Dual Factors in Chemotherapy Dose Reductions: Bridging Causal Insights with Clinical Correlates

Chemotherapy dose reductions are associated with adverse clinical outcomes in cancer patients, yet the underlying causal factors remain incompletely defined. We investigated causal determinants of chemotherapy dose reductions across three cancer-type categories (breast, colorectal, and lung/other). We compared these causal determinants with purely correlative factors identified by multivariate Logistic Regression analysis. For causal analysis, we employed the Independent Component Analysis Linear Non-Gaussian Acyclic Model (ICALiNGAM) and all causal analyses were conducted in Python within the Google Colab environment. Among 3,439 chemotherapy cycles, 322 (9%) involved dose reductions. We revealed factors demonstrating both causal and correlative significance ("dual factors"): a prior history of febrile neutropenia, cancer type, a prior history of febrile neutropenia, inpatient treatment and pre-cycle decision to use G-CSF prophylaxis. Our findings reveal distinct causal and correlative factors driving chemotherapy dose reductions, offering clearer targets for intervention to preserve optimal dose intensity.