OBJECTIVES: Apply a Bayesian dynamic borrowing (BDB) approach, calibrating both baseline and outcome differences to quantitatively determine the extent of external real-world data (RWD) that can be utilized as an external control arm (ECA).

METHODS: Participants randomized to one of two experimental arms or an internal control arm (ICA) of a phase II randomized clinical trial (RCT) were evaluated for bleeding events within 3 months. The Finnish healthcare system was used to develop an anonymized cohort, fulfilling eligibility for the RCT, to use as an ECA. ECA participants were matched to ICA participants using propensity-score (PS) methods based on 28 confounders. A power prior, taking the likelihood of the ECA and raising it by a power (0 to 1) to adjust the amount to borrow from the ECA, was used where the power is the overlap of the ECA and ICA PS distributions. An “elastic” function was used to further discount ECA data based on outcome differences, resulting in a double-adjustment for both baseline and outcome differences.

RESULTS: The RCT had n=505 participants randomized to the experimental arms and n=250 to the ICA. A pool of eligible Finnish RWD (n=3373) was established, of which n=1002 had overlapping PS with ICA participants and were selected as ECA; among the ECA, 6.8% had a bleeding event compared to 2.4% in the ICA. The double-adjusting BDB resulted in an equivalent borrowing of n=64 participants from the ECA, with a posterior mean 2.6% (95% credible interval: 0.9%, 4.5%) reduction in bleeding events (experiment vs augmented control), strengthening the evidence observed in the RCT.

CONCLUSIONS: The BDB extends PS-integrated methods by discounting external data based on baseline and outcome differences. Finnish RWD was used to apply BDB to augment the ICA of an RCT, using RWD to increase the number of controls used for comparison.