Dr. Becker raises two issues that were not explicitly addressed in our editorial. First, if the result of an induced-sputum examination is negative for P. carinii, what diagnosis is likely? Second, if the result of an induced-sputum examination is negative for P. carinii, what diagnostic procedures should be used?

Regarding the first issue, we believe that at many institutions, induced sputum examination has a high sensitivity for P. carinii pneumonia in patients who are and are not receiving prophylaxis for P. carinii pneumonia (sensitivity, >60% and >90%, respectively) [1]. However, we do not agree that a negative result of induced-sputum examination in this setting makes other pulmonary diagnoses substantially more likely than a diagnosis of P. carinii pneumonia. Huang and colleagues [2] reported that in patients who have HIV infection, low CD4 counts, a clinical picture typical of P. carinii pneumonia, and a negative result of an induced-sputum examination, the most likely diagnosis is P. carinii pneumonia. In that study [2] (which was done at an institution with considerable experience assessing induced-sputum samples), 192 of 602 (31%) patients with negative results of induced-sputum examinations were found to have P. carinii pneumonia at bronchoscopy. This diagnosis was substantially more common than that of M. tuberculosis infection (<5%) or fungal infection (<5%).

The second issue raised by Dr. Becker is what procedures should be done to establish the diagnosis of pulmonary disease when the result of an induced-sputum examination is negative. Depending on the reliability of the laboratory and the quality of the specimen, a second induced-sputum sample might be useful. Most clinicians would do bronchoalveolar lavage after the initial induced-sputum analysis and consider the merit of transbronchial biopsy during the initial bronchoscopy. Transbronchial lung biopsy is associated with a slightly increased sensitivity of the bronchoscopic procedure for P. carinii pneumonia and may either enhance the sensitivity or be required to establish a diagnosis of tuberculosis, cytomegalovirus pneumonia, fungal pneumonia, or such noninfectious processes as lymphocytic interstitial pneumonitis or some pulmonary cancers [3, 4]. It is reasonable, as Dr. Becker has done, to advocate a diagnostic procedure that includes bronchoscopy, bronchoalveolar lavage, and transbronchial lung biopsy during the initial bronchoscopic procedure for patients who present with a clinical picture that suggests P. carinii pneumonia and a negative result of induced-sputum examination. However, our general practice is to do only bronchoscopy and bronchoalveolar lavage. If those procedures fail to indicate a diagnosis, we repeat them, adding the transbronchial biopsy.

In geographic areas where tuberculosis, histoplasmosis, or coccidiomycosis is especially common or in health care settings where induced-sputum evaluation is insensitive, a different algorithm encouraging bronchoscopy with or without transbronchial biopsy may be indicated.