Earlier in my career, I remember presenting findings from a new study on the prevalence of gestational diabetes in various demographic groups. My audience that day was a group of physicians. My collaborators and I provided these health care clinicians with a variety of odds ratios, showing how some racial and ethnic groups were at elevated risk of developing gestational diabetes compared to their White peers. At the end of the presentation, the attendees asked questions and one doctor raised his hand.
“What do we do with this information?” he asked. He and his colleagues already knew certain demographic groups were more likely to develop gestational diabetes. They wanted to know what they could do as health care providers to reduce that risk and improve the health of their patients. It was an eye-opening moment for me and a turning point in my career.
I’ve always been interested in rigorous research about health, but in the past decade, I’ve become an evangelist for presenting data in formats that are most useful for decision makers. In the case of the diabetes study, we ran another analysis using the same data and were able to show how clinicians could help reduce their patients’ risk of developing gestational diabetes by focusing on their patients’ body mass index for certain racial and ethnic groups. For example, keeping Hispanic patients’ BMI in the normal range would reduce the prevalence of gestational diabetes by a third or more in this population. It was concise, practical information that gave doctors and their patients direction in pursuing healthier outcomes. During my seven years at Mathematica, I’ve had ample opportunity to apply lessons from that diabetes study. In this era of digital transformation, researchers like me have increasingly sophisticated web development tools at our disposal for displaying data in ways that empower action.
As with the diabetes study, sometimes we must take an iterative approach, incorporating feedback from our intended audience to reach a better final product. For example, a few years ago, we created an online dashboard showing the performance of accountable care organizations (ACOs) that participated in CMS’s Pioneer and Next Generation ACO Models. In an early version, we provided each organization with their performance data as compared with established national standards, such as the Medicare Shared Savings Program Quality Measure Benchmarks. What was much more informative and motivating, however, was how we recontextualized the same data to show how an organization compared with its peers across different measures. Leaders of these organizations told us it was very meaningful to them to assess their performance using data from a peer, and it created the opportunity for Mathematica to conduct follow-up coaching sessions with cohorts of ACOs so that they could learn from one another and improve.
The creation of online dashboards has become a throughline of my work at Mathematica. The subject matter can vary quite a bit, from measuring the performance of ACOs to displaying the alignment of demographic characteristics between trainees who received training in contact tracing through the TRAIN Learning Network and the communities they serve. Currently, I’m leading a partnership with HHS Technology Group to provide analytic services to the National Kidney Foundation of Michigan to support the Morris Hood III Chronic Kidney Condition Prevention Initiative. We’ve been able to make the underlying calculations of risk that inform our work for decades, but what’s different today is our ability to build interactive dashboards, including a cost calculator and a risk calculator. These tools will enable the foundation to identify and help Michigan residents who are at high risk for chronic kidney disease, and address health disparities in vulnerable populations such as Medicaid beneficiaries and racial or ethnic minorities.
Part of what appealed to me about Mathematica was the opportunity to interact directly with decision makers who can change policies, programs, and practices in ways that improve people’s lives. With online dashboards and other digital solutions, we can apply human-centered design principles and then road-test prototypes to ensure that the data we present is useful to those in a position to act. Developing these dynamic digital solutions lends itself to iterating and collaborating with audiences that would use our tools. These engagements force me and my colleagues to look beyond the validity of our data and consider their utility. To quote the clinician who challenged me to rethink how I was presenting gestational diabetes risk data: “What do we do with this information?”