Caution ahead: Research challenges of a randomized controlled trial implemented to improve breast cancer treatment at safety-net hospitals

Abstract

Purpose To implement and test a Web-based tracking and feedback (T&F) tool to close referral loops and reduce adjuvant breast cancer treatment underuse in safety-net hospitals (SNHs). Patient and Methods We randomly assigned 10 SNHs, identified patients with new stage 1 to stage 3 breast cancer, assessed their connection with the oncologist, and relayed this information to surgeons for follow-up. We interviewed key informants about the tool’s usefulness. We conducted intention-to-treat and pre- and poststudy analyses to assess the T&F tool and implementation effectiveness, respectively. Results Between the study start and intervention implementation, several hospitals reorganized care delivery and 49% of patients scheduled to undergo breast cancer surgery were ineligible because they already were in contact with an oncologist. One high-volume hospital closed. Despite randomization of hospitals, intervention (INT) hospitals had fewer white patients (5% v 16%; P = .0005), and more underuse (28% v 15%; P = .002) compared with usual care (UC) hospitals. Over time, INT hospitals with poorer follow-up significantly reduced underuse compared with UC hospitals (INT hospitals, from 33% to 9%, P = .001 v UC hospitals, from 15% to 11%, P = .5). There was no difference in underuse (9% at INT hospitals, 11% at UC hospitals; P = .8). Hospitals with better follow-up (odds ratio, 0.85; 95% CI, 0.73 to 0.98) had less underuse. In settings with poor follow-up and tracking approaches, key informants found the tool useful. The rapidly changing delivery landscape posed significant challenges to this implementation research. Conclusion A T&F tool did not significantly reduce adjuvant underuse but may help reduce underuse in SNHs with poor follow-up capabilities. Inability to discern T&F effectiveness is likely due to encountered challenges that inform lessons for future implementation research.