Abstract
Cox regression and Kaplan-Meier estimates are often needed in post-approval clinical research which requires access to individual person data (IPD). Think about a long-term treatment-effect review or a comparative effectiveness oncology study involving the IPD owned by a competitor. Here, IPD cannot always be shared due to privacy or proprietary restrictions, which complicates the making of such estimates. I propose a method that generates pseudodata replacing the IPD by only using non-disclosive IPD aggregates which are shared with a central computer and are input parameters to a Gaussian copula (GC) generating the pseudodata, allowing analysts to access completely anonymised versions of possibly multiple datasets on which to run analyses.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have