Objectives: To determine the value of clinical parameters in predicting influenza in older persons and those with underlying cardiopulmonary conditions hospitalized with respiratory illnesses.
Design: Prospective evaluation of hospitalized persons from November 15, 1999, to April 15, 2000.
Setting: A 450-bed general medical-surgical hospital in Rochester, New York.
Participants: Patients aged 65 and older or those with underlying cardiopulmonary conditions admitted to the hospital with respiratory diagnoses.
Measurements: The ability to discriminate influenza from noninfluenza cases using clinical parameters. Influenza infection was documented by culture, rapid antigen detection, reverse transcription-polymerase chain reaction, or serology.
Results: Sixty-one influenza A infections (mean age +/- standard deviation = 79 +/- 10) were identified in 332 evaluated illnesses (18.3%). Fifty-six cases occurred in 168 patients (33%) during a 9-week epidemic period; 40 were considered contagious on the basis of a positive culture or antigen test. Neither single clinical parameters nor logistic regression analysis using all parameters clearly discriminated influenza from noninfluenza cases. The complex of cough, temperature of 38 degrees C or higher, and illness duration of 7 days or less provided the best discrimination between infected and uninfected subjects; during the epidemic period, 53% of subjects with this symptom complex had influenza, compared with 18% without these symptoms (relative risk = 2.99, 95% confidence interval = 1.85-4.83). This symptom complex had a sensitivity of 78% and a specificity of 73% for contagious influenza during the epidemic period.
Conclusions: In older persons and those with underlying cardiopulmonary conditions admitted with acute respiratory illness during influenza epidemics, the presence of fever, cough, and brief illness should be useful for institution of respiratory isolation and for design of antiviral trials.