Nationwide Children’s Hospital in Columbus, Ohio, has developed and implemented a so-called “learning health system” that reduced utilization of services, lowered the cost of care and generated quality data to use in future research.

Learning health systems use electronic health records to provide more personalized treatment and drive research. However, most learning health systems use top-down models in which data are integrated from more than one hospital system, which can be difficult to implement.

Nationwide Children’s, in partnership with the Ohio State University, created a pilot program, called Learn From Every Patient, that optimizes the EHR to integrate clinical care, quality improvement and clinical-driven research. This model was tested with a group of 131 children in the hospital’s cerebral palsy program, according to Dr. William E. Smoyer, vice president of clinical and translational research and director of the Center for Clinical and Translational Research at Nationwide Children’s.

Children treated in the program during a 12-month study period experienced a 43 percent decrease in total inpatient days, a 27 percent drop in inpatient admissions, a 30 percent reduction in emergency department visits and a 29 percent decrease in urgent care visits. It also reduced health care costs by 25 percent and provided a 6-to-1 return on investment.

The program has already used the data collected to improve the care of children with cerebral palsy. For example, these young patients would receive hip X-rays each year, which is an added expense and exposes them to radiation. The physicians in the pilot questioned whether there was a subset of patients who did not need these annual X-rays. The learning health system generated enough data to determine that the X-rays were not needed for children with mild cases of cerebral palsy. This discovery directly changed the hospital’s standard of care, Smoyer says.

The local learning system, he says, “was designed as a simple solution to integrate research and clinical care, but turned out to be a disruptive innovation.”

There are some challenges to instituting a local learning health system, Smoyer notes. For instance, clinicians need to be willing to change the care they provide based on the research generated. The research questions also need to be incorporated into the EHR, not in separate databases, and the data need to be reviewed for both completeness and accuracy.

The results of the study were recently published as “‘Learn From Every Patient’: implementation and early results of a learning health system” in Development Medicine and Child Neurology and were the subject of the JAMA editorial “Creating Local Learning Health Systems: Think Globally, Act Locally.”

Smoyer says it is his “hope and dream” that this local learning health system model be expanded both within and across institutions nationwide.

“The vision is to learn from every patient so that we can systematically improve their care in the future,” he says.