The quasi-CRM shift method for partially ordered groups

Connor Celum,Bethany Jablonski Horton,Mark Conaway
DOI: https://doi.org/10.1016/j.cct.2023.107400
IF: 2.261
2024-01-01
Contemporary Clinical Trials
Abstract:This paper proposes a phase-I clinical trial design that uses ordinal toxicity to locate group-specific doses when groups are partially or completely ordered prior to the start of the trial. There has been previous work on dose-finding for groups and on dose-finding with ordinal toxicity but a solution to the problem of dose-finding for groups with ordinal toxicity has not been proposed. Simulations compared the proposed method against two methods; one that uses ordinal toxicity but does not use group information and one that uses group information but does not use ordinal toxicity. One issue with the first method is the potential for reversals, when the recommended dose for a more sensitive group is higher than the recommended dose for a less sensitive group. The proposed method avoids reversals, allocates patients to optimal doses more frequently during the trial, and selects optimal doses more frequently at the end of the trial.
pharmacology & pharmacy,medicine, research & experimental
What problem does this paper attempt to address?