Fever is probably the most frequent symptom observed in the neurointensive care unit (NICU) by healthcare providers. An oral temperature greater than 37.5℃ is considered a fever [1,2]. Hyperpyrexia is usually a diagnosis of exclusion, with temperatures exceeding 41℃ and nonresponsiveness to antipyretic treatment [3,4]. Fever is seen in almost 70% of neurocritically ill patients [5-10]. Fever of central origin was first described by in the journal Brain by Erickson [11] in 1939. A significant number of these patients have fever from noninfectious causes (47% in Kilpatrick et al. [6] and 25% in Commichau et al. [7]), but this is often treated as infectious fever. Unjustified use of antibiotics adds to the increased cost of treatment and the emergence of resistant strains, contributing to additional morbidity. Since fever has a detrimental impact on the recovery of the acutely injured brain [12-15] and contributes to an increase in length of stay in NICU [8]; timely and accurate diagnosis of the cause of fever in the NICU is imperative.

Here, we try to understand the underlying mechanism, risk factors, clinical characteristics, diagnosis and management options of central fever (CF). We also make an attempt to differentiate two noninfectious causes of fever in NICU, paroxysmal sympathetic hyperactivity and CF.

Overall in the intensive care unit (ICU), at least 50% of fever are reported to be due to noninfectious causes [16]. The incidence of noninfectious fever in the neurology ICU is 23% while in the neurosurgical ICU it is 47% [6]. Of these, the highest rates of febrile episodes occur in patients with subarachnoid hemorrhage (SAH; 50% to 65%), followed by traumatic brain injury (TBI; 4% to 40%) and intracerebral hemorrhage (ICH; 31%), with no cause of fever identified in 28% of patients, suggesting fever of central origin [7,10,17,18]. Hyperthermia is a frequent complication of acute ischemic stroke in 50% of these patients and carries a poor prognosis [19].

An abnormal rise in temperature may be physiological, environmental, or even drug-related, rather than due to infection. The mechanism of CF in the NICU is not well defined but the literature suggests some probable mechanisms. Inflammatory markers causing fever may be triggered by extreme physiologic stress in acute neurologic injury [20,21]. Brain injury may also lead to the disruption of the mesencephalic-diencephalic mechanisms responsible for the inhibition of thermogenesis [22]. Monocytes and macrophages produce the cytokines interleukin 1 (IL-1), IL-6, and tumor necrosis factor α (TNF-α), which act on the organum vasculosum of laminae terminalis. This in turn leads to the release of prostaglandin E2 (PGE2) via activation of the cyclooxygenase-2 (COX-2) enzyme. PGE2 acts on the preoptic area of the hypothalamus leading to an increase in the set point of the hypothalamus, thereby increasing the body temperature [23-34]. Systemic pyrogens, such as IL-1, appear to enter the brain at regions where there is an incomplete blood-brain barrier (circumventricular organs) and act on the preoptic area of the hypothalamus to induce fever [24,35-38]. Various neurological events that take place in febrile patients affect this pathway. Direct hemotoxic damage to thermoregulatory centers in the preoptic region, interference with tonic inhibitory inputs from lower midbrain that ordinarily suppresses thermogenesis, and stimulation of prostaglandin production leading to temperature set point elevation, have all been implicated in the causation of CF [39]. CF is speculated to result from damage to the hypothalamus, midbrain, or pons, and be enhanced by increased sympathetic activity, opening of the ventricles, damage to the frontal lobes, physical distortion, diffuse axonal injury (DAI), or toxic blood metabolites [39]. CF may also be due to the selective loss of warm-sensitive neurons, the osmotic changes detected by the organum vasculosum laminae terminalis, or from hormonal changes (progesterone, prostaglandin) modifying the firing rate of temperature-sensitive neurons in the medial preoptic nucleus [40].

Posttraumatic hyperthermia, also known as neurogenic fever, is another common cause of fever. Stimulation studies have suggested that the mechanism involves an imbalance between the hypothalamus and the various temperature regulating centers in the brainstem and spinal cord [41]. Won and Lin [42], in their study conducted on rabbits, found that inhibition of five hydroxytryptamine receptors in the anterior hypothalamus increased heat production and decreased heat loss, leading to hyperthermia. Suggested mechanisms for this effect include increased metabolic rate, increased carbon dioxide production, decreased cerebral blood flow, acidosis, brain edema exacerbation, excitotoxic neurotransmitter release, and blood-brain barrier breakdown. Disease-specific mechanisms are also described in Table 1.

Various predisposing factors have been defined for the occurrence of CF in the NICU. Independent predictors of CF on multivariate analysis include blood transfusion, SAH, IVH, tumor, or onset of fever within 72 hours of hospital admission [57]. Intraventricular catheterization is a risk factor for unexplained fever, which suggests a role for ventricular hemorrhage in the pathogenesis of CF [7]. Risk factors among various acute neurological conditions are reported as follows (Table 1).

This is a diagnosis of exclusion. CF occurs early, typically within 72 hours of admission after acute brain injury. All the cultures are negative and the chest radiograph is normal. Fever is disproportionately high and persistent. The temperature peak is higher when the fever starts earlier, and will be higher when compared to infectious fever [18]. There is generally less fever-free period and the cumulative fever load is high, accounting for a longer stay in the NICU. Generally, CF is continuous in nature without diurnal variations, plateau-like, and without spikes. Patients with CF have relative bradycardia with a notable absence of perspiration. Sustained fever is another factor in favor of CF [57]. Fever is also resistant to antipyretic medications [62,64-67].

Continuous fever, lasting longer than 6 hours for 2 or more consecutive days, has been considered persistent [57]. The combination of negative cultures; absence of infiltrate on chest radiographs; diagnosis of SAH, IVH, or tumor; and onset of fever within 72 hours of admission, predict CF with a probability of 0.90 [57]. In a study conducted by Thompson et al. [15], fever persisted for weeks to months in 4% to 37% of patients with TBI. Vasospasm with SAH is also predictive of CF [22].

The criteria for systemic inflammatory response syndrome and leukocytosis are similar to central and infectious fever. This underscores the difficulty in differentiating central and infectious fever prospectively in the critically ill population. The extreme physiologic stress provoked by acute neurologic injury can cause a rise in inflammatory markers and increased sympathetic response [21,22]. The percentage of neutrophils is higher in patients with infectious fever, suggesting that while leukocytosis may not be a reliable clinical variable to decide whether to use empirical antibiotics or discontinue antibiotics early, the presence of a left shift remains useful [57].

Extreme hyperpyrexia, defined as fever ≥41.1℃ (106°F), is usually noninfectious. Examples include CF, drug fever, malignant hyperthermia, transfusion reactions, adrenal insufficiency, thyroid storm, neuroleptic malignant syndrome, heat stroke, acalculous cholecystitis, mesenteric ischemia, acute pancreatitis, deep vein thrombosis, and pulmonary embolism [68,69]. A single fever spike of 102°F is classical for noninfectious disorders and is never due to infection. Fever associated with blood transfusions are usually transient, that is, they present as a single fever spike within <1 week [68,70].

Relative bradycardia is a feature of CF. The following applies to adult patients with temperatures >102°F and when pulse is taken simultaneously with temperature. Normally, the pulse rises in concert with the temperature, (e.g., for every degree Fahrenheit temperature is increased, the pulse should rise 10 beats/min). If the pulse rate is lower than predicted from a given temperature (>102°F), then relative bradycardia is present, unless the patient is on a beta-blocker, verapamil or diltiazem, or has a pacemaker-induced rhythm or heart block. In absence of these exclusion criteria, relative bradycardia in neurosurgical ICU patients with fever strongly suggests a central or drug fever (Table 3) [71].

Some common differential diagnoses are important to be distinguished before making a diagnosis of CF, since it a diagnosis of exclusion. Nonresponse to antibiotics in CF may lead to misdiagnosis of antibiotic failure or resistance in CF (Table 4).

Both are diagnoses of exclusion in patients with neurological injuries during their stay in neurocritical care. They both occur in the NICU in acute brain-injury patients. Differentiating one from the other is very crucial in proper care and appropriate management, and thus ultimately affects patient outcomes. Differentiating features are listed in Table 5.

Although many treatment regimens have been suggested, none have been identified as superior to others in the treatment of fever [65,105]. However, controlling fever is an important part of management in CF, owing to its detrimental effects on the brain. Pharmacologic methods include acetaminophen, acetylsalicylic acid, and other nonsteroidal antiinflammatory medications and corticosteroids [106,107].

Other methods to decrease temperature include rotary fans, sponging, and surface cooling devices. However, these have had limited efficacy and are uncomfortable for the patient. Surface cooling devices have also been reported to increase the incidence of shivering, increase oxygen consumption and even cause thermal burns [107,108]. Hypothermia blankets can lead to large temperature fluctuations [109,110]. Air blankets have been increasingly used and are found to have better efficacy and produce better patient comfort [110]. Some authors have suggested the use of sand body-conformed wraps, intravascular cooling devices, head-only cooling caps, or inhaled perfluorocarbon cooling systems [111].

Several studies have tried intravenous infusion of cold saline, showing promising results and no increase in complications [112]. A few studies have tried local (brain) cooling. This may prevent the side effects of global hypothermia, such as impaired coagulation, arrhythmias and deep vein thrombosis [113-116]. However, no large multicentre trial is available, leading to no definitive conclusions on the use of selective brain cooling.

IL-1 antagonists have been shown to produce significant improvements in rat models of TBI [117], although no human trials have been conducted. However, it has been shown that even a small temperature decrease in febrile patients can improve neurologic outcomes [118].

There are no guidelines or directions to help differentiate CF from other noninfectious causes of fever in the NICU. The literature is sparse and unclear. The diagnostic criteria are not well defined and not standardized. Treatment modalities for this clinical entity have been symptomatic only and mostly rely on over the counter drugs. No standard therapy has been defined in the literature. Multicentre large studies are required to better define CF, understand its pathophysiology, and guide standard management protocols in neurocritical care settings.

CF is an important diagnosis in neurocritical care. It not only prevents unnecessary antibiotic use, but its early recognition would also help improve patient management and prevent delayed discharge from the hospital. The current key to diagnosis in a predisposed patient is a high index of suspicion, along with thorough clinical examination, radiological, microbiological, and biochemical tests. Immediate attainment of normothermia is the current recommendation, as fever worsens the brain insult. Treatment includes various pharmacological agents and surface cooling methods to decrease body temperature. Studies are lacking on the best methods for diagnosis, treatment, and prevention of CF. Thus, more human trials are needed in this field to make any definitive recommendations.