OHDSI Home | Forums | Wiki | Github

Are you interested in reproducibility and scalability of methods in drug safety?

Hi OHDSI researchers,

Juan (@Juan_Banda ) and I (@jreps) have been asked to plan a special issue in Frontiers in Drug Safety and Regulation. We are thinking about the topic:

Reproducibility and scalability of methods in drug safety. This could include papers showing there are reproducibility issues, papers proposing methods to improve reproducibility and papers proposing drug safety approaches that go away from reactively evaluating 1 drug-event pair at a time and moving towards proactive systems that evaluate many drug-event pairs.

We wanted to see what the community thinks about this topic. Do you have any suitable research that would fit into this? Would you potentially submit to this special issue? Is there any modification to the topic you suggest?

Best wishes,
Juan and Jenna

Hi Jenna, Juan,

I would be interested in contributing because I work primarily in drug safety.
Is your focus primarily on prediction i.e drug-event pairs or signal detection?

From an epidemiological perspective, I think there are multiple interesting papers that can evolve from such research

E.g statistical models with Rare outcomes or overinflated zero outcomes (e.g death ). The challenge to reproducibility here might be that some researchers might pick a maximum likelihood estimation-based method that might not be suitable. One could use simulations to show the impact on reproducibility. This issue also has clinical significance because death is a serious adverse event and is often an endpoint within oncology research.