Premium
Beyond Randomized Clinical Trials: Use of External Controls
Author(s) -
Schmidli Heinz,
Häring Dieter A.,
Thomas Marius,
Cassidy Adrian,
Weber Sebastian,
Bretz Frank
Publication year - 2020
Publication title -
clinical pharmacology and therapeutics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.941
H-Index - 188
eISSN - 1532-6535
pISSN - 0009-9236
DOI - 10.1002/cpt.1723
Subject(s) - propensity score matching , randomized controlled trial , confounding , matching (statistics) , clinical trial , bayesian probability , medicine , control (management) , external validity , construct (python library) , gold standard (test) , physical medicine and rehabilitation , computer science , statistics , surgery , artificial intelligence , mathematics , pathology , programming language
Randomized controlled trials are the gold standard to investigate efficacy and safety of new treatments. In certain settings, however, randomizing patients to control may be difficult for ethical or feasibility reasons. Borrowing strength using relevant individual patient data on control from external trials or real‐world data (RWD) sources may then allow us to reduce, or even eliminate, the concurrent control group. Naive direct use of external control data is not valid due to differences in patient characteristics and other confounding factors. Instead, we suggest the rigorous application of meta‐analytic and propensity score methods to use external controls in a principled way. We illustrate these methods with two case studies: (i) a single‐arm trial in a rare cancer disease, using propensity score matching to construct an external control from RWD; (ii) a randomized trial in children with multiple sclerosis, borrowing strength from past trials using a Bayesian meta‐analytic approach.