Because many conditions may cause fever, an extensive discussion about each condition is beyond the scope of this chapter. However, classifying conditions associated with fever into broad categories is useful: (1) infection, (2) autoimmune disease, (3) neoplastic disease, (4) metabolic disease (eg, hyperthyroidism), (5) chronic inflammatory disease, (6) hematologic disease (eg, sickle cell disease, transfusion reaction), (7) drug fever and immunization reaction, (8) poisoning (eg, aspirin, atropine), (9) central nervous system abnormalities, and (10) factitious fever. In addition, dehydration, excessive muscle activity, and heat exposure may cause hyperthermia.

Although any disease in these categories may cause fever at any age, some diseases are more likely to occur at some ages than at others. Autoimmune disease and inflammatory bowel disease, for example, are unusual in infants but become progressively more frequent with increasing age. Similarly, febrile immunization reactions are much more common during the first year of life when most immunizations are administered.

Infections affecting the respiratory and gastrointestinal tracts account for the majority of fevers in all age groups. Most of these infections have a viral origin (eg, enterovirus, influenza virus, parainfluenza virus, respiratory syncytial virus, adenovirus, rhinovirus, rotavirus) and are generally self-limited. Knowledge of the seasonality of these viruses promotes correct and efficient diagnoses. In addition, knowledge of the typical physical findings in these infections and their course may help distinguish them from bacterial diseases. For example, high fever, irritability, posterior cervical adenopathy, and painful vesicles on the gums and tongue are characteristic of herpes gingivostomatitis. Failure to examine the tongue and gums may result in an unnecessary work-up in search of a possible bacterial infection. On the other hand, assuming that a high fever in a 2-month-old child is from roseola (exanthem subitum) would be erroneous because this infection (human herpesvirus type 6) usually does not occur at such an early age.

Failure to evaluate the fever further might result in missing a serious bacterial infection. Although viral infections may cause significant morbidity and mortality, the more aggressive course and serious outcomes of bacterial infections make early diagnosis especially important, particularly because effective antibiotic treatment is usually available. Bacterial infections may be especially devastating in younger children who are relatively immunocompromised because of their immature immune systems. An infection that remains localized in the older child may disseminate rapidly in the infant and toddler, particularly to the blood (bacteremia), the lungs (pneumonia), the meninges (meningitis), the bones (osteomyelitis), and the joints (arthritis). Because these infections may be seriously debilitating or even fatal if not recognized, the physician must be able to differentiate bacterial infections from the more benign viral infections.

The younger the child is, the more difficult it is to recognize bacterial infection. Complaints cannot be verbalized, and physical signs and symptoms are more subtle and easily missed unless a high index of suspicion is maintained. Serious bacterial disease is especially difficult to diagnose in children with no obvious focus of infection. For this reason, many attempts have been made during the last 20 years to identify children in whom fever is a sign of a serious bacterial infection,[7] particularly pneumococcal disease and infections caused by Haemophilus influenzae type B (HiB). Children between birth and 36 months of age have been of special interest because fever is most common in this age group, and they may be difficult to assess, particularly during the first 6 months of life. Efforts to improve the ability to diagnose a serious bacterial infection have focused on 3 areas: (1) data from the history and physical examination,[8] [9] (2) laboratory data,[10] and (3) response to antipyretics.[11] Of the 3 areas, the response to antipyretics has been shown most clearly to be unhelpful in distinguishing between patients who have a serious bacterial infection and those who have a more benign viral infection.[11] Children who have a serious infection respond to antipyretics no differently from those whose illness is less significant. In fact, some children who have viral illnesses do not defervesce either.

Many studies have attempted to delineate the precise combination of clinical or laboratory variables that might identify the febrile child at risk for serious disease. Defined clinical observational scales (eg, Yale Observation Scale, Young Infant Observation Scale, Severity Index) are not sufficiently discriminatory and predictive to be used alone.[12] [13] Laboratory studies continue to be necessary as well.

During the early 1990s, specific practice guidelines were published to facilitate the initial management of febrile infants and children without an obvious source of infection.[14] Although these guidelines remain controversial,[15] [16] [17] [18] [19] as many as one third of primary care physicians have found them to be helpful and have changed the way they evaluate young children with fever.[20] Nevertheless, each patient continues to require individual assessment, with application of the recommendations as appropriate to the individual context of the patient. Considerations of the inconvenience, discomfort, and cost of laboratory testing and the increasing resistance to antibiotics in the community must be weighed carefully against the risk of missing a serious bacterial infection, with its subsequent morbidity and mortality. Therefore physicians must make the best decisions possible in an environment of incomplete certainty about the presence of serious disease. Parents need to be part of these discussions, and adequate follow-up of all patients is crucial, no matter what is decided in the initial visit.

Although the early practice guidelines were helpful during the 1990s, they were formulated before the introduction, in 2001, of the heptavalent pneumococcal vaccine for infants. This vaccine provides protection against pneumococcal serotypes 4, 6B, 9V, 14, 18C, 19F, and 23F and is administered at 2, 4, and 6 months and between 12 and 15 months of age. Since the introduction of the vaccine, a decline of 60% to 80% in pneumococcal disease has occurred in children younger than 24 months.[21] Because occult pneumococcal bacteremia and other pneumococcal infections made up most of the serious bacterial infections in young children with high fever (>102°F [>39°C]) before the vaccine, the use of this vaccine has greatly lowered the incidence of serious bacterial infections in children at greatest risk—those between 2 to 3 months and 3 years of age. A similar impact occurred when the HiB vaccine was introduced in the 1980s, nearly eliminating HiB meningitis, epiglottitis, and bacteremia. Given the marked decrease in pneumococcal and HiB serious bacterial infections, the likelihood of a serious bacterial infection when high fever (>102°F [>39°C]) is present in infants and toddlers is now even smaller, and a fairly limited assessment may be more suitable at this time.[22] In addition, with an increased ability to diagnose specific viral illnesses by rapid diagnostic testing (respiratory syncytial virus [RSV], influenza, enterovirus), even the revised, updated practice guidelines[23] (Figure 181-1 and Figure 181-2) should be customized to the needs of the individual patient until further studies of their utility can be done.

Fever during the first 4 days of life has been associated with a high incidence of bacterial disease.[24] A temperature above 98.6°F (37°C) occurs in 1% of all newborns; of these children, 10% have a bacterial infection, usually caused by group B streptococcal or gram-negative enteric pathogens. A full work-up is indicated in these children, including a complete blood count and differential count, a urine analysis, and cultures of the blood, urine, and cerebrospinal fluid; antibiotics (usually intravenous ampicillin and gentamicin) should be administered until the results of cultures are known.

Similarly, neonates up to 28 days of age with fever have a significant risk of a bacterial infection (approximately 12% in some studies[15] [25]). Pneumococcal infection is uncommon; group B Streptococcus, Escherichia coli, and other enteric pathogens are more usual. Urinary tract infection (UTI) and occult bacteremias are the most common types of infection; however, with group B Streptococcus infection, the risk of accompanying meningitis is as high as 39%.[26] Low-risk criteria, such as the Rochester criteria, may not be consistently reliable to differentiate young patients with serious bacterial infection from those who have more benign disease. Although some studies have demonstrated only a 0.2% incidence of bacteremia or meningitis in neonates satisfying the low-risk criteria,[27] [28] others have found that up to 6% of neonates who satisfy low-risk criteria have a serious bacterial infection.[29] [30] Of note is that neonates with RSV infection do not have a lower incidence of serious bacterial infection when they have fever.[31] Concomitant UTIs are especially common, occurring in 5% to 7% of patients.[32]

Fever greater than or equal to 100.4°F (38.0°C) in infants between 28 and 60 days of age is associated with a 5% to 10% incidence of serious bacterial infection.[15] [29] [33] Unfortunately, neither height of fever nor apparent degree of toxicity has been a reliable predictor by itself of bacteremia or serious bacterial infection.[34] [35] Instead of using single predictors, a combination of clinical and laboratory criteria appears to be more useful in identifying infants who are at low risk for having a bacterial infection. The most well known of these combinations are the Rochester criteria[36] (Table 181-1). The infants must satisfy all of the following conditions: previously healthy (as defined in Table 181-1), no clinical signs of toxicity (in some studies[29] defined by an infant observation score of ≤10), no focal bacterial infection found at physical examination, a white blood cell (WBC) count of 5000 to 15,000 cells/mm[3] with 1500 bands or fewer, a normal urinalysis (≤5 WBCs per high-power field [HPF] with few or no bacteria found in centrifuged urine and a Gram-stained smear of stool demonstrating fewer than 5 WBCs/HPF if diarrhea is present. If cerebrospinal fluid is obtained, then the cell count should be 8 WBCs/HPF or fewer.[36] One- to 2-month-old infants who satisfy these criteria have only a 1.1% probability of having a serious bacterial infection, and a 0.5% probability of having meningitis.[37]

Because of the difficulty in determining, based solely on the degree of fever, whether an infant younger than 2 to 3 months is at a low or high risk for bacterial disease (septicemia has occurred even in infants who have low-grade fevers[38]), evaluation should be prompt and thorough whenever a fever of at least 100.4°F (38°C) exists, paying particular attention to obtaining the data necessary for classifying the child as low or high risk. Such a comprehensive evaluation should generally include a complete physical examination, total and differential WBC count, urinalysis[39] and urine culture, a Gram-stained smear of stool if diarrhea is present, blood culture, and possibly examination and culture of cerebrospinal fluid. A urine culture is especially important because UTIs are the most common bacterial infections in this age group, even in the absence of pyuria.[27] [40] [41]

If the infant appears nontoxic and meets the low risk criteria, then examination and culture of the cerebrospinal fluid and blood might reasonably be avoided as long as good observation and follow-up can be made within 24 hours and antibiotics are not administered. If antibiotics are to be administered, then a full work-up, including blood and cerebrospinal fluid cultures, should always be performed.

After obtaining a thorough history, including queries about illness of a similar nature in other family members and queries about whether the child has been immunized with the HiB and pneumococcal vaccines, the physician should assess the child for toxicity. If the child appears toxic (eg, lethargic or irritable, noninteractive, poor perfusion), then hospitalization should be considered along with further diagnostic tests to assess for serious bacterial infection. If the child does not appear toxic, a WBC count should be considered; if this count is greater than 15,000/mm[3], then a blood culture should be considered. (In the pre–pneumococcal vaccine era, children with WBC counts >15,000/mm[3] were 5 times as likely to experience bacteremia as those who had a WBC count <15,000/mm³.[17] In addition, an absolute neutrophil count of at least 10,000/mm³ correlated with an increased [8.2%] risk of pneumococcal bacteremia.) Practically, obtaining the WBC and blood culture at the same time is easiest, with the blood sent for culture only if the WBC count warrants doing so. Procalcitonin and C-reactive protein blood levels might have better sensitivity and specificity than the WBC count in predicting serious bacterial infection, but findings from various studies still vary widely with respect to the best cut-off levels to use.[42]

Given the lower incidence of pneumococcal disease now, avoiding blood tests altogether might be more cost-effective[43] and reasonable as long as the child has received at least 3 doses of the HiB and pneumococcal vaccines, does not appear toxic, has no obvious focus of infection, and has reliable health care providers with excellent follow-up capabilities.

Approximately 5% to 8% of children in the 3- to 36-month-old age group who have an undifferentiated febrile illness have a urinary tract infection (UTI).[44] Two groups of patients in this age group are especially at risk. Female infants with temperatures greater than 39°C (102.2°F) have a urinary tract infection incidence of 16% to 17%.[39] [44] Uncircumcised boys in the first 12 months of life have an 8- to 9-fold higher rate of UTI than circumcised boys.[45] Because of the high rate of UTIs in this age group, a urine culture is suggested for febrile boys younger than 6 months of age (<12 months if uncircumcised) and girls younger than 12 to 24 months.[23] A urinalysis alone is not adequate as a screening tool to determine which child should have a urine culture; 20% of children who have a UTI have a normal urinalysis, including a negative test for urinary nitrites or leukocyte esterase.[41] A chest radiograph is generally necessary only if clinical symptoms or signs suggest pneumonia (eg, cough, tachypnea, dyspnea, rales, decreased breath sounds, dullness to percussion).[8] However, at least one study has suggested that up to 20% of children with fever of at least 102.2°F (39°C) and a WBC count of more than 20,000/mm³ have pneumonia by chest radiograph, even in the absence of respiratory symptoms and signs.[46] Once the infant or child has been evaluated, a plan of management similar to those in Figure 181-1 and Figure 181-2 should be considered. As discussed earlier in this chapter, the use of these protocols is controversial.

A further consideration in the approach to a febrile infant or child in the first 3 years of life is the increased availability of rapid diagnostic viral testing. Rapid tests are now available for influenza A and B, RSV, and enterovirus. Although sensitivity and specificity vary with individual tests, a positive test may be helpful in decreasing the number of other tests that need to be performed to rule out a bacterial infection.[47] Except for neonates younger than 28 days, the rate of serious bacterial infections in febrile patients is lower if they are infected with influenza and RSV. When this rate of infection is coupled with a generally lower incidence of serious pneumococcal and H influenzae infections because of the advent of vaccines given at a young age, a reasonable strategy might be to use positive viral tests as a way to reduce blood and urine tests in vaccinated children older than 2 to 3 months who do not appear toxic.

Children older than 3 years are more likely to have signs and symptoms consistent with a recognizable illness. If they have nonspecific symptoms, an urgent consultation with a physician is probably unnecessary; however, regardless of age, all febrile children with localized signs and symptoms, such as swollen joints, meningismus, labored respirations, chest pain, dysuria, petechiae, alteration of consciousness, and severe abdominal pain, should be examined immediately.

Although many febrile children do not have signs and symptoms pointing to an obvious cause, a complete physical examination may reveal important clues to the origin of the fever. Because most infections involve the respiratory tract, this area must be examined carefully. In all instances, the tympanic membranes should be examined for otitis media, the pharynx for pharyngitis, the nose for the discharge of sinusitis or a viral upper respiratory tract infection, and the lungs for evidence of pneumonia or bronchiolitis. Conjunctivitis may be a clue to adenovirus, influenza or RSV infection, conjunctivitis-otitis syndrome, or Kawasaki disease.

The skin is no less important and may demonstrate typical viral exanthems, such as those associated with rubella, roseola, or chickenpox, or it may show the erythema marginatum of rheumatic fever or the rose spots of typhoid fever.

Generalized lymphadenopathy often occurs with viral illnesses, such as infectious mononucleosis, hepatitis, or cytomegalovirus infection, but it also may be a clue to the diagnosis of leukemia or lymphoma. Localized enlargement of lymph nodes should prompt a search for a skin infection or for a tumor. Isolated cervical lymphadenopathy may be associated with tuberculosis infection or cat-scratch disease (Bartonella infection).

The musculoskeletal system must be examined with care. Localized bone tenderness may suggest osteomyelitis, and a restricted range of motion in a warm joint may suggest arthritis. Although the latter finding may occur in many different diseases, a meticulous examination of the heart is always indicated to detect the carditis of rheumatic fever or infective endocarditis. The spine should be palpated for any evidence of diskitis, and any costovertebral angle tenderness should prompt an examination of the urine for evidence of a UTI.

Although uncommon, factitious fever is a final consideration and a well-described entity. Children as young as 8 years have been known to increase the thermometer reading artificially by rubbing the mercury thermometer bulb on the sheets or by exposing it to warm liquids. Clues at physical examination include a pulse that is not correlated with the increase in temperature, inability to document fever when it is measured rectally, and an absence of sweating during defervescence. Investigation of psychosocial disturbances within the family is usually necessary.

A discussion of the physiology and management of fever is provided in Chapter 53, The Ill Child.

Figure 181-1 Algorithm for the management of a previously healthy infant 28 to 90 days of age with fever without source at least 100.4°F (38°C). (Baraff LJ. Management of fever without source in infants and children. Ann Emerg Med. 2000;36:602-614. Copyright © 2000, Elsevier, with permission.)

Figure 181-2 Algorithm for the management of a previously healthy child 91 days to 36 months of age with fever without source. (Baraff LJ. Management of fever without source in infants and children. Ann Emerg Med. 2000;36:602-614. Copyright © 2000, Elsevier, with permission.)

1	Infant appears generally well
2	Infant has been previously healthy Born at term (≥37 weeks' gestation) Did not receive perinatal antimicrobial therapy Was not treated for unexplained hyperbilirubinemia Had not received and was not receiving antimicrobial agents Had not been previously hospitalized Was not hospitalized longer than mother
3	No evidence of skin, soft tissue, bone, joint, or ear infection
4	Laboratory values: Peripheral blood WBC count 5.0 to 15.0 × 10[9] cells/L (5000 to 15,000/mm[3]) Absolute band form count ≤1.5 × 10[9] cells/L (≤1500/mm[3]) ≤10 WBC per high-power field (×40) on microscopic examination of a spun urine sediment ≤5 WBC per high-power field (×40) on microscopic examination of a stool smear (only for infants with diarrhea)

From Jaskiewicz JA, McCarthy CA, Richardson AC, et al, and Febrile Infant Collaborative Study Groups. Febrile infants at low risk for serious bacterial infection—an appraisal of the Rochester criteria and implications for management. Pediatrics. 1994;94:390-396.