$1.4M NIH Award Funds Research to Integrate Electronic Health Records

It’s a common “medical” complaint: You switch to a new doctor and find that electronic health records have not moved with you—cue the déjà vu while filling out the same forms it seems you’ve completed a hundred times before. 

But beyond the frustration, what does it mean for your health when your providers aren’t seeing your complete medical history? 

With support from a $1.4 million National Institutes of Health grant, a University of Maryland health policy expert is collaborating with a University of California, San Francisco researcher to answer that question.

Nate Apathy, a professor of health policy and management at UMD’s School of Public Health and A. Jay Holmgren of the UCSF School of Medicine will examine how doctors make clinical decisions with the data at hand, how their decisions change when they see all the information including “outside data” from elsewhere, and what impact that change ultimately has on a person’s health. 

“There are lots of layers that introduce friction to the use of this outside data,” said Apathy. “There are still a lot of important outstanding questions about how the use of outside records changes doctors’ decision-making and, furthermore, how those gaps in records could impact a patient’s health.”  

The rate at which doctors look at “outside data” to make clinical decisions is, in general, surprisingly low, Apathy said. In an outpatient setting, less than 30% of them on average review outside data. For other doctors, the rate is 5-10%. (Apathy noted that some outside data may not be relevant to the patient’s health concern.)

Over the next four years, the researchers will examine data from two academic health systems: UCSF Health and the University of Maryland Medical System. The IT systems of both institutions were recently updated to seamlessly include outside data (as permitted by patients) in patient electronic health records, allowing researchers to compare if and how clinicians used it for clinical decisions before and after the system upgrade. 

Using machine learning and AI tools, the researchers plan to create different patient “phenotypes”—profiles of observable traits like sex, age, disease diagnoses and medical care history. These phenotypes will help flag relevant outside records for a given patient so doctors can proactively look at the relevant data for their decision making. 

“When health data travels seamlessly between institutions, there is immense promise to drive down health care costs and reduce health care use and duplicative paperwork, all of which can improve patient health and satisfaction,” Apathy said. “We hope this research and the open-source tools we will create from it will contribute to improved decision-making and health outcomes.”