Minnesota Nursing magazine

Lisiane Pruinelli, PhD, RN

Finding patterns

A big data model to improve liver transplant care
by
by: 
Barb Schlaefer

While working as a multi-organ transplant surgery coordinator at a major medical center in Brazil, Lisiane Pruinelli, PhD, RN, found she had a growing number of questions without answers.

She noted that survival and quality of life varied widely for patients who had similar transplant procedures. Her curiosity and desire to optimize outcomes for patients drove her to learn more.

“I wondered what we could learn about the modifiable health characteristics that patients bring prior to transplant that predict outcomes,” said Pruinelli, assistant professor and OptumLabs visiting fellow. “I recognized there were some factors we could not change. So I began to ask what conditions and symptoms patients bring that we can actually influence prior to surgery to maximize outcomes.”

Liver transplantation became a viable and common treatment in the 1980s for people with diseased or damaged livers. While the majority of people survive the complex transplant surgery today, survival rates and patients’ post-operative well-being vary widely.

Strength in numbers

Today, equipped with access to data on 25,000 liver transplant patients through the School of Nursing’s partnership agreement with OptumLabs, Pruinelli is developing a promising framework to identify risk factors prior to liver transplant surgery that are predictive of better outcomes.

Building on methodology from Pruinelli’s earlier study and existing literature, the team is grouping patients into clusters that are algorithmically derived subgroups. Using this method, liver transplant patients with similar combinations of problems can be assessed over time to identify which problems influenced outcomes most, either positively or negatively.

“The research to date is primarily disease-focused,” said Pruinelli. “So we bring a holistic approach for data modeling that encompasses more. Our model is comprehensive, including 175 possible health problems including mental health, frailty and conditioning, musculoskeletal, psychosocial, behavioral, environmental, pain and neurologic conditions that have seldom been considered as impacting liver transplant outcomes.”

Improving survival rates

Pruinelli is working with the Liver Transplantation Program at University of Minnesota Health to gain insights and continuously refine her research.

“Lisiane’s program of research uses a unique methodology of clustering that is useful in helping us think in patterns,” said Timothy Pruett, MD, professor and chief of the Division of Transplantation at the University of Minnesota. “This novel approach, combined with the statistical power of the data, could give practitioners in disparate disciplines more refined predictive capacity to learn how we might better allocate resources or change care to improve survival rates for specific patients in the future.”

This study is aimed at generating evidence that can drive effective nursing interventions and ultimately decrease health care costs, Pruinelli said. “If we discover, for example, that patients entering transplantation with optimal nutrition and lower pain scores have a much higher likelihood of thriving after surgery than those who don’t, we have a case for investing in greater nutrition therapy and pain management,” she said.

OptumLabs is an open, collaborative research and innovation center founded in 2013. Its core linked data assets include de-identified claims data for privately insured and Medicare Advantage enrollees and de-identified electronic health record data from a nationwide network of provider groups. The database contains longitudinal health information on enrollees and patients, representing a diverse mixture of ages, ethnicities, races and geographical regions across the United States. The EHR data reflects all payers, including uninsured patients.

The research is funded by a Grant in Aid. Led by Pruinelli, the research team includes Gyorgy Simon, PhD, and Timothy Pruett, MD.

Tags: