ABSTRACT
Background: A miliary pattern on chest imaging is often attributed to tuberculosis (TB) infection. However, a myriad of conditions can cause a miliary pattern, many of which are imminently life-threatening. Research Question: The primary aim of our study is to elucidate the potential causes of miliary chest imaging patterns to improve workup and empiric therapy selection. The secondary aims are to discern the predictors of miliary disease etiology and to determine whether appropriate empiric antimicrobial therapies were given. Study Design and Methods: In this retrospective cohort study, we searched a radiology database for patients with chest imaging studies described by the word "miliary". Subjects were excluded if they were under 18 years of age and if there were insufficient objective data to support a miliary disease etiology. A radiologist independently reviewed all imaging studies, and studies that did not appear to have a true miliary pattern were excluded. The collected data include patient demographics, immunocompromising risk factors, conditions associated with miliary disease, ß-D-glucan levels, serum eosinophil count, and empiric therapies received. Results: From our 41-patient cohort, 22 patients (53.7%) were clinically diagnosed with coccidioidomycosis, 8 (19.5%) with TB, 7 (17.1%) with metastatic solid cancer, 1 (2.4%) with lymphoma, 1 (2.4%) with other (Mycobacterium simiae), and 3 (7.3%) with unknown diseases (the sum equals 42 patients because one individual was diagnosed with both coccidioidomycosis and TB). All six patients with greater than 500 eosinophils/µL were diagnosed with coccidioidomycosis. Of the 22 patients diagnosed with coccidioidomycosis, 20 (90.91%) were empirically treated with an antifungal regimen. Of the eight patients with TB, six were empirically treated for TB. Interpretation: Based on our data from a Coccidioides-endemic region with close proximity to tuberculosis-endemic areas, the leading cause of miliary disease is coccidioidomycosis, although TB and cancer are also common etiologies. Serum eosinophilia and elevated ß-D-glucan levels were strongly predictive of coccidioidomycosis in our patient cohort with a miliary chest imaging pattern.
