Fever of unknown origin: is there a role for empiric therapy?

Empiric therapy has little or no role to play in cases of classic fever of unknown origin with three important exceptions: cases that meet criteria for culture-negative endocarditis; cases in which findings or the clinical setting suggests cryptic disseminated tuberculosis (or, occasionally, other granulomatous infections); and cases in which temporal arteritis with vision loss is suspected. Several studies indicate that patients with prolonged, undiagnosed fever of unknown origin generally have a favorable prognosis. A small and largely anecdotal literature suggests a small role for symptomatic use of corticosteroids or nonsteroidal anti-inflammatory agents in highly selected cases.