Real-World Impact of a Decision Support Tool on Colony-Stimulating Factor Use and Chemotherapy-Induced Febrile Neutropenia Among Patients With Breast Cancer.

Background: White blood cell colony-stimulating factors (CSFs) decrease the incidence of chemotherapy-induced febrile neutropenia (FN). Widespread use of CSFs that is not guideline-concordant has been reported. Among patients with breast cancer receiving chemotherapy, the ability of evidence-based decision support tools to promote risk-appropriate reductions in CSF use without increased incidence of FN has not been examined. Methods: A retrospective cohort design and US commercial claims data were used. The impact of CSF decision support was analyzed among women with breast cancer receiving first-cycle chemotherapy from April 1, 2013, to March 30, 2015. The tool was implemented as part of a prior authorization process in 9 states starting July 1, 2014. Patients were assigned to intervention (ie, states where the decision support tool had been implemented) or nonintervention states (ie, 39 states where the tool had not been implemented). CSF use and subsequent incidence of FN were compared using difference-in-difference (DID) regressions adjusting for baseline differences in FN risk factors such as comorbidities and various infections. Results: The study sample of 7,224 patients (intervention states: pre-implementation, 1,991 and post-implementation, 2,010; nonintervention states: pre-implementation, 1,569 and post-implementation, 1,654) showed no significant difference in risk factors. Before and after implementation, a significant decrease in the proportion of patients with CSF use was observed in the intervention states (75% to 69%) compared with no significant change in the nonintervention (72% to 71%) states (DID, -5.4%; 95% CI, -6.0% to -4.7%; P =.006). No significance increase in FN incidence occurred in intervention (5.0% to 5.5%) and nonintervention (5.4% to 4.8%) states (DID, 0.2%; 95% CI, -0.20 to 0.30; P =.78). Similar results were obtained in subgroups by comorbidities and in sensitivity analyses by claims-based FN definitions. Conclusions: CSF use decreased modestly after implementation of the decision support tool, with no observed changes in FN rates. Such tools can reduce practice variation to improve care standards.