Effect of sputum induction on arterial oxygen saturation and spirometry in HIV-infected patients

We wanted to study the effect of sputum induction on forced expiratory volume in one second (FEV1) and on oxygen saturation in normal controls, asymptomatic human immunodeficiency virus (HIV)-seropositive individuals, and HIV-seropositive patients under investigation for suspected Pneumocystis carinii pneumonia (PCP). Over a five month period, sputum induction with ultrasonically nebulized 3% saline was performed on 110 HIV-seropositive patients with suspected PCP, 10 asymptomatic HIV-seropositive patients, and 15 normal controls. Oxygen saturation (peak, trough and change in oxygen saturation (delta O2)) was measured throughout the procedure using pulse oximetry, and these results compared with the chest radiograph and the final pulmonary diagnosis. In addition, the effect of sputum induction on FEV1 was measured in the 15 control subjects and 10 asymptomatic HIV-seropositive patients. Compared with bronchoalveolar lavage, sputum induction had a diagnostic sensitivity for PCP of 76%. Chest radiography was 79% sensitive, and had specificity of 83%. Patients with PCP had lower peak and trough oxygen saturation values compared with the non-PCP group (mean peak 95 vs 97%; mean trough 88 vs 91%), and greater falls in O2 saturation during the procedure (mean delta O2 7.6 vs 5.5%). One subject desaturated to 76%, requiring supplemental oxygen. Sputum induction caused significant but temporary falls in FEV1 both in control and HIV-seropositive groups (mean maximum fall in FEV1 10.4 vs 12.5%). We conclude that although sputum induction causes significant falls in oxygen saturation and FEV1, it remains sensitive and safe, and provides a useful alternative to bronchoscopy for the diagnosis of PCP.