Search tips
Search criteria 


Logo of jnnpsycJournal of Neurology, Neurosurgery and PsychiatryVisit this articleSubmit a manuscriptReceive email alertsContact usBMJ
J Neurol Neurosurg Psychiatry. 2007 November; 78(11): 1278–1280.
PMCID: PMC2117587

Non‐infectious fever in the neurological intensive care unit: incidence, causes and predictors


Background and objective

Non‐infectious causes of fever are often considered in critical neurological patients but their true significance has not been formally studied. The aim of this study was to evaluate the incidence, causes and predictors of fever in patients with acute neurological/neurosurgical disease and no documented infection.


Prospective data collection of consecutive patients admitted to the neurological intensive care unit (NICU) of an academic medical centre for more than 48 h was carried out. Fever was defined as body temperature [gt-or-equal, slanted]101°F (38.3°C) documented on at least one measurement for 2 consecutive days. Patients were enrolled only if a diagnostic workup, including cultures of [gt-or-equal, slanted]2 body samples, was performed before antibiotic use. Febrile patients with no proven evidence of infection were considered to have non‐infectious fever.


93 patients were included in the final analysis. Fever was non‐infectious in 31 patients (33%). There were no differences between the infectious and non‐infectious fever groups in terms of age, use and duration of invasive catheters, daily duration of fever and number of days with fever. Documented infections tended to be more common among febrile patients with traumatic brain injury (52% vs 36%; p = 0.06). Non‐infectious fever was more frequent among patients with subarachnoid haemorrhage (48% vs 18%; p = 0.01) in whom it was associated with vasospasm (p = 0.03) and symptomatic vasospasm (p = 0.05). Non‐infectious fever started earlier (mean 2.6 vs 4 days; p = 0.007) and onset of fever within the first 72 h of admission predicted negative evaluation for infection (p = 0.01). Subarachnoid haemorrhage and fever onset within the first 72 h were independent predictors of non‐infectious fever on multivariable analysis.


Fever in the absence of documented infections occurs commonly in the NICU, especially among patients with subarachnoid haemorrhage and vasospasm. Early onset of fever predicts a non‐infectious cause.

Between one‐quarter1 and more than one‐half2,3,4 of patients admitted to the neurological intensive care unit (NICU) develop fever. The cause of fever in these patients often remains unexplained. As hyperthermia is strongly detrimental for the recovery of the acutely injured brain5,6,7,8,9 and contributes to an increase in the length of stay in the NICU,4,10 timely and accurate diagnosis of the cause of fever in the NICU is crucial.

Infections are the most common causes of fever in the NICU population but they are only documented in half of all febrile patients.1 Central fever related to loss of the physiological regulation of body temperature by the hypothalamus is often proposed as a possible cause for persistent fever in acute neurological patients with no evidence of infection. There are no current means to confirm the diagnosis of central fever. Thus we can only assess its frequency by studying patients in whom fever remains unexplained after thorough investigations for infectious and other non‐infectious causes.

The objective of this study was to determine the incidence, causes and predictors of non‐infectious fever in patients with acute neurological disease.

Patients and methods

All patients admitted for more than 48 h to the 24 bed NICU of Jackson Memorial Hospital (Miami, Florida, USA) between June 2004 and April 2005 were screened for the presence of fever. In all patients, body temperature was measured continuously by bladder sensors and documented hourly throughout their NICU stay. Fever was defined as body temperature [gt-or-equal, slanted]101°F (38.3°C) documented on at least one measurement for two consecutive days. Clinical data were prospectively collected in all febrile patients during the study period. If more than one episode of fever occurred, only the first episode was analysed.

Documented infectious fever was defined according to the following criteria11,12: (1) positive culture of body fluids, tissue samples or indwelling catheters by recognised pathogenic microorganisms, (2) sinusitis documented by CT and transnasal puncture, (3) diarrhoea and positive Clostridium difficile toxin in stools, (4) lung, abdominal or pelvic abscess documented by CT, (5) complicated surgical wound or (6) clinical evidence of cellulitis. Cultures of [gt-or-equal, slanted]2 body samples before antibiotic use were required to enter the study.

Patients without documented infection were categorised as having non‐infectious fever. All possible non‐infectious causes of fever, excluding central fever, were noted using predefined diagnostic criteria.11,12 All febrile patients were uniformly treated according to a practice protocol with acetaminophen and mechanical cooling measures (cooling blankets, ice packs). Induced hypothermia was not routinely used in our NICU during the study period.

Collected data included demographics, primary neurological diagnosis, fever onset and duration, presence of persistent fever (operationally defined as continuous fever for more than 6 h over 2 or more consecutive days), presence of central venous catheter, ventriculostomy catheter and endotracheal tube upon fever onset and time from their placement, cultures performed and their results, possible non‐infectious causes of fever, use of empiric antibiotics, length of ICU stay and mortality. In patients with subarachnoid haemorrhage (SAH), we documented the presence of vasospasm (ie, mean blood flow velocity >120 cm/s on transcranial Doppler or vessel narrowing on angiography), associated neurological symptoms (symptomatic vasospasm) and timing of vasospasm in relation to fever onset.

We analysed predictors of non‐infectious fever using the Fisher exact test for nominal (binary) variables and univariate logistic regression for continuous variables. We then tested the independence of the associations using a multivariable logistic regression model. A p value <0.05 was considered significant. The Institutional Review Board of the University of Miami School of Medicine approved the study design.


Ninety‐three febrile patients were included in the final analysis. Fever was not associated with documented infection in 31 patients (33%). In only four of these patients, all with head trauma, did we identify potential non‐infectious causes for the fever other than the primary brain injury (alcohol withdrawal in three and phenytoin toxicity in one). Hence, 27 patients (29%) had unexplained, non‐infectious fever. Positive culture of respiratory secretions, present in 46 patients (49% of the total population and 74% of patients with documented infection), was the most common criterion determining the diagnosis of infectious fever. Possible causes of non‐infectious fever were noted in 10 patients with documented infection (16%), including pancreatitis in four, deep venous thrombosis in two, pulmonary contusions in two and drug toxicity/withdrawal in two.

There were no differences between the infectious and non‐infectious fever groups in age, sex, rate and duration of use of central venous and ventricular drainage catheters or tracheal tube, number of days with fever and prevalence of persistent fever. Patients with traumatic brain injury tended to have infectious fever more commonly (52% vs 36%; p = 0.06) while non‐infectious fever predominated in patients with SAH (48% vs 18%; p = 0.01). Vasospasm was diagnosed in 12/15 (80%) patients with SAH and non‐infectious fever (symptomatic in seven; preceding onset of fever in six) and in 4/11 (36%) patients with SAH and documented infection (symptomatic in only one; preceding onset of fever in three) (p = 0.03 for vasospasm and p = 0.05 for symptomatic vasospasm). Neither radiological grade (Fisher score) nor placement of intraventricular catheter was related to non‐infectious fever.

Median number of sites cultured was 3.5 and 94% of patients had cultures from [gt-or-equal, slanted]3 sites. Empiric antibiotics were used in 72% of cases of non‐infectious fever but these treatments were usually stopped when the negative infectious workup was confirmed (mean duration of antibiotic therapy 3.7 days).

Fever started significantly earlier in patients without evidence of infection (mean 2.6 vs 4 days; p = 0.007). In fact, onset of fever within the first 72 h of NICU admission was associated with negative results on the evaluation for infection (p = 0.02). Multivariable analysis confirmed that SAH and fever onset within 72 h of admission were strongly associated with non‐infectious fever. It also showed a trend towards an inverse association between age and occurrence of non‐infectious fever.

Table 11 summarises the comparison between the groups of patients with infectious and non‐infectious fever on univariate analysis. Table 22 presents the results of the multivariable logistic regression analysis.

Table thumbnail
Table 1 Comparison of patients with documented infectious fever and non‐infectious fever in the neurological intensive care unit
Table thumbnail
Table 2 Multivariable analysis of predictors of non‐infectious fever among 93 patients with acute neurological illness admitted to the neurological intensive care unit


Our results indicate that unexplained, non‐infectious fever is common in the NICU. Unexplained fever is significantly more frequent among patients with SAH and is associated with the development of vasospasm and symptomatic vasospasm. Non‐infectious fever tends to start early (within 72 h of admission) and remains present for several days.

Nearly one‐third of our patients had fever with no identified cause other than the primary brain injury. Although central fever cannot be proven, it is the most likely explanation for the hyperthermia in these cases. Impaired central mechanisms of thermomodulation may have also contributed to the occurrence of fever in some patients with documented infection, but this possibility cannot be assessed with our study design. The coexistence of infection and disturbed central thermoregulation in some patients may explain the lack of difference in the duration of fever between the infectious and non‐infectious groups.

Early fever onset predicted a non‐infectious cause in our population. Thus it may be reasonable to withhold antibiotics within the first 72 h in the absence of other signs of infection until culture results become available. If empiric antibiotics are prescribed early, they should be promptly discontinued if cultures are negative. Persistent hyperthermia is often regarded as suspicious for central fever, as opposed to the spiking pattern most commonly seen with infections. However, persistent fever, at least as defined for this study, was not a distinctive feature of non‐infectious fever. More detailed analyses of the fever curve, including assessment of sustained maximal temperatures, should be part of future studies.

Accurate and timely recognition of non‐infectious fever can avoid unnecessary use of antibiotics, thus reducing the risk of toxic effects, drug interactions and emergence of multi‐resistant microorganisms. In patients with unexplained or suspected central fever, treatment should be focused on cooling measures to prevent the deleterious effects of hyperthermia on the injured brain.5,6,7,8,9,13,14,15 Novel techniques to reduce body temperature, such as surface or intravascular cooling devices, can be particularly useful in these situations.16,17,18 The need for more effective treatments to combat hyperthermia is underscored by the prolonged duration of fever in our population.

We found that SAH is associated with unexplained fever and, more specifically, with vasospasm and symptomatic vasospasm, as suggested by previous results.1,5,13 Fever worsens functional outcome in SAH, thus febrile patients must be treated with cooling therapies early and aggressively.16 It is unclear if SAH shares a common pathophysiology with vasospasm (such as upregulation of prostaglandin E2)19 or if it is caused by vasospasm induced damage on the hypothalamus and related structures.13

This study has several limitations. We did not include parameters of severity of acute systemic illness in our analysis; however, it is unlikely that including these parameters would have modified our main results. The effects of treatment of hyperthermia on our results cannot be determined, but they represent standard practice. We only collected data on the first episode of fever during the NICU admission. Thus our results do not fully reflect the situation of patients with prolonged admissions who may have multiple episodes of fever from different infectious and non‐infectious causes. Reliance on positive cultures to define infectious fever may have led to an overestimate of the frequency of infections; thus the true prevalence of central fever could actually be higher.

Disturbances of central mechanisms of thermoregulation are likely responsible for the common occurrence of non‐infectious hyperthermia in NICU patients. The causes of this phenomenon remain speculative. Possibilities include direct damage to thermoregulatory centres, temperature set point elevation induced by cytokines, disruption of mesencephalic–diencephalic physiologically responsible for tonic inhibition of thermogenesis and impairment of peripheral mechanisms of heat dissipation by the overriding effect of sympathetic activation.1,20


NICU - neurological intensive care unit

SAH - subarachnoid haemorrhage


Competing interests: None.


1. Commichau C, Scarmeas N, Mayer S A. Risk factors for fever in the neurologic intensive care unit. Neurology 2003. 60837–841.841 [PubMed]
2. Kilpatrick M M, Lowry D W, Firlik A D. et al Hyperthermia in the neurosurgical intensive care unit. Neurosurgery 2000. 47850–855.855 [PubMed]
3. Albrecht R F, Wass C T, Lanier W L. Occurrence of potentially detrimental temperature alterations in hospitalized patients at risk for brain injury. Mayo Clin Proc 1998. 73629–635.635 [PubMed]
4. Stocchetti N, Rossi S, Zanier E R. et al Pyrexia in head‐injured patients admitted to intensive care. Intensive Care Med 2002. 281555–1562.1562 [PubMed]
5. Oliveira‐Filho J, Ezzeddine M A, Segal A Z. et al Fever in subarachnoid hemorrhage: relationship to vasospasm and outcome. Neurology 2001. 561299–1304.1304 [PubMed]
6. Reith J, Jorgensen H S, Pedersen P M. et al Body temperature in acute stroke: relation to stroke severity, infarct size, mortality, and outcome. Lancet 1996. 347422–425.425 [PubMed]
7. Rossi S, Zanier E R, Mauri I. et al Brain temperature, body core temperature, and intracranial pressure in acute cerebral damage. J Neurol Neurosurg Psychiatry 2001. 71448–454.454 [PMC free article] [PubMed]
8. Schwarz S, Hafner K, Aschoff A. et al Incidence and prognostic significance of fever following intracerebral hemorrhage. Neurology 2000. 54354–361.361 [PubMed]
9. Thompson H J, Tkacs N C, Saatman K E. et al Hyperthermia following traumatic brain injury: a critical evaluation. Neurobiol Dis 2003. 12163–173.173 [PubMed]
10. Diringer M N, Reaven N L, Funk S E. et al Elevated body temperature independently contributes to increased length of stay in neurologic intensive care unit patients. Crit Care Med 2004. 321489–1495.1495 [PubMed]
11. Marik P E. Fever in the ICU. Chest 2000. 117855–869.869 [PubMed]
12. O'Grady N P, Barie P S, Bartlett J. et al Practice parameters for evaluating new fever in critically ill adult patients. Task Force of the American College of Critical Care Medicine of the Society of Critical Care Medicine in collaboration with the Infectious Disease Society of America. Crit Care Med 1998. 26392–408.408 [PubMed]
13. Rousseaux P, Scherpereel B, Bernard M H. et al Fever and cerebral vasospasm in ruptured intracranial aneurysms. Surg Neurol 1980. 14459–465.465 [PubMed]
14. Baena R C, Busto R, Dietrich W D. et al Hyperthermia delayed by 24 hours aggravates neuronal damage in rat hippocampus following global ischemia. Neurology 1997. 48768–773.773 [PubMed]
15. Hickey R W, Kochanek P M, Ferimer H. et al Induced hyperthermia exacerbates neurologic neuronal histologic damage after asphyxial cardiac arrest in rats. Crit Care Med 2003. 31531–535.535 [PubMed]
16. Badjatia N, O'Donnell J, Baker J R. et al Achieving normothermia in patients with febrile subarachnoid hemorrhage: feasibility and safety of a novel intravascular cooling catheter. Neurocrit Care 2004. 1145–156.156 [PubMed]
17. De Georgia M A, Krieger D W, Abou‐Chebl A. et al Cooling for acute ischemic brain damage (COOL AID): a feasibility trial of endovascular cooling. Neurology 2004. 63312–317.317 [PubMed]
18. Mayer S A, Kowalski R G, Presciutti M. et al Clinical trial of a novel surface cooling system for fever control in neurocritical care patients. Crit Care Med 2004. 322508–2515.2515 [PubMed]
19. Frosini M, Sesti C, Valoti M. et al Rectal temperature and prostaglandin E2 increase in cerebrospinal fluid of conscious rabbits after intracerebroventricular injection of hemoglobin. Exp Brain Res 1999. 126252–258.258 [PubMed]
20. Shibata M. Hyperthermia in brain hemorrhage. Med Hypotheses 1998. 50185–190.190 [PubMed]

Articles from Journal of Neurology, Neurosurgery, and Psychiatry are provided here courtesy of BMJ Publishing Group