OBJECTIVES: To examine the impact on costs and outcomes that may occur in neutropenic patients when treating for suspected invasive fungal infections (IFIs) caused by Aspergillus with typical empirical approach (EA) versus the recently proposed “diagnostic-driven” (DD) approach in China. METHODS: A decision-analytic model was used to estimate total costs and predicted survival associated with EA and DD approaches in Shanghai, China. The population included patients aged >= 18 years with hematological malignancies or autologous/allogeneic stem cell transplantation expected to be neutropenic for >=10 days, and without prophylactic antifungal treatment. Rates of IFI incidence, IFI captured by EA, overall mortality, and IFI-related mortality (10.9%, 30%, 10.7% and 28.6%, respectively) were obtained from the literature. Survival rates for each strategy were generated based on the proportion of patients with identified and appropriately treated IFI. Treatment patterns with EA and DD approaches and resource use assumptions were based on the opinion of five clinicians from three top hospitals in Shanghai. The total medical costs (in 2014 Chinese Yuan) included antifungal drug cost, treatment-related adverse events cost, and cost of other medical resources. City-specific costing sources were used wherever possible. RESULTS: Both approaches had similar survival rates (90.76% vs. 91.33% for EA and DD, respectively). Antifungal drug cost per patient was ¥2,813 for EA and ¥2,307 for DD strategy. Although DD patients incurred a higher cost on PCR/GM testing (¥111 vs ¥88), the total medical costs of DD were substantially lower (¥2,563) than that of EA strategy (¥4,298) due to fewer patients receiving antifungal agents (DD: 7.4%; EA: 12.5%) with targeted IFI treatment.     CONCLUSIONS: This study suggests that the DD approach has the potential to initiate antifungal treatment in a more targeted population. It is expected to be a cost saving management strategy for immunocompromised patients with suspected IFI in the context of China.