Impact of clinical data on chest radiography sensitivity in detecting pulmonary abnormalities in immunocompromised patients with suspected pneumonia

Purpose Chest radiography (CXR) of immunocompromised patients has low sensitivity in the early evaluation of pulmonary abnormalities suspected to be infectious. The purpose of the study was to evaluate whether the knowledge of clinical data improves the diagnostic sensitivity of CXR in the particular setting of immunocompromised patients after hematopoietic stem cell transplantation (HSCT). Materials and methods Sixty-four CXRs of immunocompromised patients with clinically suspected pneumonia were retrospectively and independently evaluated by two radiologists to assess the presence of radiological signs of pneumonia, before (first reading) and after (second reading) the knowledge of clinical data. A chest computed tomography (CT) performed within 3 days was assumed as the standard of reference. For each reading, sensitivity of both radiologists was calculated. Results Readers showed a sensitivity of 39% and 58.5% for the first reading, and 43.9% and 41.5% for the second reading, respectively. For both readers, these values were not significantly different from those obtained at first reading (McNemar’s test, p >0.05). Interobserver agreement at second reading was fair (Cohen test, k=0.33). Conclusions The sensitivity of CXR is too low to consider it a stand-alone technique for the evaluation of immunocompromised patients after HSCT with suspected pneumonia, even if the radiologist knows detailed clinical data. For these patients, an early chest CT evaluation is therefore recommended.