A pragmatic cluster-randomized trial of a computerized clinical decision support system to improve colony stimulating factor prescribing for patients with cancer receiving myelosuppressive chemotherapy (SWOG S1415CD).

1525 Background: Primary prophylactic colony stimulating factors (PP-CSF) are prescribed to patients undergoing chemotherapy to reduce the risk of febrile neutropenia (FN). Prior studies have shown that 55-95% of CSF prescribing is inconsistent with practice guidelines. We conducted a cluster randomized trial to determine if guideline-informed standing orders for PP-CSF improved prescribing and reduced the incidence of FN. Methods: Patients age ≥18 with breast, colorectal or non-small cell lung cancer initiating first cancer-directed therapy with NCCN-recommended regimens were eligible. The intervention consisted of automated PP-CSF orders for high FN risk chemotherapy regimens and an alert not to use PP-CSF for low FN risk regimens. Regimen FN risk was based on NCCN guidelines. Clinicians could override the orders. Primary and secondary outcomes were PP-CSF use among patients receiving high and low risk regimens FN incidence within 6 months of initial therapy. Sample size estimates assumed an FN risk of 25% for high-risk chemotherapy. 32 NCI Community Oncology Research Program (NCORP) practices randomized 3:1 to the order entry system (intervention) versus usual care (UC) provided 90% power to detect a 50% reduction in FN at a planned enrollment of 90 patients per site. Mixed effect logistic regression models were used to test differences among randomized sites. 13 practices with pre-existing PP-CSF order sets enrolled in a parallel cohort study. Patients and other stakeholder groups informed study design, conduct and reporting. Results: Between January 2016 and April 2020, 2,946 patients were randomized (2287 intervention, 659 UC); 718 were enrolled in the cohort. Mean age across arms was 58.1. 77% of patients were female; 61% diagnosed with breast cancer. Among patients receiving high-risk regimens, PP-CSF use did not differ between arms (89.2% intervention; 95.8% UC, adjusted p = 0.21) and was similar to the cohort patients (93.0%). The FN rate for high-risk patients was 5.7% in intervention clinics and 4.2% in UC clinics (adjusted p = 0.26); FN was 14.9% among high-risk patients who did not receive PP-CSF. Among patients receiving low-risk regimens, PP-CSF use did not differ between arms (intervention 6.3%, UC 5.5%, adjusted p = 0.74) and was slightly lower than the cohort (8.3%). FN rates did not differ between low risk groups (intervention 1.5%, UC 0.8%, adjusted p = 0.51). Conclusions: Guideline-informed standing orders did not increase PP-CSF use in high-risk patients, nor did it decrease use in low-risk patients. Adherence to guidelines in both risk groups exceeded historical reports. FN rates among patients not receiving PP-CSF were substantially below those reported in CSF guidelines. Automated standing orders for PP-CSF do not appear to be helpful or necessary. Clinical trial information: NCT02728596.