April 1987

A Simple Index to Identify Occult Bacterial Infection in Adults With Acute Unexplained Fever

Author Affiliations

From the Department of Medicine (Drs Mellors and R. Horwitz) and Epidemiology (Dr R. Horwitz) Yale University School of Medicine, New Haven, Conn, and the Department of Medicine, University of Alabama School of Medicine, Birmingham (Dr Harvey). Dr S. Horwitz is presently with the Institution for Social and Policy Studies at Yale University.

Arch Intern Med. 1987;147(4):666-671. doi:10.1001/archinte.1987.00370040048009

• Patients with acute fever (less than three weeks' duration) and no localizing symptoms or physical findings to suggest a source (unexplained fever) may have self-limited illness or occult bacterial infection requiring prompt treatment. To develop a management strategy for patients with unexplained fever, we studied 880 adults who were evaluated for acute fever in an emergency room. At presentation, 135 (15%) patients had unexplained fever. Occult bacterial infection was found in 48 (35%) of these 135 patients, and 21(44%) of 48infected patients had bacteremia. Four bacteremic patients were incorrectly discharged from the emergency room without antimicrobial therapy. Neither a "toxic" appearance of the patient nor an initial temperature of greater than or equal to 39.4°C (103°F) were predictive of occult bacterial infection. An index of predictive features was developed that included: age 50 years or older; diabetes mellitus; a white blood cell count greater than or equal to 15 000 /mm3 (15×109/L); a neutrophil band cell count greater than or equal to 1500 /mm3 (1.5×109/L); and a Wintrobe erythrocyte sedimentation rate greater than or equal to 30 mm/h. In patients with 0,1, 2, or 3 or more index features present, the proportions having occult bacterial infection were 5% (1/21), 33% (15/45), 39% (15/38), and 55% (17/31), respectively. All four bacteremic patients incorrectly discharged had two or more of the index features. Adults presenting with acute unexplained fever often have life-threatening bacterial infection. A simple clinical index can be used to estimate the likelihood of occult infection and may reduce the frequency of diagnostic error.

(Arch Intern Med 1987;147:666-671)