Turning big data into better health outcomes

Population health management is a multifaceted, many-layered endeavor that nevertheless has a common theme: the need for data and the ability to mine it for actionable information.

A broad spectrum of health care players -- individual providers, hospital systems, payers, local public health departments and federal agencies -- are all in some way addressing population health management. The approach involves identifying populations, assessing their disease status and developing appropriate responses, such as management programs for chronic diseases. Those activities require access to data -- and plenty of it.

"You can't do population health management without data," said Fred Goldstein, interim executive director of the Population Health Alliance.

Dr. Jon White, director of the health IT portfolio at the Agency for Healthcare Research and Quality, said data -- or the lack thereof -- influenced his decision to leave his medical practice and join the government. AHRQ's work in population health includes evaluating hospital safety via data analysis.

"It became painfully obvious that having the right information at the right time is really important for delivering great care," White said. "I didn't have the information where I needed it when I was in practice."

The good news for organizations pursuing population health is that they have more data than ever. The Centers for Medicare and Medicaid Services have provided financial incentives for the adoption of certified electronic health record (EHR) systems by physicians and hospitals through its meaningful-use initiative. Information formerly confined to paper charts is now in electronic form, and a new generation of wearable health-oriented devices promises to generate another stream of data.

However, privacy and security considerations complicate the collection of data, and technology also contributes to the problem. Although EHRs free data from paper records, they can inhibit the aggregation of data across medical providers using different systems. Other issues are quality and the "normalization" of data so analysts can make meaningful comparisons. And then there's the need to build an infrastructure capable of crunching all those numbers.

Why it matters

Population health management is a key element of the Obama administration's efforts to reform health care. The Medicare shared-savings component of the Affordable Care Act (ACA), for example, focuses on Medicare beneficiaries as a population. Under the law, accountable care organizations -- groups of providers who coordinate the care of Medicare patients -- receive a portion of the savings that result from better quality and lower costs. According to the law, the program "promotes accountability for a patient population."

In general, population health management dovetails with the ongoing shift from reimbursing providers for the number of procedures they perform to paying them based on the value they deliver.

"Part of that value is measuring your ability to manage the health of the population that has been assigned to you," said Cynthia Burghard, research director for accountable care IT strategies at IDC Health Insights.

In a report released in April, IDC Health Insights said the increasing interest in population health and data analytics is also being influenced by the objectives of the "Triple Aim," which the Institute for Healthcare Improvement defines as improving the patient experience of care, enhancing the health of populations and reducing the per-capita cost of care.

More than 75 percent of health care costs can be attributed to chronic conditions such as diabetes and heart disease, according to the Centers for Disease Control and Prevention. White called chronic diseases the foundational problem in health care and said chronic disease management and population health programs share a common IT remedy.

"The tools we need to more effectively attack it are the same ones we need for population health," he added.

The Department of Veterans Affairs is among the federal agencies building IT systems for population health management. Last year, VA's Business Intelligence Service Line consolidated regional data warehouses into a central Corporate Data Warehouse. VA uses the data to identify high-risk populations that need extra care and examine readmission rates, among other activities, said a spokeswoman for Microsoft, which is working with the VA on the project. The Corporate Data Warehouse uses Microsoft's SQL Server, business intelligence tools, Windows Server and System Center management tools.

Much of the warehouse's holdings stem from the VA's EHR system, called the Veterans Health Information Systems and Technology Architecture. But other sources contribute as well. Sean Murphy, chief technology officer for federal health care at Microsoft, said the VA's strategy is to "bring together many disparate data sources...to enable reporting and analytics across many differing and complementary domains of data." The VA's data store contains more than 500 billion rows of data, he added.

Other population health efforts include the Healthy People program managed by the Department of Health and Human Services. Healthy People provides a set of national 10-year objectives for boosting the health of Americans. An HHS spokesperson said the objectives serve as a framework for public health activities nationwide.

The current iteration of the program, Healthy People 2020, covers 42 health topics, including one labeled "health communication and health IT." That topic area seeks to use communication strategies and health IT "to improve population health outcomes and health care quality, and to achieve health equity," the spokesperson said.

AHRQ, part of HHS, also pursues population health management with an emphasis on data. The agency works with a group of patient safety organizations to collect data on safety events in hospitals and other health care settings, White said. Researchers aggregate and analyze the data to identify concerns related to devices and medications. In one case, AHRQ's analysis found a problem with bloodstream infections caused by bacteria traveling down the catheter used to administer drugs and nutrients to patients -- so-called central line infections.

"With the data we collected, we were able to recognize that was a big problem and figure out an intervention," White said.

As a result of the research, the AHRQ-funded Keystone ICU project advanced procedures for the safe handling of central lines, and the infection rate plummeted, he added.

The fundamentals

The Population Health Alliance has created a conceptual framework that breaks the population health management process into four stages: health assessment, risk stratification, patient-centered interventions and impact evaluation. The assessment phase looks at the health risks and disease status of a population. Based on the assessment, the population is categorized as low, moderate or high risk, Goldstein said. Interventions are then designed for individuals based on their risk category. The final step measures the clinical and financial impact of those interventions.

Market watchers are already seeing an increase in IT spending triggered by the push for population health management. The IDC Health Insights report describes the field as an emerging and active technology market.

"The requirement for managing a population is really driving the technology investment," Burghard said.

Population health management technology offers three levels of functionality. First is an analytics capability that lets organizations identify patients who would benefit from participating in some form of care management program. Second is a workflow function and the ability to create personalized care plans for individual patients, Burghard said. And third, a population health platform provides a communications component that enables organizations to communicate with individual patients and the wider community.

Specialized vendors and payers have the most experience in setting up population health management systems, Goldstein said. Vendors include Accenture, Healthways, U.S. Preventive Medicine and Viridian. Payers operate their own systems or partner with vendors.

EHR suppliers are joining those traditional players and adding population health management features to their products, Goldstein said. Health care providers are also creating systems to manage populations. The VA, for example, is partnering with vendors such as Microsoft to establish its data analytics and population health infrastructure.

The hurdles

Population health management faces a few obstacles, one of which is privacy. Any organization amassing large stores of health data is bound to invite scrutiny.

Indeed, a JASON group report created for AHRQ titled "A Robust Health Data Infrastructure" cited concerns about the use of health data as a major challenge. JASON is a government advisory group administered by Mitre.

"There is a natural tension between the private and public use of health-related data," states the report, which was released in November 2013. "Individual patient health data are sensitive and therefore must be carefully safeguarded, whereas population health data are a highly valuable, and largely untapped, resource for basic and clinical research."

The availability of health data is less of a barrier today given the wider adoption of EHRs. But problems persist when health care organizations attempt to aggregate that data.

"There are so many [EHR] vendors out there [that] you still have to figure out how to link these together on a broader level," Goldstein said.

But for Burghard, the underlying technology of population health management is relatively straightforward. She said the greater difficulty lies in "getting good data in and getting actionable insights out" of a population health management program. She cited the example of laboratory systems that calibrate test results differently; organizations must normalize the data from different labs to make sure they are comparing apples to apples.

"The complexity of that data and the quality of that data become a real challenge," Burghard said.