Big data, small state: R.I. database offers the potential for efficient, improved health care

Researchers at the Brown-based, federally funded Advance-CTR program are using Rhode Island’s All-Payer Claims Database to improve health care and train the next generation of health care scholars.

PROVIDENCE, R.I. [Brown University] — From the machine learning that allows Alexa to understand users who ask “What’s the weather today?” to the systems that predict the paths and impacts of hurricanes, big data is being used to accomplish remarkable tasks across a wide range of sectors.

The field of health care is no exception — physicians are using artificial intelligence tools to better identify breast cancer, scholars are analyzing patient data to identify inefficiencies within the health care system, and the possibilities for using data to solve problems in health care are nearly boundless.

Through a new partnership between the Rhode Island Department of Health and Brown-based, federally funded Advance Clinical and Translational Research (Advance-CTR), five health care researchers affiliated with Brown University, Lifespan and the University of Rhode Island now have access to the state’s All-Payer Claims Database (APCD). APCDs are mandated, state-run archives that contain information from medical and pharmacy bills for the state’s residents, whether they’re insured by Medicaid, Medicare or private insurance companies.

Dr. Ira Wilson, the partnership’s co-leader, has been studying the sources of health care cost increases in Rhode Island for the past 30 years. His team is comparing costs between hospitals or provider groups for different procedures in order to help reduce cost increases.

“This project is all about making a link between a state asset, the APCD, and a university-based scientific collaboration, Advance-CTR, to ensure that our health care system works to benefit the residents of Rhode Island,” said Dr. Ira Wilson, the partnership’s co-leader and a professor at Brown’s School of Public Health and Warren Alpert Medical School. “Rhode Island spends $8 to $9 billion on health care each year. If we can figure out how to cut just a fraction of a percent, it will more than pay for the necessary infrastructure.”

The five research teams have had access to the database for more than a year, and Advance-CTR’s partnership license will remain active until at least April 2020.

The primary goals of the APCD partnership? To use big data to benefit residents of Rhode Island, and the nation, by figuring out how to minimize waste, improve health care and train the next generation of health care scholars to be proficient in a data-driven world — all while properly safeguarding every patient’s personal information.

Minimizing waste, improving care

While a half dozen states have made APCDs accessible to researchers to date, Rhode Island’s database is ideal for researchers for several reasons, said Neil Sarkar, director of the Brown Center for Biomedical Informatics, an associate professor of medical science and co-leader of the partnership.

Neil Sarkar, director of the Brown Center for Biomedical Informatics and co-leader of the partnership, is leading a team to determine what health conditions are amenable to being modeled using a claims database and which are not.

First, because Rhode Island has a population just over 1 million, it is possible to track nearly all instances of heart attacks or shoulder surgeries, as two examples, instead of selecting a sample of patients with the hope that the sample is statistically representative, Sarkar said. Because the APCD includes residents with government-sponsored health insurance and private insurance, it’s a nearly complete dataset for studying the rates of certain conditions among Rhode Islanders.

In addition, because approximately 70 percent of the state’s residents remain in Rhode Island, it’s possible to track patients over the long term even if they switch from private insurance to Medicaid and back, Sarkar said. Switching between private and government-sponsored health insurance can make tracking long-term health outcomes particularly challenging, but is an area where APCDs are critical.

“One of the things that is really exciting about conducting research in Rhode Island, if you work at the population level, is that you can actually get full longitudinal histories of individuals and track health care trajectories over time in a much more controlled environment,” Sarkar said. “It is an excellent test-bed for research.”

Currently, Rhode Island’s APCD includes medical bills, or claims, from 2011 through 2018, with more data being added on a quarterly basis as it becomes available, Wilson said. The program’s leaders aim to continue the partnership for many more years and to expand the number of seats to support additional researchers.

To date, five projects enabled by the partnership are now underway.

Dr. Brett Owens, chief of sports medicine at Miriam Hospital and a professor of orthopedics at Brown, is studying the long-term outcomes of orthopedic procedures for shoulder injuries. Shoulder instability can occur if the shoulder ligaments are stretched or torn when the shoulder is dislocated; some cases can be treated with physical therapy while other cases require surgery. Unlike primary care physicians, specialists often don’t get the chance to track patients over time, unless there are severe aftereffects.

Dr. Orestis Panagiotou, an assistant professor (research) in the Department of Health Services, Policy and Practice at Brown, is investigating health care services that provide no or minimal value at high cost, such as diagnostic imaging for low-risk conditions without medical warning signs, or services that may in fact be harmful to patients such as unnecessary laboratory tests. The team is looking at the rates at which various treatments are used so that health care providers can improve the care they provide.

Sarkar is leading a team to determine what health conditions are amenable to being modeled using a claims database and which are not. For example, diabetics with high blood pressure also have a high risk for heart attacks and cardiovascular conditions. This pattern and others — such as risk factors for blindness among premature babies — should be apparent from analyzing the data. Determining what conditions can be studied using APCD will lead to new studies and further insights that could result in medical care improvements. Sarkar is also leading the analysis of APCD to provide data-driven insights for a Rhode Island General Assembly special commission studying the impacts of insurer payments on access to health care.

Wilson has been studying the sources of health care cost increases in Rhode Island (and the rest of the nation) for the past 30 years. With access to the APCD, his team is comparing costs between hospitals or provider groups for different procedures, determining conditions that require more preventive interventions to reduce costly and distressing emergency department visits, and assessing the use of low-value, high-cost treatments for various conditions. The goal is to help the state reduce health care cost increases in a responsible manner.

Steven Cohen, an assistant professor at the University of Rhode Island, is using the APCD to study the socioeconomic factors that impact health care use and medical outcomes in Rhode Island. His team is examining the database by ZIP code to see if certain areas have particularly high or low usage of emergency services and medical tests such as MRI or CT scans. This information could inform policymakers and health care practitioners in order to increase health equity and decrease costs.

Training students

Another major goal of the partnership is to train the next generation of health care researchers in how to best conduct research using the state’s APCD.

“We live in a big data world,” Wilson said. “Part of our mission as a school is to train appropriately qualified and knowledgeable people at every level, whether they’re an undergraduate, master’s, Ph.D. student, postdoc or faculty member to use all kinds of data. That includes publicly available, de-identified data you can download from the internet, to data that are only available under aegis of strict data-use agreements, like the APCD.”

Hace Oh, a doctoral student at Brown’s School of Public Health who intends to explore health care economics during his thesis work, gained experience using data from the APCD. For the past year, he has worked with Wilson’s team on the health care cost trends project.

Oh’s primary role has been to combine the lists of physicians and nurse practitioners contracted with the four largest private health insurance providers in the state, which involved matching doctors on the lists and standardizing naming conventions, he said. This work will assist in future cost analyses by allowing the research team to link specific providers to claims and compare different insurance systems.

Now he is working with Megan Cole — a 2017 Brown Ph.D. graduate, assistant professor at Boston University and adjunct assistant professor at Brown’s School of Public Health — on the project, adjusting for demographic differences between organizations.

Oh has found working with Cole to be a meaningful learning experience.

“She’s willing to take the time to explain what she’s doing,” Oh said. “She’s been very helpful with the day-to-day activity — like learning to code with new data — and how to think more rigorously.”

While ensuring the data was complete and identifying missing elements, Oh had the opportunity to collaborate with doctors, competing insurance companies and state officials.

“As a student, being given the responsibility to talk to state officials and work with leaders in different organizations has been incredibly rewarding. This experience has made me want to continue working with big data and big databases and get even better at it.”

In addition to providing students with invaluable hands-on experience, training students to use the APCD also benefits the state, Sarkar said.

“The students, as they analyze these data and start using them, are able to provide the state real-world documentation in terms of ‘Here’s how you actually get at this kind of a dataset and the kinds of interpretations you can get out of it,’” he said.

Some of the students who learn how to conduct APCD research may end up working at state departments of health, in Rhode Island or elsewhere, Sarkar noted.

Protecting patient data

Before Advance-CTR-affiliated researchers could gain access to the APCD to begin their studies to improve health care, the state and Advance-CTR leaders, including Sarkar and Wilson, explored and implemented a robust set of safeguards to ensure that every patient’s personal information is properly protected.

These measures include an encrypted system that requires multiple permissions and authentication steps in order to view the de-identified data, and limitations to ensure that researchers can only export the results of their own analyses — not the raw data — Sarkar and Wilson said.

The APCD is available through a secure system designed to support health care data analytics created by a partnership between the Center for Biomedical Informatics and Brown’s Computing and Information Services department.

Security doesn’t stop there. Everyone from the lead researchers to the students take protecting sensitive information seriously, Oh said.

“Just to get into the servers to work with the data, there’s a multi-step authentication process that we have to complete, which is a reminder to everybody that we’re dealing with really sensitive data. It’s something we think about every day.”

Sarkar added that the Advance-CTR leaders also take steps to ensure that the research conducted using the APCD is taken on for the right reasons — the potential to improve health care systems in ways that ultimately benefit Rhode Islanders and patients across the nation.

“Any research that we do has to have a good reason to be done,” he said. “Not just because it’s an interesting question, but because it’s an interesting question that has health implications.”

News from Brown

Big data, small state: R.I. database offers the potential for efficient, improved health care

Minimizing waste, improving care

Training students

Protecting patient data

Social Navigation

Big data, small state: R.I. database offers the potential for efficient, improved health care