Jue Hou Designed Protocol to Use EHR Data to Supplement Randomized Controlled Trials

Jue Hou, PhD, was part of an international team of researchers who developed an approach to use Electronic Health Record (EHR) data to study clinical treatments.

Dr. Hou is an Assistant Professor for the Division of Biostatistics in the School of Public Health at the University of Minnesota and a full member of our Innovative Methods & Data Science Program. He designed the protocol for the study.

The research team proposed a “pipeline” approach that researchers can use to supplement randomized controlled trials (RCTs) with EHR data. Their approach is split into four modules.

Module 1 is mapping clinical variables from an RCT design onto different sources of data in the EHRs.

Module 2 is identifying which patients have the condition or disease of interest. The authors proposed starting out with a subset of data that covers all patients who potentially meet the criteria for the study. From there, researchers can narrow down which patients have the condition using an algorithm. Finally, they can determine which of these patients received treatments relevant to their study.

Module 3 is extracting eligibility criteria and endpoints from the data. Eligibility criteria are requirements patients need to meet to be part of the study. Endpoints are events that can be measured to determine if the treatment is beneficial. These can be extracted from structured, textual, and image EHR data using the mapping that occurred in Module 1.

Module 4 is detecting data errors and adjusting for biases in the data. This involves validating the data, using robust statistical methods that can account for biases, and adjusting for confounding factors.

The full protocol can be found here.