Sioux Falls, S.D.-based Sanford Health hopes to improve its own clinical outcomes by sharing patient data with nearby universities.
Through the Sanford Data Collaborative project, six institutions — including Sanford Research, the health system's research arm — will receive patient data stripped of private information. The data, including clinical, financial, operational and quality information, is gathered from patient visits to Sanford's 45 hospitals and almost 300 clinics.
Each of the six institutions pitched their own research project, and will work separately to conduct studies on cancer, diabetes and hypertension, among other conditions. From the universities' analyses, Sanford hopes to gain actionable insights for patient care with a focus on predictive risk, chronic disease management, diagnostic testing and pediatric and women's health.
Investigators from the six participating organizations will collaborate with Sanford providers to apply research findings directly to the health system's patient interactions.
"One of the tenants we have when we're speaking with our academic partners is this [research] isn't just for publication," says Benson Hsu, MD, vice president of data analytics at Sanford. "Although we're extraordinarily interested in expanding the way that we deliver healthcare and in contributing to the literature, what we're really interested in is how we can impact patients immediately."
In early 2017, Sanford's clinical and research leaders selected six projects from more than 20 research institutions that applied to use the health system's data. In addition to Sanford Research, the projects hail from Madison, S.D.-based Dakota State University; Brookings-based South Dakota State University; Grand Forks-based University of North Dakota; Vermillion-based University of South Dakota; and Grand Forks-based University of North Dakota School of Medicine & Health Sciences.
The Sanford Data Collaborative team worked with the health system's compliance, legal and patient safety offices to ensure patient data was appropriately deidentified before delivery to participating universities. Dr. Hsu underscored the importance of open communication with the universities about what patient data Sanford would and would not release.
"I believe the biggest challenge we had when we set this up was to ensure patient safety and privacy," he explains. "There's this push and pull, because the more information that is released, the better the algorithms can become. Yet, at the same time, fundamentally, as a core value of this organization, patient privacy is key."
In one of the Sanford Data Collaborative projects, principal investigator Carole South-Winter — an assistant professor at University of South Dakota Beacom School of Business — plans to develop a risk score for patient readmission following heart surgery. However, unlike existing predictive models, this algorithm will also inform preventive measures for Sanford to use.
"The algorithm ultimately will become a prescriptive tool to not only tell us the risk of readmission, but also impart what we can do to decrease that risk," Dr. Hsu says. "What it is about this particular patient that we can intervene on to actually decrease the readmission, as opposed to just saying, 'this person is at risk.'"