In the coming years, federal, state and private sector purchasers will implement a wide range of new and resurrected reimbursement schemes to effect a variety of key improvements in care.

 

 

These schemes include:

  • penalties for poor care (e.g., high levels of hospital readmissions);
  • bundles and episode payments that provide fixed reimbursement for all care required for a surgical procedure, such as a hip replacement, or all care needed by a diabetic in the course of a year;
  • reimbursement reward for care above or below national averages (e.g., value-based purchasing);
  • financial rewards for efficient care (e.g., shared savings programs).

These schemes vary in their maturity and will be introduced over the course of the decade.

To respond to these payment pressures, providers will increase their efforts to establish care processes such as care coordination and population management. And they will participate in widely discussed new care models such as patient-centered medical homes and accountable care organizations.

Providers will need to enhance their measurement and reporting so they can implement these new care models at the individual patient and population levels and manage the diverse payment arrangements. For example, ACOs will want to measure the effectiveness of care protocols, such as exercise compliance, for a population of diabetic patients. Surgical service providers will need to understand the costs and quality of proposed procedure bundles. Understanding what works and what does not is key to ensuring reimbursements, controlling costs and, most importantly, providing the best care for the patient.

As such, business intelligence and analytics will become essential in this new care delivery environment. These tools will enable providers to assess, for example, risk factors of a defined population in relation to a care protocol, the measurement of the effectiveness and efficiency of a program, and data mining to develop clinical evidence for best care practices.

The Focus and Nature of Business Intelligence and Analytics

The various payment reform alternatives (and the associated new care models) have several common themes. They all encourage:

  • a view of care as longitudinal rather than a series of disconnected encounters;
  • patient risk prediction and stratification models;
  • funding of lower-cost services that substitute for higher-cost admissions, emergency department visits or face-to-face physician encounters;
  • using lower-cost providers for tasks they are competent to perform;
  • patient self management;
  • systems of care in which information flows within and between provider organizations.

Health care reform will force significant changes in the data needs and strategies of provider organizations. Providers will need to link patient encounters by disease and to assess the care activities that occur over lengthy periods in different care settings. The data must allow providers to determine the specific care activities, e.g., medications prescribed, laboratory test results and the costs of care delivered. The finance analytics databases will need to include a much broader and deeper set of longitudinal clinical data.

Providers also may need to incorporate encounter data from outside the organization (enabled by health information exchanges) and eventually data that patients enter through personal health records. And the provider's analysis capabilities should include software that allows analysts to identify patients who are at greater risk of needing care, enabling the organization to strengthen support for those patients.

Moreover, providers will need to assess the costs and quality of alternative care settings and to model the implications of moving patient care to settings other than the hospital or physician's office. The analytics software should support an organization's intent to optimize performance, control key processes and decisions, and react to changes in the environment as well as deviations from performance.

Overall, providers will need to be more focused and aggressive in managing the organization and their patients. Changes in reimbursement will require providers to predict which patients will need extra care, determine the financial implications of changes in quality scores, assess the performance of core organizational processes such as transitions of care, determine conformance to medical evidence, and report quality measures to purchasers. Business intelligence and analytics will also need to be more real-time and accessible at the point of care.

In addition to the need for enhanced business intelligence capabilities, the "context" of business intelligence will change due to the technology, shifts in the provider business model, and the fact that administrators and clinicians are growing increasingly comfortable with information technology.

For example, consider the following:

  • Analytics from the data warehouse are going to be expressed in much more of a real-time and distributed fashion, pushing information as close to the point of decision as possible.
  • Traditional command and control structures to monitor adherence, track performance variances, and take corrective action through retrospective quality improvement and training initiatives no longer will be the only way to align an organization's goals. Information will be pushed to the front lines, guiding decisions at the point of highest possible impact.
  • Relationships and meanings can be inferred from available data, some of which are structured and some of which are unstructured, and all of which are expanding rapidly as new technologies become available to capture information. What information means, its relationships and potential combinations, will become less reliant on exactly how it is captured, extracted and stored. This diminishes the burden on clinical processes and resources.
  • As new relationships are discovered, predictive models can emerge from the data as a byproduct of machine learning capabilities, rather than as an exclusive derivative of the human investigative process and statistical analysis. Analysis technology will become more "invisible" to the end users even as they take greater advantage of analytical applications in their day-to-day workflow.

Ramifications for the BI and Analytics Management Ecosystem

To respond to the business intelligence needs that result from payment reform, providers will need to implement next-generation analytics and address the changes in the data context and in its use. As is always the case, effective use of technology requires implementing a sound ecosystem of management practices and policies, as described below.

Establishing BI governance. Data governance is the most critical contributor to the effective day-to-day use of BI. Good quality data is of greater organizational value than state-of-the-art analysis software. Data governance involves creating and continually managing the organization structures, policies and processes needed to define, control and ensure the quality of the data. Moreover, the governance function helps providers determine what types of analyses are the next focus of BI efforts.

The form and organizational location of the governance function will vary, largely driven by the scope and intent of the BI effort. If the organization is focused on using BI to improve the revenue cycle, then the governance structure may report through the finance department, and the governance steering committee will comprise stakeholders who manage operations that are key components of the revenue cycle. The governance body may spearhead changes in process and information systems to ensure the accuracy of insurance information acquired at registration and to develop a single definition of the data element "visit."

If, on the other hand, the organization is focused on using BI to measure care quality, then the chief medical officer might oversee the governance structure, and the steering committee could comprise a large number of clinicians and managers of ancillary departments. The governance body may commission system changes to improve data capture on medical error events and to define clinical best practices for treating patients with a chronic disease based on data derived from the electronic health record.

Developing data use policies. Data exchange among health care entities raises data management questions for both the senders and recipients of data. Under what conditions can data from one organization be used by another organization, e.g., for care operations assessment or clinical research? And if one organization needs to amend data it has exchanged with others, how is that amendment propagated to the various recipients?

Broad EHR adoption will open the door to secondary uses of clinical data: clinical research, care improvement, population health and post-market medication surveillance. Early efforts to use EHR-based data in these fields show promise, but also have exposed data quality issues. Providers also may be approached by data aggregators who are interested in pooling data from multiple organizations to pursue these secondary uses. The contributing organizations will need to establish policies and agreements that enable them to benefit from these arrangements, but also to protect themselves.

Defining data management practices. Data management refers to the steps an organization takes to ensure that its data have a well-understood meaning, are of good quality, are appropriately used, are protected and have potent analyses tools. Data management is very difficult. The work is not sexy — it can involve slogging through process changes, cajoling clinicians, engaging in difficult discussions on data meaning and edits, reducing multiple data silos, and fixing application software with insufficient edits.

Enhancing the quality of data will become more complex and daunting. For example, steps to improve data capture often hinder operational efficiency. This challenge will spread from registration areas to the exam room. In addition, managers are often uncertain about the quality of data in their reports and are unsure of the "source of truth." As the organization's data encompass sources beyond the organization, confidence and truth become difficult to assert.

Determining business needs and value. Managers should discuss why they want to make the investments required to establish and maintain an effective BI effort. The rationale must be more compelling than a statement such as "we need better data." It can focus on reasons such as organizational imperatives to understand cost structure, care quality or revenue-cycle performance. Regardless of the rationale chosen, the case should be clear to managers.

The value of having better analysis capabilities can be difficult to measure. It may not be obvious if an investment of several hundred thousands of dollars to understand care quality will be "worth it." BI investments often rest on the strength of management's judgment that this investment is necessary if the organization is to achieve its goals and objectives. This should not preclude efforts to measure the return on investment of BI investments. Rather, managers should understand that ROI alone will not be a good assessment of BI value.

Developing an end-to-end vision. The organization will need to develop an overall vision and road map that defines the scope of the BI effort and the plan to achieve that scope. This vision often starts with efforts to improve data collection practices at the front end, e.g., the processes used to identify referring physicians in the specialty clinics. The vision moves all the way through definitions of the composition and organizational location of the BI analysis staff. The vision will describe the types of analysis desired and the classes of users who will authorize the analyses and receive the results.

Defining target areas for BI implementation. It is not practical to implement BI technologies and disciplines throughout the organization all at once. Rather, managers should target areas for initial implementation based on business needs and value. As leaders learn from these pilot efforts, they can make decisions about the breadth and pace of BI implementation.

The Learning Years

Health care providers in the United States are entering a decade of profound change driven by reimbursement changes and care models that are intended to elevate quality and reduce costs. While there have been attempts to use reimbursement to drive change, the decade before us has no parallel in terms of the complexity, significance and potency of these efforts. They will alter the structure of health care provision, and providers will need to make investments in a comprehensive portfolio of information technology.

A critical investment will be in BI, analytics and ensuring that a sound management ecosystem is in place for successfully using the technology.

In the next few years, we will all be challenged to develop a more comprehensive understanding of analytics and BI in health care. As we rely on techniques like data normalization and consolidation, we will learn to exploit other data types and analytical techniques to provide the answers we need to deliver optimal care to the communities we serve.

The choice will not be between retrospective or predictive analytics, consolidated or distributed data, and human or machine-driven decision-making. We must do all of these.

Indeed, there will be much to learn and new knowledge to share as we move boldly into this data-rich future — a future we ought to be preparing for now.

John Glaser, Ph.D., is the CEO, health services, at Siemens Healthcare in Malvern, Pa. He is also a regular contributor to H&HN Daily.