Sharing of Individual Patient-Level Data by Trialists of Randomized Clinical Trials of Pharmacological Treatments for COVID-19, Esmail et al, 2023

[cassava7]

Access to individual patient-level data (IPD) of randomized clinical trials (RCTs) increases the reproducibility of research, facilitates exploration of effect modifiers, and leverages research investments beyond the original research question.

This study aimed to describe access to IPD of RCTs of pharmacological treatments for COVID-19 and to determine whether the intent to share IPD as reported in the registry and publication or preprint was consistent with actual access to IPD.

JAMA - https://jamanetwork.com/journals/jama/fullarticle/2803583

—

Summary by epidemiologist Gideon Meyerowitz-Katz:

Interesting paper - of 224 COVID-19 RCTs preprinted or published by May 2021, only 22% had accessible patient data on request

Studies done in high-income countries, with mixed funding, with more patients, and preprints were more likely to provide data on request

Data availability statements were, on the whole, untrue. Of the 138 RCTs that said they would share data, only 37 (27%) actually sent it through after a request and multiple reminders

 

[rvallee]

Data availability statements were, on the whole, untrue. Of the 138 RCTs that said they would share data, only 37 (27%) actually sent it through after a request and multiple reminders

I've seen this quite a few times in LC papers. They mention data availability on request, but reserve the right to withhold for any reason. It's usually put in some form of "give us a good study proposal and we'll consider it", but of course that's exactly the same as no sharing, as they would only share the data to like-minded researchers whose outcomes align.

So with this standard, the only data sharing would be between people who agree with one another, challenges are refused. Which is what we've seen in BPS land. This whole system is built to fail and abuses patient privacy issues to shield researchers from any analysis that isn't fully deferential.