The Learning Health System in Action: Studying Guideline Changes at Kaiser Permanente Washington
Sascha Dublin, M.D., Ph.D., describes the learning health system framework as a circle: design a change, apply a plan, collect data to study the impact, disseminate the results and make adjustments. That’s how you bring best practices and knowledge into your health system.
A senior investigator with the Kaiser Permanente Washington Health Research Institute, Dublin recently gave a presentation to the NIH Collaboratory Grand Rounds on her organization’s work to study a major change in gestational diabetes screening.
She described the background of their research: If a woman has gestational diabetes, her baby may grow too large, raising the risk of cesarean delivery. After birth, the child may experience hypoglycemia. Dublin explained that there is some disagreement about the best method of screening for gestational diabetes. Most U.S. providers follow a two-step process, where all women receive a basic screening test and only women with an abnormal result get longer and more intensive testing. But in 2010, the International Association of the Diabetes in Pregnancy Study Groups proposed that all pregnant women undergo the longer, more intensive test.
This led to some confusion and controversy, she said. Because the evidence wasn’t clear, Kaiser Permanente Washington chose follow the new international recommendations and screen all pregnant women with the single test.
In researching the impact, Dublin and her team were able to have something of a natural control group because pregnant women insured by Kaiser Permanente Washington who were cared for by outside, contracted providers kept using the two-step method, whereas physicians in Kaiser Permanente’s own group practice largely switched to the one-step screening. She said this made it more likely that whatever change they found would reflect the effects of the switch in screening process.
Implementers developed an Epic smart set to order the new test and clinical leadership visited clinics and talked to them about adopting the new guidelines, which were widely adopted internally.
The researchers found that the guideline change led to a 41 percent increase in diagnoses of gestational diabetes in the group practice, but it did not lead to a reduction in cesarean deliveries, high-birthweight babies or admissions to the newborn intensive care units. “The hoped-for benefits did not materialize in ways we could measure,” Dublin said. “The take-home message to delivery system leaders was that there was no benefit to adopting the one-step approach,” she added, so the organization decided to return to the two-step testing.
The one-step approach had been more time- and resource-intensive, so the clinicians were eager to hear the results of the study. Closing that loop gave them satisfaction, she said. Going back to the one-step process gave them more time and energy to screen for depression and other topics. There were comments that some clinicians felt they were spending too much time on blood sugar, and not spending enough time on other things.
Reflecting on the value of the project, Dublin said that it was important that the research question came from Women’s Health clinical leaders. “We maintained engagement with clinical leaders, asking them to choose the outcomes to be studied,” she said. “The leaders who were in a position to disseminate findings and change care practices were part of the research team from the beginning.”
Finding the funding for this type of research is always a challenge, she admitted. But when a healthcare system wants to change a practice, she noted, it should consider planning for evaluation from the beginning, and that includes investing in the data infrastructure to make that research possible.
