The concept of big data is getting a lot of buzz in business journals and especially marketing circles. Healthcare leaders should pay attention, because big data can drive improvements in care processes, delivery, and management.
At a Glance
- Big data is a concept that is being widely applied in the retail industries as a means to understand customers’ purchasing habits and preferences for follow-up promotional activity.
- It is characterized by vast amounts of diverse and rapidly multiplying data that are available at or near real-time.
- Conversations with executives of leading healthcare organizations provide a barometer for understanding where the industry stands in its adoption of big data as a means to meet the critical information requirements of value-based health care.
“The more companies characterize themselves as data-driven, the better they perform on objective measures of financial and operational results.”
—Andrew McAfee and Erik Brynjolfssona
The advent of big data in health care marks a critical turning point for healthcare providers: Those that seize opportunities to use big data in improving value will become market leaders, and those that do not will likely be left behind.
Big data is a concept that is generally used to describe vast amounts of diverse data, both structured and unstructured, that organizations can access quickly and analyze using innovative new tools that help pinpoint opportunities to better manage and improve value. As providers learn how to harness this resource, they are likely to see a tremendous increase in the potential applications of data for transforming care delivery.
Today’s healthcare leaders are recognizing the critical role that data will play in the transition toward value-based business models, where value is defined as the relationship between quality of care and cost. But what are the opportunities for using big data—and how should providers begin to seize them? Healthcare organizations will certainly be able to use big data to develop care protocols and explore population health management. Big data also can help providers target patient engagement initiatives that can help them control costs and improve outcomes. Eventually, systems will be able to deliver information in real-time to caregivers based on predictive models utilizing big data.
These are only a few examples, and countless others will emerge as healthcare providers not only apply the lessons learned from other industries in how to make use of big data, but also discover myriad ways that big data has relevance uniquely for health care. Here, we provide insight into ways healthcare organizations are pioneering the use of big data in meeting the challenges of health care.
Big Data Defined
In their previously cited article, McAfee and Brynjolfsson describe three key attributes of big data:
- Volume—much greater amounts of rapidly multiplying data than were ever previously available
- Velocity—data at or near real-time
- Variety—many sources of data, including some that are relatively new, such as information from social networks
For marketers, the process of refining how best to target, or segment, customers has evolved through several generations of approaches. In the early years of segmentation, marketers tended to base their strategies on observations of simple, one-dimensional groups of individuals with some common characteristics. Over time, segmentation became much more multidimensional, with organizations eventually realizing they could run tests with individual customers. (For example, an organization could reach out to a customer with various promotional offers in succession until the customer accepted an offer, thereby giving the organization a basis for further promotions.)
Using big data goes beyond testing to include the development of sophisticated models that can anticipate what a customer needs or will accept. It also provides a powerful means to reinforce predictive modeling and enhance decision support.
In short, it is a more sophisticated basis for segmentation than tools used in the past—analogous to the difference between a standard microscope and an electron microscope. It facilitates a more precise assessment of circumstances to allow for more refined and effective interventions.
Big data also can be applied to a wide range of issues. Consider just a few examples of potential applications in health care:
- To identify populations and health situations in which a phone call or visit is most likely to prompt a change in behavior that will lower costs and/or enhance future health status
- To determine when behavioral or lifestyle data can be added to physical data to improve an outcome
- To evaluate which collection approaches are most likely to be effective, thereby also improving ROI of the collection process
- To predict the future healthcare needs of a population, thereby allowing the healthcare organization to tailor actuarial estimates of costs, future staffing, and other resource utilization decisions for this population
The range of possible uses in health care is almost infinite—from diagnosis, to treatment, to forecasting of utilization, to population health management, to capital and strategic planning.
Lessons from Other Industries
Other industries have advanced much further than health care in applying big data for strategic benefit. The concept has already captured the imaginations of marketers in the retail industry, for example, and the way major retailers have used big data can provide a lesson for healthcare organizations. For example, Target and the online retailer Amazon use predictive modeling driven by individuals’ past purchases to provide specific coupons or product recommendations to carefully segmented customers.
In The Power of Habit: Why We Do What We Do in Life and Business, Charles Duhigg traces the development of big data in the marketing strategies of organizations such as Target. The work began with a simple question, “How can we identify pregnant women before other marketers do?” The answer is, by mining the organization’s own data.
Current Applications in Health Care
To determine the extent to which big data is being applied in health care today, we talked with senior executives at a number of leading organizations, who shared their ideas about how they will use data to position their organizations for the future. All of the leaders acknowledged that their organizations’ activities in these areas are in their formative stages.
Four areas of focus emerged from the conversations as being important to an organization’s early efforts to apply big data, particularly for predictive modeling:
- Establishing and implementing data governance/data discipline
- Assembling the quantities of data required
- Combining data sources and integrating them into predictive models
- Applying approaches in multiple directions, leading to different decisions
Providing access to the sheer amount of data required for such efforts, and getting it to the ultimate user or decision-maker, is a particular concern. Tarek Elsawy, MD, chief medical officer for the Cleveland Clinic’s Community Physician Partnership and Quality Alliance, notes: “Data are the currency that drives improvements. The ultimate goal is accurate and actionable data—not data that are three months old, but data that can be made available to providers in real time. The ability to integrate large amounts of clinical,financial, and demographic data will get us much closer to that goal.”
Michael Apkon, MD, PhD, senior vice president, medical affairs, and chief medical officer of the Children’s Hospital of Philadelphia, offers a similar assessment: “We want not only a larger patient database, but also a database that encompasses the largest possible genetic diversity. Then we have to use it in predictive models and translate it to the caregiver.”
Evolving Use of Big Data in Health Care: Case Examples
A closer look at the current strategies of five organizations provides interesting insight into the types of strategies healthcare organizations have pursued, to date, to make use of big data.
Baptist Health South Florida. Ralph E. Lawson, FHFMA, CPA, executive vice president and CFO of Baptist Health South Florida in Coral Gables, Fla., sees tremendous power in the ability of data to transform care delivery. Lawson points out that a key early stage in the journey to using big data is “effective data governance”—that is, defining data elements to be consistent across all systems (clinical, financial, other). Baptist has recently appointed a data governance committee, headed by a vice president, to undertake this “difficult and slow” work. The health system applies predictive modeling in its experiments with value-based payment methods, including bundled payments and shared savings arrangements.
The organization is participating in an ACO-like payment arrangement for oncology services with a payer and a local physician group. As part of this arrangement, the payer is providing longitudinal claims data that enable Baptist Health to analyze patient utilization both inside and outside of the system. For Baptist Health, access to this type of information is unprecedented and, according to Lawson, incredibly valuable. Baptist Health anticipates publishing results of this arrangement within the next few months. The keys to success? “Focus, and access to claims data,” Lawson says.
Methodist Health System. Linda K. Burt, CPA, CFO of Omaha, Neb.-based Methodist Health System, says her organization is initially applying the concept of big data related to population health management, with an emphasis on capturing data from both the hospital and medical group EHRs (electronic health records) and claims data from its largest payers. The evaluation of opportunities to offer bundled services is another area where the use of big data has been critical, she says.
In studying ways that a big-data approach could be applied at Methodist, Burt and her colleagues have come to recognize the health system’s need for information from external sources in addition to what it generates internally. The additional information needed not only includes data from physicians and claims data, but also could eventually include demographic data and the findings of intensive market research of patients, their needs, and their experiences with physicians and hospitals.
Working through a jointly owned accountable care organization (ACO), Methodist and a local competitor have selected a population health management tool to analyze historical utilization (and monitor current utilization) within their employee populations. The two health systems have decided to begin their journey jointly toward management of the health of a population with their own self-insured employee groups. Although the organizations currently work with different third-party administrators (TPAs), both TPAs will be providing historical claims data to be analyzed within the population health management tool. An initial assessment, including comparisons with actuarial benchmarks, will provide directional opportunities to reduce costs. The medical management committee of the ACO will be the key user of the information produced by the tool. The opportunities identified by this tool will be used to form new work groups or to add focus and priority to existing initiatives, including focuses on ambulatory sensitive conditions, the Choosing Wisely campaign, and emergency department utilization.b
Billings Clinic. “We don’t use the term big data,” Nicholas Wolter, MD, president and CEO of the Billings Clinic in Billings, Mont., says. “However, our overall data strategy is to bring together data elements from different systems to provide information that lets us think differently about strategy.”
Billings Clinic has used an EHR for many years. Currently, the clinic is installing a data warehouse and a decision support tool and is evaluating options for systems that “attach to” its data warehouses to allow more real-time access by analysts and clinicians.
Billings Clinic recently became sole owner of New West, a Medicare Advantage plan. The organization’s leaders see ownership of the health plan as a strategic investment in data capabilities, which in turn support population health management. “We are hoping the ability to look at clinical and claims data with this population will help us manage the plan and grow the membership,” Wolter says.
Billings Clinic has pursued other opportunities to augment its clinical data to drive improvements. For example, the clinic participates in a project with the American Medical Group Association in which Billings Clinic submits data and receives a larger set of de-identified patient data in return. These data are beginning to help the clinic profile its results with high-volume and complex conditions and compare the results with results and outcomes of other group practice and integrated systems
Geisinger Health System. Kevin F. Brennan, FHFMA, CPA, executive vice president, finance, and CFO of Danville, Pa.-based Geisinger Health System, describes big data as “a giant data warehouse that is a comprehensive depository of clinical information from physicians, group practices, hospitals and health plans.” He notes that all of these constituents feed the data warehouse, allowing for “infinite possibilities for data mining and research.”
The value to Geisinger, Brennan says, is in being able to extract data in creative ways to advance business interests.
Relative to many other healthcare organizations, Geisinger is already advanced in its data-related capabilities. It is renowned for its ability to use data from its EHRs and costing systems to identify opportunities for clinical and cost improvement, to standardize care protocols, and to promote the consistent use of those protocols across the system.
As an integrated system with a health plan, Geisinger has decades of experience in population health management. However, Brennan notes that even an organization as advanced as Geisinger still has gaps in the data required for population healthcare delivery. As an example, Brennan points out, “Sometimes patients’ underlying problems emanate from behavioral health, and behavioral health tends to be carved out by insurance companies for separate management by them or their subcontractors, leaving a gap in the patient’s medical EHR.” This disconnect makes it much more difficult to access the data needed to address the problems. Another example relates to pharmacy data: “At Geisinger, we are undertaking a major reorganization to better link pharmacy data to advance our knowledge and performance,” Brennan says.
Brennan sees tremendous promise in big data: “Imagine marrying gene typing with all of the clinical information we have. In the future, we may be able to customize care protocols based on a patient’s unique genetic makeup.”
Already, Geisinger participates in arrangements in which it sells de-identified data to pharmaceutical companies. The companies provide Geisinger with data specifications. Brennan notes that the pharmaceutical companies find great value in having the longitudinal data because the data help them more quickly identify what works and what does not work, thereby reducing research and development time. “This could transform how clinical trials are done,” Brennan says.
Kaiser Permanente. Murray Ross, PhD, vice president, Kaiser Foundation Health Plan and director of the Kaiser Permanente Institute for Health Policy in Oakland, Calif., notes that, like other organizations, his organization is working to define big data. Kaiser Permanente leaders have engaged in internal discussions about what it means for their organization. In a cross-functional meeting of the IT department, the organization’s Care Management Institute, and other stakeholders, the group generally agreed that big data extends beyond the clinical data that Kaiser Permanente captures in its EHR (called HealthConnect). The group viewed the concept of big data as involving the ability to combine data sets and mine the data real-time.
Among healthcare organizations nationally, Kaiser Permanente is arguably one of the most advanced in terms of data capture. The health system has used the same EHR for many years. In certain pockets of its organization, Kaiser Permanente is able to combine databases of genomics, EHR, and census data. The organization has developed panel support tools that marry clinical guidelines with the EHR. The panel support tool can help a physician see whether his or her patients have carried out recommendations and more closely evaluate patients who do not have chronic disease who may have gaps in care.
In Ross’s view, having data that are truly “big” involves much more than simply “adding an additional characteristic about a patient.” Instead, it means “having enough data that it leads to new types of decision-making.” As an example, it might involve adding as much data on individual patient preference as possible into the system. Once aggregated and studied, such information could eventually promote new decision-making.
Next Steps for Health Care
Linda Burt of Methodist Health System describes the organizational implications of big data and predictive modeling succinctly: “The future is in making this investment; if we don’t, we won’t be here.” With this view in mind, Methodist’s board and senior management have shown a clear willingness to invest in big data. Methodist is working with outside firms that provide specialized information and predictive analytics.
Similarly, a payer interviewed for HFMA’s Value Project noted that any organization that can determine how to collect and act on patient data “will be the successful entity.”
To enter the world of population health management, genetic knowledge leaps, and patient-driven health care without the benefit of technological advancements to support such efforts seems foolhardy. Healthcare organizations should be asking important questions regarding how best to make use of those advancements:
- How should a healthcare organization gain access to the volume and diversity of data required? For instance, should it be through partnerships, affiliations, vertical integration, or contracts with vendors?
- How should an organization ensure that it has the discipline and integrity within its own data to be able to combine those data meaningfully with data from outside the organization?
- Given the almost limitless number of applications, and the significant cost and time of pursuing each, where is the biggest ROI for an organization in developing big data capabilities? And what are the biggest risks in not developing such capabilities?
- How can an organization stay ahead of what is likely to be a fast-changing opportunity for improvement in population health?
As useful as big data has been in marketing—for firms such as Target and Amazon—some of the longest reaching uses of big data may well be in health care. As healthcare information systems link more effectively with one another, as health systems share data with one another more globally, as predictive modeling in health care matures, and as information based on predictive models is delivered real time to decision-makers, we are likely to look back and say today’s uses of data were only a crude beginning for evidence-based decision-making.
Keith D. Moore is CEO, McManis Consulting, Denver, and a member of HFMA’s Colorado Chapter.
Katherine Eyestone is a senior consultant, McManis Consulting, Denver, and a member of HFMA’s Colorado Chapter.
Dean C. Coddington is a senior consultant, McManis Consulting, Denver.
a. Andrew McAfee, A., and Brynjolfsson, E., “Big Data: The Management Revolution,” Harvard Business Review, October 2012.
b. Get more information about the Choosing Wisely initiative.
Publication Date: Thursday, August 01, 2013