Development and internal validation of a model for predicting 60-day risk of invasive mould disease in patients with haematological malignancies

Objective: Our objective was to develop a model that predicts a patient's risk of developing invasive mould disease (IMD) within 60 days of admission for treatment of a haematological malignancy. Methods: We analysed 19 risk factors for IMD in a cohort of 1944 adult patients with haematological malignancies over 4127 admissions at a haematology referral centre in Northern Italy (2007-2016). We used a multivariable logistic regression to estimate the 60-day probability of developing probable or proven IMD. The model was internally validated using a bootstrap resampling procedure. Results: The prevalence of IMD was 3.3% (90 probable cases, 43 proven cases). Seven risk factors were retained in the final risk model: (1) uncontrolled malignancy, (2) high-risk chemotherapy regimen, (3) high-dose corticosteroids, (4) severe lymphopenia, (5) CMV reactivation or disease, (6) prolonged neutropenia, and (7) a history of previous IMD within 90 days. The model displayed good calibration and discrimination in both the derivation (aROC 0.85, 95% CI 0.84-0.86) and validation (aROC 0.83 95% CI 0.79-0.89) populations. Conclusions: Our model differentiated with 85% accuracy whether or not patients developed IMD within 60-days of admission. Individualized risk assessment, aided by validated prognostic models, could assist IMD management and improve antifungal stewardship.