Detecting lung cancer as a cause of hemoptysis in patients with a normal chest radiograph: bronchoscopy vs CT.

OBJECTIVE: To determine whether fiberoptic bronchoscopy (FOB) or CT results in the lowest number of tests needed to diagnose (NTND) lung cancers in patients presenting with hemoptysis and a normal chest radiograph (CXR).
DESIGN: Calculation of the NTND in a hypothetical cohort of patients presenting with hemoptysis and a normal CXR.
INTERVENTIONS: In the primary analysis, either FOB or CT is performed to detect lung cancers. FOB is used to diagnose endobronchial abnormalities, and transthoracic needle aspirate is relied on to diagnose parenchymal findings. Patients then undergo serial follow-up CXRs. In a secondary analysis, sputum cytologic tests are performed prior to FOB and CT. Abnormal cytologic results require FOB. Unremarkable cytologic results allow a choice between FOB or CT. MEASUREMENTS: NTND and number of lung cancers detected during serial follow-up CXRs.
RESULTS: Performing FOB results in a much lower NTND than CT with a similar number of lung cancers detected during serial follow-up with each approach. Reducing the false-positive rate for lung cancers of airway evaluations by CT reduces the NTND for the CT strategy. Performing both FOB and CT results in a large NTND. Adding sputum cytology as a guide for performing FOB substantially reduces the NTND for the FOB approach.
CONCLUSION: A strategy relying on initial sputum cytologic testing as a screen for choosing either FOB as an immediate diagnostic step or serial follow-up CXR to detect lung cancer in patients presenting with hemoptysis and a normal CXR results in the lowest NTND with only a marginal reduction in the early detection of all cancers.