Cancer trials

This branch covers trials of cancer treatments, and its members divide them by the part of the trial in question: what the trial is measured on and what each endpoint lets its report claim, how the design changes once a targeted agent or immunotherapy breaks the assumption that toxicity and efficacy rise together, and where bias enters the design and its reporting. It leaves time-to-event analysis to the topic on survival analysis, bias in studies of every kind to the topic on bias and confounding, and the biology of cancer and its targeted treatments to the branch on cancer in the scientific background.

• Endpoints in cancer trials — overall survival against the progression and response endpoints that arrive sooner, the response definitions they depend on, the molecular, pharmacodynamic and phase I endpoints, and how far a surrogate endpoint supports a claim
• Trial design for targeted therapies and immunotherapies — what changes phase by phase once toxicity and efficacy stop rising together, what the adaptive designs require of a protocol, and the patient selection and late toxicity an immunotherapy trial plans for
• Bias in the design and reporting of oncology trials — the conditions a noninferiority, single-arm or crossover design carries, how far each design weakness inflates an effect estimate, the estimand a confirmatory trial needs, and the publication bias, outcome reporting bias and spin a writer controls