Clinical and radiographic predictors of the etiology of pulmonary nodules in HIV-infected patients.
STUDY OBJECTIVES: To determine the etiology and the clinical and radiographic predictors of the etiology of pulmonary nodules in a group of HIV-infected patients. DESIGN: Retrospective analysis. SETTING: A large urban hospital in San Francisco, CA. PATIENTS: HIV-infected patients evaluated at San Francisco General Hospital from June 1, 1993, through December 31, 1997, having one or more pulmonary nodules on chest CT. MAIN OUTCOME MEASURES: Three physicians reviewed medical records for clinical data and final diagnoses. Three chest radiologists blinded to clinical data reviewed chest CTs. Univariate and multivariate analyses were performed to determine clinical and radiographic predictors of having an opportunistic infection and the specific diagnoses of bacterial pneumonia and tuberculosis. RESULTS: Eighty seven of 242 patients (36%) had one or more pulmonary nodules on chest CT. Among these 87 patients, opportunistic infections were the underlying etiology in 57 patients; bacterial pneumonia (30 patients) and tuberculosis (14 patients) were the most common infections identified. Multivariate analysis identified fever, cough, and size of nodules < 1 cm on chest CT as independent predictors of having an opportunistic infection. Furthermore, a history of bacterial pneumonia, symptoms for 1 to 7 days, and size of nodules < 1 cm on CT independently predicted a diagnosis of bacterial pneumonia; a history of homelessness, weight loss, and lymphadenopathy on CT independently predicted a diagnosis of tuberculosis. CONCLUSIONS: In HIV-infected patients having one or more pulmonary nodules on chest CT scan, opportunistic infections are the most common cause. Specific clinical and radiographic features can suggest particular opportunistic infections.
Jasmer, RM; Edinburgh, KJ; Thompson, A; Gotway, MB; Creasman, JM; Webb, WR; Huang, L
Volume / Issue
Start / End Page
International Standard Serial Number (ISSN)
Digital Object Identifier (DOI)