In 1961, Petersdorf and Beeson proposed their now classic criteria for fever of unknown origin (FUO): (1) an illness of at least 3 weeks’ duration, (2) measured temperatures greater than 38.3°C on several occasions, and (3) no etiologic diagnosis after 1 week of hospitalization.1 For adult patients, Durack and Street2 modified the third criterion of this definition to account for potential outpatient evaluation by substituting “3 outpatient visits or at least 3 days in the hospital” for “1 week of hospitalization.” Studies focusing on pediatric patients have occasionally used fever duration of 2 rather than 3 weeks. Patients presenting with 2 to 3 weeks of fever but not meeting all the criteria for FUO are often referred to as having prolonged fever. Differentiating prolonged fever from FUO can be impractical, because the diagnostic criteria for FUO do not specify the type or extent of evaluation required. Final diagnoses are determined in a number of ways (e.g. natural history, biopsy, imaging, serologic studies) and no single patient either receives or requires every diagnostic test. This chapter discusses the evaluation of a child with prolonged fever in whom the cause of fever is not readily apparent.
For purposes of this chapter, FUO means (1) fever of prolonged duration (>2 weeks), (2) documented temperature higher than 38.3°C (101°F) on multiple occasions, and (3) uncertain cause. Up to 50% of patients referred for evaluation of FUO have multiple, unrelated, self-limited infections, parental misinterpretation of normal temperature variation, or complete absence of fever at the time of referral.3-5 Therefore the initial history should include the method used to determine temperature (e.g. “felt warm” vs. actual measurement), the duration of thermometer insertion, location of insertion (tympanic membrane, oral, axillary, rectal), time of day, and confirmation by more than one person. Studies examining parents’ ability to subjectively determine the presence of fever found a sensitivity of 74% to 84% and a specificity of 76% to 91% for tactile examination.6-9 Some studies of tactile examination reveal more accurate results for higher temperatures (>38.9°C) than for lower ones. The location of measurement also affects the accuracy of the reading. The oral temperature can be 0.3°C to 0.7°C lower than a simultaneous rectal measurement.10 Oral measurements are more easily influenced by other factors such as oral liquid intake and respiratory rate. Tandberg and Sklar reported that the difference between rectal and oral measurements increased from 0.5°C to almost 1°C in patients with tachypnea.11 Axillary and tympanic measurements are at least 0.5°C lower than simultaneous oral measurements.12 To compensate for such discrepancies, parents are sometimes instructed to add 0.5°C or 1°C to oral or axillary measurements to approximate the “real” temperature. Such corrections are often inaccurate, however, because the amount of underestimation varies from person to person and with the body temperature (differences are exacerbated at higher body temperatures); these corrections may further cloud the evaluation ...