Biostatistics Journal Club: Leveraging External Data to Test Heterogeneous Treatment Effects

Novel cancer treatments may benefit some patient subgroups more than others, but these differences are often difficult to anticipate when trials are designed. As a result, a randomized clinical trial may miss a real treatment benefit if that benefit is concentrated in a smaller subgroup. Led by Boyu Ren, PhD, of McLean Hospital, this talk will examine how external data from prior clinical studies and electronic health records can help identify treatment benefits that may be limited to specific patient subgroups. It will introduce a statistical test that improves the ability of randomized clinical trials to detect these heterogeneous treatment effects while controlling false-positive findings. The method remains valid even when the external data differ from the trial population or contain unmeasured confounding. Its performance will be illustrated through simulations and a retrospective analysis of glioblastoma clinical trials.

Required Reading:

Leveraging External Data for Testing Experimental Therapies with Biomarker Interactions in Randomized Clinical Trials [PDF]

Speaker:

Boyu Ren, PhD, McLean Hospital