Linking data sources
When he and his colleagues wanted to assess the cost of not achieving RA remission, no single data source provided all of the information they needed. They used both CORRONA registry data and health claims data to look at various outcome measures across disease activity categories and with adjustment for comorbidity clusters. They previously reported on the feasibility and validity of the approach.
“We’re currently doing another project where one of the local Blue Cross plans said ‘I’m interested to support you to see how efficient you are; we will donate or loan you our claims data [and] let you link it to your practice so you can actually tell us ... cost conditional on [a patient’s] disease activity,’ ” he said.
Another example involves a recent study looking at biomarker-based cardiovascular disease risk prediction in RA using data from nearly 31,000 Medicare patients linked with multibiomarker disease activity (MBDA) test results, with which they “basically built and validated a risk prediction model,” he said.
The point is that such data linkage provided tools for use at the point of care that can predict CVD risk using “some simple things that you and I have in our EHR,” he said. “But you couldn’t do this if you had to assemble a prospective cohort of tens of thousands of arthritis patients and then wait years for follow-up.”
Patient-reported outcomes collected at the point of care and by patients at home between visits, such as digital data collected via wearable technology, can provide additional information to help improve patient care and management.
“My interest is not to think about [these data sources] in isolation, but really to think about how we bring these together,” he said. “I’m interested in maximizing value for both patients and clinicians, and not having to pick only one of these data sources, but really to harness several of them if that’s what we need to take better care of patients and to answer important questions.”
Doing so is increasingly important given the workforce shortage in rheumatology, he noted.
“The point is that we’re going to need to be a whole lot more efficient as a field because there are going to be fewer of us even at a time when more of us are needed,” he said.
It’s a topic in which the ACR has shown a lot of interest, he said, noting that he cochaired a preconference course on mobile health technologies at the 2018 ACR annual meeting and is involved with a similar course on “big data” ahead of the 2019 meeting.
The thought of making use of the various digital health and “big data” sources can be overwhelming, but the key is to start with the question that needs an answer or the problem that needs to be solved.
“Don’t start with the data,” he explained. “Start with [asking] ... ‘What am I trying to do?’ ”
Dr. Curtis reported funding from the National Institute on Arthritis and Musculoskeletal and Skin Diseases and the Patient-Centered Outcomes Research Institute. He has also consulted for or received research grants from Amgen, AbbVie, Bristol-Myers Squibb, CORRONA, Lilly, Janssen, Myriad, Novartis, Roche, Pfizer, and Sanofi/Regeneron.