A clinically actionable nomogram for predicting invasive pulmonary aspergillosis: a nested case-control study

Invasive pulmonary aspergillosis (IPA) is a life-threatening infection with high mortality, often misdiagnosed due to non-specific symptoms. The lack of effective early prediction models delays treatment and worsens outcomes. To identify independent IPA risk factors and develop a clinically actionable nomogram for early prediction, we conducted a nested case-control study within a 10-year cohort of 27 100 pulmonary infection patients. The cohort was split into training (70%) and testing (30%) sets. In the training set, 1 002 IPA cases (proven or probable, as defined by EORTC/MSGERC 2020 criteria) were included, alongside 2 004 randomly selected pneumonia controls (1:2 ratio). A nomogram was developed using multivariable logistic regression in the training set and was evaluated by held-out internal validation in the testing set. Model performance was primarily assessed via area under the curve (AUC), calibration, and decision curve analysis (DCA). Nine independent IPA predictors were identified: bronchiectasis (OR = 7.07), pulmonary tuberculosis (OR = 2.20), diabetes (OR = 2.09), positive serum galactomannan (GM) test (OR = 1.76), mechanical ventilation (OR = 1.74), connective tissue disease (OR = 1.73), positive serum (1,3)-β-D-glucan (G) test (OR = 1.48), sputum (OR = 1.33), and neutrophil-to-lymphocyte ratio (OR = 1.01). The nomogram demonstrated moderate stable discrimination (training AUC: 0.73; testing AUC: 0.75), with excellent calibration (Brier scores: 0.183 and 0.110). DCA confirmed clinical utility across wide risk thresholds. In conclusion, this novel nine-variable nomogram using routine clinical data provides a practical tool for early IPA risk estimation, potentially guiding timely decisions and improving outcomes. External validation is warranted.