Despite its frequency, there is no generally accepted definition of fever. A practical definition is a temperature above 38°C (100.4°F) by mouth or above 38.4°C (101°F) by rectum. Lower temperatures have been proposed as a definition of fever but are not practical and contribute to excessive concern, especially because it is not abnormal for a child to have a rectal temperature between 100°F and 101°F in the afternoon or after exercise.¹

Fever is not equivalent to hyperthermia. Fever is an adaptive response that is well regulated by the body and is not dangerous in and of itself (although the cause of the fever may be quite serious). Temperatures due to fever almost never exceed 41°C (105.8°F). In contrast, hyperthermia is an elevated body temperature caused by a dysregulation of the normal mechanisms and can be very dangerous, with temperatures exceeding the body’s set point. Examples include heat stroke (in which body temperature is elevated by external means) and malignant hyperthermia (caused by markedly increased heat production via uncoupling of oxidative phosphorylation).

Normal body temperature shows a diurnal variation, being lowest before awakening and highest in the late afternoon or evening. Fever curves usually follow this diurnal pattern also.

Conversion between Centigrade and Fahrenheit degrees can be made using the following formulas:

Axillary temperature measurement is slow, has poor sensitivity in detecting fever, and is not recommended for use in infants and children in outpatient settings.⁵ In rapidly breathing patients, even when there is no obvious mouth breathing, oral temperatures may be erroneously low.⁶ The so-called tempadots (papers that are placed on the child’s forehead) and temperature-taking pacifiers are inaccurate and should not be used. Tympanic membrane thermometers (“ear thermometers”) have the advantage of being rapid, noninvasive, and painless; however, their accuracy has been questioned in several studies. Mothers’ estimation of high fever in young children (< 2 years old) and their estimation that a child has no significant fever are very accurate most of the time, without the use of a thermometer.⁷ However, the determination of so-called tactile fever should not be relied on in infants in the first 2 months of life.⁸

Medical lore holds that children with severe central nervous system (CNS) abnormalities have decreased temperature regulation and exaggerated fever responses, and this is supported by some studies.^9,10

Corticosteroids, like antipyretics, can obscure the presence of fever. Sometimes, high fever occurs in children receiving high doses of corticosteroids, and other times, fever is obscured by steroids, especially in tuberculosis.¹¹

In each child with a fever, the physician should first ask whether any symptomatic treatment is really necessary. The phrase fever phobia has been used to describe undue worry by parents about fever.^12,13 A recent study showed that fever phobia persists despite efforts at education. About 25% of parents became worried enough about their child’s fever that they measured his/her temperature five or more times a day, and an equal percentage of parents slept in the same room as the child with fever.¹⁴ Fear of fever has also been documented to exist among medical professionals.¹⁵

Guidelines in counseling parents have been recommended for use for children over 6 months of age.^1,13 These include retaking the temperature after the child rests for one-half hour, defining 105°F as a high fever, teaching parents that normal body temperature regulation will keep the fever from going “out of control,” and using antipyretic therapy only when fever makes the child uncomfortable.¹³ Sponging has been recommended only if the temperatures are over 40°C (104°F) and then only if the temperature has not improved within 1 hour of giving an antipyretic. The child is not to be awakened for temperature taking or antipyretic administration because sleep is more important. The parents are taught that temperature taking or “breaking the fever” is not a substitute for more important observations, such as watching for dyspnea, change in consciousness, or pain.^1,13

In general, excessive clothing or blankets should be removed to a point of comfort. Hydration with oral fluids is usually advised but should not be forced.

Sponging with tepid (but not cold) water is a comfortable and effective method that is traditional for high fever. One study showed that over the first 30 minutes sponging was more effective than either acetaminophen or ibuprofen, but by 60 minutes it was inferior to both of these agents.¹⁶

Sometimes hospitalized patients with high fevers are placed on cooling blankets to decrease their body temperature. A study of febrile adults in an intensive care unit found that the use of cooling blankets plus acetaminophen was no more effective in reducing fever than the use of acetaminophen alone. In addition, the use of cooling blankets was associated with wide fluctuation in temperatures and with rebound hypothermia.¹⁷

Unlike antipyretics, external cooling acts not by reducing the elevated set point but by overwhelming metabolically expensive effector mechanisms that have been evoked by the elevated set-point.¹⁸

For patients with hyperthermia, external cooling may be lifesaving. However, for patients with fever, the use of cooling blankets is nonsensical, as it causes forced peripheral vasoconstriction at a time when the body is attempting to dissipate heat by vasodilatation.

Sometimes, the treatment of fever produces more problems than the fever. Complications of treatment include antipyretic overdosage, particularly aspirin poisoning (salicylism), which was more common in the past. Acetaminophen can produce delayed liver toxicity or the rare complications of dermatitis, hypoglycemia, agranulocytosis, thrombocytopenia, and methemoglobinemia. Overdosage may produce prolongation of the prothrombin time, vomiting, hepatic failure, and death after 2-7 days. The toxicity of aspirin and acetaminophen have been reviewed in detail.²⁴

Children administered ibuprofen have a higher rate of gastrointestinal bleeding than those treated with acetaminophen. Renal side effects have been reported, too, especially in children with preexisting renal problems or intravascular volume contraction.²⁵ There has been some concern that the use of ibuprofen for antipyresis in children with varicella may increase the risk of necrotizing fasciitis caused by Group A streptococcal superinfection. The evidence in favor of this association is one case-control study and one retrospective cohort study. The case-control study showed an odds ratio of 11.5 for ibuprofen use in children with necrotizing fasciitis versus the control group;²⁶ in many of these cases, ibuprofen use was initiated after the child had developed symptoms of secondary infection. This fact is used by many to say that perhaps ibuprofen use was simply a marker for more severe infection. However, studies in a rabbit model suggest that ibuprofen is capable of preventing neutrophil adhesion, suppressing azurophil granule secretion, and decreasing the production of superoxide,²⁷ all of which are mechanisms by which ibuprofen potentially could increase the risk of progression of an existing superinfection. The cohort study showed that if ibuprofen was dispensed to a child within the 30 days prior to the onset of varicella, the rate of bacterial skin superinfection was 3.1 times higher; if only superinfections severe enough to warrant hospitalization were included, that ratio rose to 5.1.²⁸ Problems with this study include that there is no evidence the children actually took the medication and that the confidence intervals of both odds ratios are wide, and the P values nonsignificant. Nevertheless, the suggestion of a possible association, coupled with the severity of the disease in question and the ready availability of an alternative antipyretic agent would seem to argue against the use of ibuprofen in children with varicella.

Poisoning can occur from inhalation of alcohol used to sponge the skin and may produce hypoglycemia and coma.^29,30

In hospitalized patients who are being treated with an antibiotic, the use of antipyretics for moderate fever can interfere with the interpretation of the effectiveness of the antibiotic. However, if the infecting organism is known and the therapy known to be appropriate, then use of antipyretics for discomfort is appropriate.

It is helpful to have a classification of fever syndromes for use in the problem-oriented approach. The classification in Table 10-2 was developed by Dechovitz and Moffet in 1968 by analyzing the records of 155 children hospitalized for fever and is the basis for subsequent individual sections in this chapter.³¹

Prolonged unexplained fever is defined as a documented fever higher than 38.4°C (101°F) occurring daily for more than 10 days. This pattern is often called “fever of unknown origin (FUO).”

Fever without localizing signs is defined as a fever of less than or equal to 10 days’ duration, with no signs of the source of the infection on physical examination and normal urine. Newborn infants younger than 2 months old are excluded from this category, and children from 2 months to 2 years of age are considered a special subgroup because of an increased risk for occult bacteremia. At the time this classification was devised, many physicians were using the diagnosis of FUO wrongly to describe children with fever of brief duration without localizing signs. Of course, not every child with fever has a urinalysis and/or urine culture done in practice, but it is useful to retain this requirement in the definition to remind the clinician that the possibility of a urinary infection has not been excluded.

Fever with Nonspecific signs is defined as a fever of less than or equal to 10 days’ duration and some abnormal physical findings that direct investigation to a specific area.

Fever complicating a chronic disease is defined as fever in a patient with a disease known to have a predilection for a particular febrile complication. Many chronic diseases have a particular expected febrile complication, such as subacute bacterial endocarditis or brain abscess, which must be considered in fever with congenital heart disease.

Fever in an immunocompromised host should be considered a separate category because the diagnostic considerations must be broadened to include opportunistic infections and because the urgency of providing specific therapy is usually greater (see Chapters 20, 22, and 23).

Recurrent fever is defined as separate episodes of definite high fever.

Periodic fever is defined as recurrent episodes of definite fever that occur at regular or predictable intervals.

Influenza-like illness is defined as fever with prominent respiratory symptoms of cough and sore throat without remarkable respiratory signs such as dyspnea or rales and usually including myalgia. This syndrome is discussed in Chapter 7. “Viral syndrome” is an inadequate working diagnosis, as it lacks the specificity of the previous diagnostic phrases.

Pseudofever is the term used when parents bring a child to medical attention for temperatures that are above their concept of “normal” but do not meet the definition of fever given earlier; that is, temperature greater than 37°C (98.6° F) but less than 38°C (100.4° F) oral or 38.4°C (101°F) rectal. Because this is not true fever at all or is just diurnal or exercise temperature variation, the clinician should try to avoid reiterating the parent’s use of the word fever. It is helpful to explain to the parents that this is a normal variation of temperature, especially when the child’s appearance and examination are normal, as discussed later.

Fever without localizing signs is a tentative or working diagnosis and is best defined as:

A urine dipstick examination that includes testing for leukocyte esterase and bacterial products such as nitrites might be substituted for a microscopic study. However, if a clean-catch urine has been obtained, the additional time involved in the more thorough study is usually minimal. It is not rare to discover important urinary tract disease when a proper urinalysis is done after a number of febrile illnesses attributed to respiratory infections (see Chapter 14).

The preliminary diagnosis of fever without localizing signs should be reserved for patients who do not appear seriously ill. Suspected septicemia should be the preliminary diagnosis if the patient is seriously ill or hypotensive.

Fever without localizing signs (FWLS) also should be defined to exclude the newborn and young infant in the first 2 months of life (Table 10-2). Neonatal fever is a better preliminary descriptive diagnosis and raises the question of sepsis, discussed in Chapter 19.

Fever without localizing signs as a preliminary diagnosis helps to avoid using more exact, but less certain, diagnoses. Other descriptive diagnoses with a similar meaning include “fever only,” “undifferentiated febrile illness,” “undiagnosed fever,” and “fever, not seriously ill-appearing.” All of these preliminary diagnoses are acceptable, but the term fever without localizing signs is being used more widely and is readily understood.

Although “flu syndrome” has been used to describe this pattern, it is not accurate, because influenza-like illnesses characteristically have prominent respiratory symptoms, especially cough and sore throat, as described in Chapter 7. “Viremia” has also been used to describe this syndrome, but this term is neither useful nor accurate, because documentation of a virus in the blood is rarely possible using currently available methods.

“Upper respiratory infection” (URI) is also an inappropriate diagnosis, because these patients do not have sufficient upper respiratory signs or symptoms to account for the magnitude of the fever. URI is too vague a diagnosis even when respiratory symptoms are present.

The preliminary diagnosis of fever without localizing signs may be changed to another diagnosis as the course of the illness evolves. There are several courses the illness may take:

“Self-limited febrile illness” is a retrospective descriptive diagnosis used for a fever persisting for several days from which the patient recovers without antibiotic therapy and without any localized infection, rash, or other signs.³² A urinary tract infection must be reliably excluded to make this diagnosis. If antibiotic therapy is used, the term self-limited is not appropriate. This syndrome is usually presumed to be a viral illness and has also been called “3-day fever,” “acute undifferentiated febrile illness,” “systemic infection,” and “febrile illness of short duration.”

In the child over 2 years of age with fever without localizing signs, clinical decisions concerning hospitalization and therapy can be based on the clinical findings, ability of parents to judge changes, and convenience and reliability of follow-up contacts. Usually, the child can be followed by telephone if done carefully, with return visits as necessary for reexamination and without antibiotic therapy if there is no specific likely explanation for the fever. New symptoms or signs may be noted on follow-up, the illness may resolve without a diagnosis, or hospitalization may be needed for further study.

Beginning in the early 1970s, there was an increasing appreciation of the occurrence of unrecognized (“occult” or “outpatient”) bacteremia in young children (younger than 2 years and younger than 3 months became established as the major age breakpoints). Within recent years, the cutoff age for hospital admission for suspected sepsis in babies with fever has been changed to 2 months in some centers, and to 4–6 weeks in others. Prospective studies have looked at many clinical, laboratory, and epidemiologic variables, so that more accurate information is now available about one of the most difficult clinical diagnostic problems: high fever in young children.

It is not surprising that age breakpoints have become so helpful in assessing the probabilities for various infections, as pediatrics is a field that makes progressively finer age distinctions at younger ages.

Infants younger than 2 months are susceptible to more diseases and are often treated similarly to newborns, but they also may have many of the same diseases as older infants. Early in this period, there is less social eye contact and less social smiling, and as a consequence, the general appearance is harder to evaluate.⁴⁵ Additionally, babies are unable to communicate, which makes interpretation and diagnosis of their illnesses somewhat trickier. Finally, young infants often develop similar, nonspecific signs and symptoms in response to a wide variety of illnesses; this makes distinguishing the cause of the fever difficult.

Most physicians have routinely hospitalized infants less than 2 months of age because of fever, regardless of the results of preliminary laboratory studies. This policy is discussed later. In one series of 169 children aged 3 months or younger, serious infections were found in 8 (36%) of 22 infants with a temperature of 40°C (104°F) or higher and in 15 (10%) of 147 infants with fever of 37.5°C to 39.9°C (100°F to 103.9°F) who were brought to an emergency room.⁴⁶

In another series of infants less than 6 months of age, the frequency of serious illness was higher in those with very high temperatures.⁴⁷ This general correlation of magnitude of fever with seriousness
of illness has been found in many studies, but additional clinical and laboratory findings can be used to develop more predictive accuracy than the fever alone. An exception to this rule is a temperature 41.1°C (106°F) or more, which is frequently associated with serious illness regardless of other findings.⁴⁸ In general, hospitalization can be individualized rather than routine for those older than 1 month old, depending on the findings and the ability and accessibility of the parents (Table 10-3).

The separation of febrile infants with early serious infections from those with benign self-limited illnesses is one of the most difficult decisions in the care of children. Every experienced pediatrician has known the feeling of not recognizing an early serious infection.

Prospective studies have been done, especially by McCarthy and colleagues, to help give guidelines for both clinical and laboratory evaluation.^{66,67,68,69,70} Unfortunately, rapid and unexpected change is typical of illness in young children. No laboratory tests or clinical observations detect all of the children who progress to a serious illness, so that follow-up clinical observations by the physician and parents are essential.

What observations help the parent or physician detect a worsening course? In studies directed at this question, the child’s playfulness and eye contact were helpful guides.^69,70 Other useful criteria were the degree of alertness and of consolability, essential observations needed to determine potential severity. Other definable components include the infant’s use of its eyes to observe people, spontaneous arm and leg movement, appropriate smiling or crying, playing, sucking, reaching, and vocalizing. Generally, patients as a group can be divided into those who are at high risk and those who are at low risk of bacteremia or invasive disease. However, this ability does not extend to the individual patient level; in other words, some children initially designated as being at low risk for serious infection prove to develop serious infection, and some children initially thought to be at high risk prove to have trivial self-limiting illnesses.

Studies of outpatients in the early 1970s showed that blood cultures were positive in a small percentage of children with fever, particularly those under 2 years of age with higher fever (104°F, 40°C) or leukocytosis (greater than 20,000/L).⁶³ S. pneumoniae and H. influenzae were the most frequent species detected, but Group A streptococci, S. aureus, Salmonella, and meningococci were also detected, along with a number of skin contaminants. One emergency room study reported that 25% of the positive blood cultures were attributable to contaminants.⁶⁴ In the newer study mentioned previously, however, only 2.1% of positive blood cultures were thought to be secondary to contamination.⁵⁷ Using continuously monitored blood culture systems such as BACTEC may be helpful in differentiating true pathogens from contaminants: The mean time to positivity in patients with true pathogens was 15 hours; it was 31 hours in those with contaminants.⁵⁷

In other studies, some of the children did not seem seriously ill or had seemingly trivial illnesses, but a high fever (140°F, 40°C) or a high leukocyte count (20,000/μL) was clearly a risk factor.

These studies led to the definition and recognition
of occult bacteremia. Many pediatricians in private practice have been skeptical that outpatient blood culture is superior to careful follow-up, especially because many bacteremic children do not develop serious illness rapidly or without clinical findings on follow-up examination. In one study of 4,151 visits to private practices, with 145 children aged 3–24 months deemed “at risk” for occult bacteremia, blood cultures, sedimentation rates, and WBC counts were seldom used, but hospitalizations were rare, and no deaths or notable sequelae were encountered.⁸⁸

Bacteremia with H. influenzae is certainly a potentially serious disease. One study of 69 children with culture-proven Hib bacteremia found that on follow-up, 36 (52%) of the children were still clinically ill; 17 developed meningitis, 5 got pneumonia, 3 had epiglottitis, 5 had cellulitis, 3 had septic arthritis, and 3 were persistently febrile. Even the patients who felt well at follow-up were apparently not safe; 3 (9%) of 33 were still bacteremic despite being afebrile, and 5 (15%) developed a secondary focus of infection.⁸⁹ This is in sharp contrast to the 96% spontaneous cure rate observed with pneumococcal bacteremia in the retrospective emergency department study cited earlier.⁵⁷

The criteria have not been entirely clarified for the use of blood cultures in the office of the private pediatrician or the walk-in clinic, which may deal with a different patient population. One of the best analyses of this dilemma deserves quotation: “When good observation or a well-functioning relationship with the medical system does not exist, the blood culture may serve an ‘administrative’ function by bringing to attention those infants in need of prompt recall and reevaluation.”⁹⁰

For some physicians, “obtaining a blood culture binds anxiety” (by giving the feeling that one is ‘doing something’), but it is not by itself therapeutic for the infant.”⁹⁰ At the present time, most investigators emphasize that blood cultures obtained selectively, using some clinical and laboratory criteria, do result in the detection of infants at higher risk of developing serious disease. However, it has also been noted that the same criteria for obtaining blood cultures can be applied for using outpatient antibiotic therapy until culture results are known.⁹¹ The efficacy of such treatment is unknown, and the patient must still be followed carefully. The rate of false-positive blood cultures (2% to 40%) and the cost of cultures also make them only another test, the value of which depends on the clinician’s judgment and experience with its use. Sloppy, injudicious, and unwise use and interpretation of blood cultures in the setting of fever can lead to enormous costs, both financial and psychological. In one extreme example, 41% of positive blood cultures contained only contaminants; despite this, phone calls, return visits, extra diagnostic procedures, and hospital admissions resulted, amassing a cost of $78,904, or $642 for every true pathogen isolated.⁹² It should be noted that, in most centers, the rate of false positive blood cultures is less than 5%.

Antibiotics are usually not indicated in children over 2 years of age with fever without localizing signs, because most of these illnesses are viral. If an antibiotic is used and fever persists, there may be continued confusion and changing of antibiotics. In addition, antibiotics may disguise localizing signs of infection and sometimes allow unrecognized progression of tissue damage. Antibiotics do not reduce bacterial complications of acute viral diseases. In addition, antibiotics can be a cause of persistent fever and rarely may have serious, even life-threatening, toxicities.

Hospitalization is more likely to be indicated if the child is very young or appears “toxic” or if the parents are too anxious or too unreliable to be alert for changes in the general appearance as described earlier. Symptoms of dyspnea, somnolence, or areas of pain or tenderness also call for hospitalization.

Hospitalization need not be done routinely for the infant less than 2 months of age, as iatrogenic risks and financial costs argue against routine admission.^45,101 Infants who are deemed to be at higher risk for severe infection because of prematurity, poor growth, or other parameters are routinely hospitalized. Clinical and laboratory criteria (Table 10-3) for identifying babies over 1 month of age who are at low risk for serious bacterial infection have been identified, studied, and even established as the standard of care in some hospital emergency departments.^102,103 The Rochester criteria have a greater focus on the babies’ health history, whereas the Philadelphia criteria place more weight on clinical appearance and laboratory findings (Table 10-3).

Most pediatricians routinely hospitalize babies who develop fever during the first 28 days of life, probably because of a perception that the risk of severe infection is higher in these babies. Strictly speaking, this perception is true. There is no magic shift in immune function that occurs at the age of 29 days, however, and children between the ages of 28 days and 2 months with high fever remain candidates, on a case-by-case basis, for hospitalization. Some authors are attempting to apply the approach of screening with selective treatment even to babies less than 28 days of age. In one study, only one (0.8%) of 131 febrile neonates younger than 28 days who appeared well, had no focal physical examination findings, had a peripheral WBC count of between 5,000 and 15,000/μL, a neutrophil band form count of less than 5,000/μL, a spun urine specimen that had less than 10 WBCs per high-power field, and a C-reactive protein of less than 20 mg/dL had a bacterial infection.¹⁰⁴ Further study along these lines will be required before the standard of care is altered. Most experts agree that babies less than 28 days old with fever should undergo a complete work-up for sepsis, be hospitalized, and be given intravenous antibiotics expectantly. Treatment is generally ampicillin plus either gentamicin or cefotaxime. If the reason for hospitalization is related more to parents’ ability to observe or return, then observation without antibiotics is satisfactory.

The frequencies of the various possible causes of prolonged undiagnosed fever are considerably different in children than in adults, and children are not included in most reviews of fever in adults. There are only a few reviews to consult for the possible causes of prolonged fever in children.^{106,107,108,109} The possible causes are described following and are listed in Box 10-1.^{106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122}

The text follows the sequence of the table. For completeness, some very rare possibilities are listed, which often are based on a few case reports. It may seem difficult to see how some of these possibilities were not diagnosed earlier, but retrospective analysis always seems simpler.