PCORnet® is a full-scale research infrastructure designed to make studies faster, easier, and more impactful.
And helping researchers design studies with confidence in their representativeness is a recent focus for PCORnet.
Because PCORnet is a national network connected to approximately 47 million patients who receive care each year, the network can be used to conduct a wide array of health studies, from pragmatic trials to retrospective observational research.
By comparison, the American Community Survey samples approximately 3.5 million households per year and while there is no national census for patients who receive healthcare in the U.S., the large size of PCORnet and the network’s demographic comparability to the U.S. suggest that a wide array of topics can be studied using the network. (See PCORnet Population Insights to review the collected data.)
Over the last year, the PCORnet Representativeness Project team has been working through the nuances of how to help clinical researchers optimize their use of PCORnet to enhance and evaluate the representativeness of a clinical study. The project will culminate in early 2026 with new resources for guiding researchers in their use of the PCORnet infrastructure.
“When we develop conclusions from research, we want these conclusions to apply to the people we will ultimately treat in the healthcare system,” said Carly Brantner, Assistant Professor of Biostatistics and Bioinformatics at the Duke School of Medicine and the Duke Clinical Research Institute (DCRI) and a member of the PCORnet Representativeness Project team.
“Part of the challenge of this project is conceptualizing who those people are and thinking through the best ways to compare a study sample with this target group of people to whom we want our findings to apply,” said Brantner.
Brantner recently joined the Casual Inference podcast to talk about how the PCORnet infrastructure can support investigators in thinking about ways to generate relevant samples using a national network.
“Because PCORnet is so large, descriptive data in prep-to-research data queries can help investigators identify patient populations to recruit from and benchmarks to which they can compare their study samples,” she said.
The Casual Inference podcast is hosted and sponsored by the American Journal of Epidemiology. It holds casual, accessible conversations with guests around topics in epidemiology, statistics, data science, causal inference, and public health.
Laine Thomas, Professor of Biostatistics and Bioinformatics at Duke and Deputy Director of Data Science and Biostatistics at the DCRI and the lead of the PCORnet Representativeness Project, acknowledged the inherent challenges in achieving representativeness.
“While the target population can be conceptualized based on eligibility criteria, having a dataset of all people that meet those criteria can be a tall order,” said Thomas. “What makes a dataset a gold standard is how closely the actual data set matches the ideal conceptual target population.”
New Tools
Helping investigators get as close a match as possible is the goal of the representativeness project, said Thomas. “Through this work we hope to show that PCORnet is not only a data network but a design tool to make research more representative from the start.”
The PCORnet Representativeness Project is developing resources to help researchers assess their PCORnet® Studies for representativeness.
The project is expected to conclude by March 2026, said Thomas, resulting in a tutorial on approaches for assessing representativeness of a study design relative to the network’s many resources. The team is also preparing manuscripts for peer review and has contributed content to the online PCORnet® Playbook.
Thomas herself will be able to use these new tools directly because she is part of the leadership team of a newly designated PCORnet® Study called Comparative Effectiveness of Emerging Medications in Children with Inflammatory Bowel Disease (COMPARE).
In her podcast discussion, Brantner looked ahead to these new tools.
“When a researcher does a study using PCORnet, we want to help make their studies representative,” said Brantner. “PCORnet can help guide that process.”
Interested in conducting national-scale research? PCORnet may be used by all interested investigators, regardless of affiliation or source of funding. Contact the PCORnet® Front Door to get started.