Tailored Cancer Outcome Prediction and Informed Consent

Clinically relevant disparities in the outcomes of cancer treatment between institutions are well established. Some authors have argued that physicians have an ethical obligation to disclose these disparities as part of the informed consent process.1 We believe that preservation of patient autonomy requires disclosure of treatment institution-specific outcomes and that statistical prediction models tailored to individual patients are the best way to frame this discussion. We describe a utopian system to gather and disseminate cancer outcome data based on the United Network for Organ Sharing and articulate why accurate and equitable prediction models are feasible both scientifically and logistically. We also discuss strategies of design and oversight required to mitigate any unintended negative downstream consequences of such a system.