a worse prognosis, likely because their immune system is too impaired to fight the infection. This poor fever response is common in many populations including the elderly and immunocompromised patients (eg, drugs [immunosuppressants, steroids], HIV, asplenia). And septic patients who are hypothermic (<35.5ºC/95.9ºF) have the worst prognosis, with reported mortality being up to twice as high as febrile septic patients. Approximately 10% of sepsis cases fall in this group.[3, 5]
It is therefore important not to wait for fever to develop before suspecting sepsis. Although the SIRS temperature criterion is T>38.3ºC (101ºF) and hospital protocols calling for a more aggressive workup typically don’t kick in until this threshold is reached, the clinician should not base a diagnosis or initiation of treatment solely on this temperature threshold. A correct assessment of the potentially septic patient must consider the entire clinical picture including risk factors, clinical signs, relevant lab results, imaging, and so on. Waiting for a fever to develop before initiating the fever workup can waste precious time and prove to be a costly mistake in some patients.
- A robust fever response in sepsis is indicative of a properly functioning immune system and correlates with lower mortality
- Allowing fevers to go untreated to 40-41ºC may have beneficial effects on infection clearing and reduce mortality, but larger studies are needed (HEAT trial)
- High fevers over 41ºC may be harmful and should be treated; these fevers are associated with non-infectious causes such as malignant hyperthermia
- Fever increases metabolic rate, oxygen consumption and heart rate. Caution and possibly lower temperature thresholds for treating fever should be used in patients with cardiopulmonary disease, CNS injury and pregnant women
- Fever is not always present in sepsis; hypothermic sepsis is associated with higher mortality
- Non-infectious causes of fever and hyperthermia should be ruled out
We would like to receive your feedback on this topic. How do you or your institution treat fever in septic patients, is it promptly treated or allowed to take its course? Do you agree or disagree with the studies that show a potential benefit of fever, and have you seen this effect in your own practice? How important is fever in the initiation of a more extensive sepsis workup in your institution? If you wish to share your feedback or have any comments or suggestions you may email us at [email protected].
Daniel Nichita, MD
- Zhou L, Maviglia SM, Mahone LM, et al. Supratherapeutic Dosing of Acetaminophen Among Hospitalized Patients. Arch Intern Med. 2012;172(22):1721-1728 [available here]
- MacLaren G, Spelman D. Fever in the intensive care unit. UpToDate Online. [available here]
- Hasday JD, Garrison A.Â Antipyretic Therapy in Patients with Sepsis. Clin Infect Dis. 2000 Oct;31 Suppl 5:S234-41 [PubMed]
- John E. Hall. Guyton and Hall Textbook of Medical Physiology 12th ed. Philadelphia, PA. Elsevier Health Sciences;2011
- Kothari VM, Karnad DR. New Onset Fever in the Intensive Care Unit. JAPI. 2005; 53:949-953 [available here]
- Schulman CI, Namias N, Doherty J, et al. The effect of antipyretic therapy upon outcomes in critically ill patients: a randomized, prospective study. Surg Infect (Larchmt). 2005 Winter;6(4):369-75. [PubMed]
- Swenson BR, Hedrick TL, Popovsky K. Is fever protective in surgical patients with bloodstream infection? J Am Coll Surg. 2007;204:815-823. [PubMed, article available here]
This article is strictly informational. ESCAVO does not endorse any drug or treatment discussed here and has no relationship with any companies or organizations potentially mentioned. The treatments, drugs and indications discussed may not be approved by local regulators such as the FDA. Use caution and your own clinical judgment when deciding to use any treatments discussed here.