TY - JOUR T1 - Diagnosis of Pneumocystis carinii pneumonia in HIV antibody positive patients by simple outpatient assessments. JF - Thorax JO - Thorax SP - 1005 LP - 1009 DO - 10.1136/thx.47.12.1005 VL - 47 IS - 12 AU - D E Smith AU - A Forbes AU - S Davies AU - S E Barton AU - B G Gazzard Y1 - 1992/12/01 UR - http://thorax.bmj.com/content/47/12/1005.abstract N2 - BACKGROUND: As increasing numbers of patients with immunosuppression induced by the human immunodeficiency virus (HIV) present with respiratory symptoms it is important to differentiate Pneumocystis carinii pneumonia from other chest diseases rapidly and start treatment early. The management of pneumocystis pneumonia could be improved if clinicians could diagnose this condition confidently on the basis of simple clinical assessments. METHODS: Three hundred and eighteen patients with evidence of immunosuppression due to HIV infection and suspected pneumocystis pneumonia were investigated. A clinical history was taken and arterial blood gas analysis, chest radiography, oximetry during exercise, and sputum induction or bronchoscopy (or both) were performed. RESULTS: Pneumocystis pneumonia was confirmed microscopically from induced sputum or bronchoalveolar lavage fluid in 154 patients; 118 had other chest disease. The remaining 46 patients had no definitive diagnosis. The best single independent predictors of a diagnosis of pneumocystis pneumonia were exercise induced oxygen desaturation and obvious interstitial infiltrates on the chest radiograph (odds ratios of 4.88 and 5.44 respectively). The symptom triad of exertional dyspnoea, cough, and fevers; the absence of pneumocystis pneumonia prophylaxis; and resting arterial hypoxaemia were less predictive (odds ratio 2.07, 3.72, and 0.69). An algorithm was developed that gave a positive predictive value for confirmed pneumocystis pneumonia of 95% and also identified those patients with a very small chance of having pneumocystis pneumonia (negative predictive value 85%). CONCLUSIONS: The diagnosis of an initial episode of pneumocystis pneumonia can be confidently made in a large proportion of immunosuppressed patients with respiratory symptoms on the basis of clinical symptoms, the absence of prophylaxis, chest radiographic appearances, and oxygen desaturation during exercise as shown by oximetry. Using these simple features clinicians can rapidly assign patients to the appropriate type of management at presentation. ER -