Major Academic Groups Push Forward with Socialization of Trial Designs to Improve COVID-19 Trial Quality & Hence Outcomes

Major Academic Groups Push Forward with Socialization of Trial Designs to Improve COVID-19 Trial Quality & Hence Outcomes

Elizabeth Ogburn, an associate professor in the Department of Biostatistics at Johns Hopkins Bloomberg School of Public Health, and a significant group of colleagues are working on a disruptive new clinical trials design platform, open-sourced for all to consider, called the COVID-19 Collaboration Platform. This effort responds to the over 3,000 COVID-19 clinical trials launched in the past nine months since the onset of the pandemic. Michael Rosenblum from Johns Hopkins University also just recently contributed to what will hopefully lead to better designed clinical trials associated with not only COVID-19 studies but all others.

The Problem with COVID-19 Clinical Trials

Challenges with too many trials, either improperly designed or as a result of a somewhat haphazard reaction, the team led out of Baltimore sought the ability to distinguish meaningful study signals from what is described as “confounding statistical noise.” After all, many critics have already called out that perhaps a majority of clinical trials associated with COVID-19 may add no clinical value in the end. With the many “underpowered trials” (not enough participants) hence lacking “statistical power” to concerns of “false-positive results,” given many studies just weren’t designed properly in the first place.

Socialize Clinical Research?

Now, the group of researchers aims to change the situation described above. By developing an open, source-based knowledge repository for open and shared access to proper protocols, data models and other mission-critical elements involved with the study designed the team announced the COVID-19 Collaborative Platform. Professor Ogburn was quoted in Michael Eisenstein’s piece in the Johns Hopkins University Hub, declaring, “We’re trying to facilitate matchmaking for clinical questions.” Ogburn and team are on a mission to set up a sort of externalized share services for not only the pooling of efforts and study design best practices and artifacts but also working to evidence direct value-add, as measured by a reduction in squandered resources or underpowered studies.

Sharing and IP

Often, in the competitive world of biopharma drug development, it’s taboo to share anything, unless it’s via some formalized and accepted forum or conference, where, what is actually shared, remains as cursory and not granular. This kind of competitive tension also occurs among major academic medical centers, as they often compete amongst each other for major grants, for instance. Or, for that matter, relatively small independent trial site organizations can actually team up with networks to access reusable processes, best practices, artifacts and the like. But the general permeating culture remains one of scarcity, security and suspicion in regard to the outsider.

As Professor Ogburn commented, there is often a dearth of incentive (from publishers, funders, etc.) to share key trial design intelligence.

Enter the COVID-19 Collaboration Platform

The COVID-19 Collaboration Platform represents a major collaborative undertaking including the following academic medical centers and research-oriented institutions:

With the Executive Committee, including Barbara Bierer (Harvard Medical School), as well as Betsy Ogburn and Dan Scharfstein (both with Johns Hopkins University), Joseph Lee runs the group’s operations. Existing protocols can be viewed here.

The Streamliners

Another Johns Hopkins team member involved with biostatistics, associate professor Michael Rosenblum has developed a statistical strategy that streamlines clinical trials while not impacting quality at all. He and team have published their work this month in Biometrics. In this work, the team simulated clinical trials with the use of medical records from hospitals containing COVID-19 patients. Interestingly, the team focused on what’s known as confounding factors in study cohorts (e.g., age, gender, pre-existing conditions, etc.); these, of course, can directly impact the randomization dynamics of the study. The study summarizes that through the use of a covariate adjustment, the team could lessen the impact of such confounding factors in the study’s simulated trials. Now, Rosenblum is developing specific open-source software to incorporate these new techniques into existing study design initiatives.

Call to Action: Perhaps this group is on to something important. You can contact them at [email protected] To reach professor Elizabeth Ogburn see her profile. To access Michael A. Rosenblum see this profile.