VCU news

By Jackie Kruszewski
C. Kenneth and Dianne Wright Center for Clinical and Translational Research

Registered nurse Rachel Pryor had collected 2,000 COVID-19 antibody tests and 2,000 surveys. And she needed to know, in the early weeks of the pandemic, what they said about the risk of the novel coronavirus for health care providers.

“The prevalence of antibodies in providers was one way for the health care system, and the community in general, to better understand the prevalence of COVID-19,” said Pryor, a data analyst and research coordinator in the Infection Prevention Department at VCU Health. “And it was important to understand if we were protecting our health care workers.”

Logistically, though, the number of tests and surveys was daunting. Data needed to be confidential, but linked to surveys that conveyed meaningful information about the person. So like researchers before her, Pryor turned for help to the informatics team at Virginia Commonwealth University, housed in the C. Kenneth and Dianne Wright Center for Clinical and Translational Research.

Informatics has long been the silent backbone of health research at VCU. Data experts who help faculty find what they need in notes and numbers work through 250 to 300 requests for support in a typical year.

But the pandemic kicked their role into high gear. In addition to their regular load, they worked on nearly 50 pandemic-related projects last year, adapting quickly to the needs of health researchers.

The team was behind the scenes of many of VCU’s most notable COVID-19 research projects. They made a confidential registry of people who recovered from the virus and were willing to donate their plasma. They helped researchers quantify the alarming surge of opioid overdoses that doctors were seeing. They surveyed people who noticed changes in their ability to taste and smell. And they helped researchers manage the vast amounts of data accrued in their scientific inquiries.

“When investigators are faced with data capture and management challenges — large or small — they come to the Wright Center informatics team, because we have the infrastructure and expertise to scale quickly,” said Tim Aro, the Wright Center’s manager of clinical research informatics.

Data tells researchers what’s possible

A researcher with a question to investigate first needs to know if such an investigation is even possible.

“Investigators traditionally might say, ‘Well, I sense, based on my practice, that we can do this study,’” Aro said. “But once the study or trial is underway, they come to realize they’re falling short of their recruitment goals. Many trials don’t get off the ground because they don’t have a large enough patient population to support the research.”

Previously, researchers might have reviewed charts, manually pulling hundreds of individual patients and transcribing the data, to know if they could attempt a study.

These days, researchers can query electronic health data securely and easily with the help of the informatics team. And tools supported by the informatics team can expand those searches beyond VCU, facilitating multisite trials that increase the effectiveness of a study.

“Investigators can get a quick and fairly accurate count of a potential patient population here at VCU Health, without exposing identifiable information,” Aro said. “And with national platforms, they can query 50 to 60 health care organizations from their computer in a secure manner, which can provide tremendous insights into study feasibility and design.”

When VCU researcher Arun Sanyal, M.D., a liver specialist and gastroenterologist at VCU Health, brought a COVID-19 clinical trial to VCU in March, the informatics team was there to create a secure portal for the patients of doctors at other institutions. With each referral, automated notifications sent information to the study team.

The referral form was operational within 24 hours, tested and validated to capture the information Sanyal needed to receive patients. And the drug Sanyal was testing, remdesivir, later became one of the first drugs authorized by the federal Food and Drug Administration for use in treating COVID-19.

Early on in the pandemic there were a lot of questions. Had more health care providers than we thought had the virus asymptomatically? Are they safe now? Does personal protective equipment work?

Unlocking the secrets inside clinical research data

Once a clinical trial or health study is up and running, it can generate an enormous amount of data from a variety of sources.

For Rachel Pryor’s project, the informatics team was instrumental in linking those 2,000 tests and surveys in a way that allowed for meaningful study.

“Early on in the pandemic there were a lot of questions,” Pryor said. “Had more health care providers than we thought had the virus asymptomatically? Are they safe now? Does personal protective equipment work?”

She needed to test providers for antibodies, which would indicate whether someone already had the virus, and combine that data with surveys about work and post-work habits.

“I created a survey of behaviors that could potentially put a person at risk for increased likelihood of developing COVID-19,” Pryor said. “What unit did they work on? Did they travel internationally? Do they practice social distancing and wear a mask outside of work?”

Creating the survey was simple enough, but scheduling the tests and matching the results to the surveys would be a big undertaking at a sample size of 2,000. It was a multiform survey — with an initial screening questionnaire, consent forms to review and a notification system for participants.

“The informatics team came in and linked all these together for us, which was quite complex and certainly beyond my level of understanding,” Pryor said. “And they were able to de-identify the data so that we could share results without exposing the names of individual people.”

Years ago, Aro said, most investigators used Excel spreadsheets to maintain their study data. Now, his team manages, trains and helps researchers use a more secure and functional platform. That’s what they used for Pryor’s study.

“We really couldn't have done this without the team at the Wright Center,” Pryor said. “They were crucial to the performance of it.”

In addition to studies like Pryor’s, the informatics team has been vital to streamlining the process for getting interested patients at VCU Health into COVID-19 treatment trials, such as Sanyal’s remdesivir study.

A process they set up identifies patients based on chart data entered into the system and sends real-time emails to researchers when someone comes into the emergency room who might qualify for their study. It’s been a key part of VCU’s nimble reorganization of clinical research to adapt to the unique challenges of the pandemic.

Better data makes future research possible

Since the pandemic began, the informatics team has helped VCU researchers access and build datasets that will inform years of future coronavirus study.

Internal to VCU, they helped one doctor and professor build a biobank that connected the dots of health data from COVID-19 patients. The biobank registry compiled medications, labs, procedures and histories from de-identified patients so future researchers could study how the virus affects different people.

And it doesn’t simply populate numbers from a medical chart into a database. There are what Aro calls the “infamous” clinical notes — information about a patient that a physician or nurse might record in a narrative form that doesn’t lend itself to easy analysis when considering a large set of patient data.

“Clinicians often use their own shorthand, or might describe the same symptom in different ways,” Aro said. “That’s where natural language processing comes in, programs with text searching capabilities that can find what researchers need inside clinical notes.”

Taking that same goal beyond VCU, the informatics team at the Wright Center helped VCU join the National COVID Cohort Collaborative, led by the National Institutes of Health, which securely collects and organizes clinical and diagnostic data from patients across the country — to create an even broader dataset for researchers.

The database included the records of more than 3 million patients as of Feb. 22.

“The days have been long, like for so many people who work in and around health care,” Aro said of the workload for his team. “But we’re bound and determined, because so many researchers are doing so many cool, time-sensitive things. We want to be that cog in the machinery that doesn’t slow them down. We want to speed them up if we can.”

As for Pryor’s study, she said the results confirmed what similar studies have shown: The prevalence of COVID-19 antibodies in people, early in the pandemic, was lower than initially thought.

“Everyone wanted to think they’d had COVID,” Pryor said. “‘Oh, I had that weird flu in March.’ Well, that probably wasn't COVID.”

And it showed that personal protective equipment works. Like other studies, they found that the main, statistically significant risk for getting COVID-19 was having a household contact test positive.

“It wasn’t where they worked or the patients they cared for,” Pryor said. “It was encouraging to see that the masks and gowns and equipment were, and are, protecting our health care workers.”

VCU researchers can request informatics support for their research via the Wright Center.