AEGiS-11IAC: Predicting Pneumocystis carinii pneumonia: results of a multivariate analysis.

11th International AIDS Conference

Vancouver, British Columbia — July 7-12, 1996

Predicting Pneumocystis carinii pneumonia: results of a multivariate analysis.

Int Conf AIDS 1996 Jul 7-12; 11:229 (abstract no. Tu.B.110)
Huang L, Hecht FM, Gruden JF, Kearns K, Turner J, Stansell JD, Hopewell PC; San Francisco General Hospital, San Francisco, CA, USA. Fax: 415-695-1551. E-mail: LHUANG@ITSA.UCSF.EDU.

Background/

OBJECTIVES: PCP remains the most frequent AIDS-defining opportunistic infection in the U.S. Definitive diagnostic tests are expensive and, in the case of bronchoscopy, invasive. Selection of appropriate patients for diagnostic testing relies on clinician experience in evaluating clinical variables. This study determined the association of clinical variables with PCP using bivariate analysis and the independent predictive power of these variables for PCP using a multiple logistic regression model.

METHODS: Prospective study of all patients with suspected PCP referred to the Division of Pulmonary and Critical Care at SFGH for diagnostic evaluation over a 5 months period. Clinical and lab variables were verified by a single researcher, chest radiographs (CXR) were graded by a chest radiologist, and patients without a microscopic diagnosis of PCP were followed for 60 days to assess for the subsequent development of disease.

RESULTS: (1) 164 patients with suspected PCP were evaluated. PCP was subsequently diagnosed in 60 (36.6%). (2) Clinical variables that were associated with PCP in a bivariate analysis are shown below: (table: see text)

RESULTS: of a multiple logistic regression model: The strongest independent predictors of PCP were a CXR with granular opacities (OR=66.9, p=0.0001), a CD4 count less than 50 cells/uL (OR=11.0,p=0.009), and absence of purulent sputum (OR=43.5, p=0.004). No patient with purulent sputum and a CXR without granular opacities had PCP.

CONCLUSIONS: (1) The strongest independent predictors of PCP are a CXR with granular opacities, a CD4 count less than 50 cells/uL, and the absence of purulent sputum. (2) The combination of purulent sputum and a CXR without granular opacities indicates a low risk for PCP; these patients should be treated for another process and do not need to undergo diagnostic testing for PCP.

Keywords: AEGIS, Pneumonia, Pneumocystis carinii, Multivariate Analysis, Acquired Immunodeficiency Syndrome, CD4 Lymphocyte Count, Logistic Models, Radiography, Thoracic, Prospective Studies, Bronchoscopy, Lung, Diagnostic Tests, Routine, Lung Diseases, United States, Human, radiography, ICA11

960707
TuB110