Randy L. Thomas

It has often been said that health care is "data rich and information poor."

There are lots of transactional systems collecting lots and lots of data, but it is often hard to mine the data to identify trends and opportunities. Executives and administrators in healthcare organizations are often frustrated when they know the data they need are "in there," but getting to them to support planning and decision making proves difficult and time-consuming.

This situation has been getting steadily worse with the increasing adoption of advanced clinical information systems-electronic health records. These systems are designed to provide fast response time, one patient at a time. They are not designed to easily support cross patient analysis and reporting. Organizations spend tens of millions of dollars acquiring and implementing these systems with a focus on clinical transformation-improving clinical efficiency and outcomes-only to be stymied when trying to access the data to evaluate how EHR adoption is affecting the performance of the organization. The time has arrived for health care to seriously pursue methods to turn data into information.

Keys to Building a Data Warehouse

An effective data warehouse is essential to your EHR and other systems. There are four key components to building a well-organized and easy-to-use data warehouse:

  • A plan that is driven by how the data will be used, from the top down
  • A decision framework that guides prioritization and supports decisions into the future
  • An inventory of the data analytic tools and databases currently in place
  • A process for discussing conflicts and prioritizing needs

Leveraging Data for Clinical Excellence

The need for clinical, financial, and administrative analytics and reporting spans a broad spectrum of clinical and business requirements. Frequently, the focus is on quality outcome and patient safety initiatives. There is also the need to provide data for mandatory reporting to the Joint Commission on Accreditation of Healthcare Organizations (core measures) or the Centers for Medicare and Medicaid Services (quality indicators). New pay mechanisms, such as pay for performance and related initiatives, need aggregate data reported on clinical outcomes. In organizations that have a research focus, there is a growing need to combine genotypic and phenotypic data with clinical operations data to understand how genetic profile, clinical process, and therapies affect outcomes. This analysis can, in turn, lead to discoveries of new and improved approaches to clinical care.

This problem of leveraging the vast store of clinical and other data to support improvements in clinical excellence is not solved by simply dumping all the data into a giant data warehouse and giving folks (appropriate!) access with a reporting tool. More than one CIO has referred to this process as creating a data "landfill." The data warehouse might contain everything-including the kitchen sink-but the data are not necessarily organized in a way that the warehouse can be used effectively. People trying to run reports to analyze a problem-why a particular contract is not as profitable as anticipated, for example-have to sift through a lot of useless stuff before finding what they need to answer their question.

Bringing together disparate data from multiple sources across the organization requires a plan that is driven by how the data will be used. In other words, the data warehouse needs to be built on the foundation of the clinical and business issues the organization intends to address. The issue of building an analytics environment needs to be approached holistically, from the top down, keeping in mind the overall clinical and business objectives of the organization. This will result in an integrated, enterprise informatics plan that will carry the organization through a multiyear design, build, and execute process.

Often, organizations begin the process of building a data warehouse by focusing on the technology it will use. Choosing database tools becomes the focal point. But before a data model can even be started, many issues around governance structure and process about how to manage the data must be put in place. The organization needs to define a knowledge management process. Not all needs can be met simultaneously. New sources of data and new needs for information will continually arise. A process for discussing the inevitable conflicts and prioritizing needs must be in place before an organization can begin the more tactical process of creating an appropriate analytics environment.

Taking Inventory

Once a knowledge management process is put in place, the organization should inventory all the various data analytic tools and databases currently in place. Understanding what exists and how it is used adds additional information to the overall integrated informatics plan. These existing tools can also be leveraged in the building of the new enterprise analytics environment.

Through this planning stage, which typically takes an organization three to four months, a decision framework needs to be defined that not only guides prioritization and decisions now, but helps support decisions into the future. This decision framework should be composed of a set of "principles" that describe the needs and aspirations of the organization around analytics. It should reflect the clinical and business goals and objectives of the organization. The decision framework should indicate things such as the balance between departmental or facility autonomy and enterprise objectives, the prime drivers behind an integrated informatics initiative, and the balance between clinical care delivery efficiency and data needs for research.

Armed with a decision framework, an inventory of current capabilities and a consensus of priorities on the clinical and business needs the analytics environment will support, the organization can then determine how to get from "here" to "there." This is the plan or "road map" the organization will follow during the design, build, and execute process.  This process will result in an orderly data warehouse appropriate to the goals of the organization and the end-user analytic tools that will readily support the clinical and business queries that are difficult to answer now. While it could easily take five years to fully build out the analytics environment, if properly planned, "first productive use" should come about in 18 months or so-with a focus on meeting the most pressing needs first. With the foundation of the data warehouse finished, subsequent needs should then be addressed every few months, according to the original plan.

Of course, health care is not static. New conditions arise and needs change. Your informatics road map will likely need to adapt over time to meet these changing conditions. This is where the value of the decision framework comes into play again. As new issues come up, you can use the decision framework to "course correct" within the construct of your original objectives-knowing specifically what you are changing, why you are changing it, and what effect it will have on your overall plan.

Mining the Gold in Your IT System

There's "information gold" trapped in the "data hills" of your transactional systems. A thoughtful, focused planning process can help you mine these data, providing valuable information your organization can use to continue to transform care and improve clinical and business outcomes.

Randy L. Thomas, FHIMSS, is an associate partner, Healthlink, a division of IBM, part of IBM Global Business Services, Marlton, N.J. (thomasra@us.ibm.com).

Publication Date: Friday, September 01, 2006

Login Required

If you are an existing member, please log in below. Username and password are required.



Forgot User Name?
Forgot Password?

If you are not an HFMA member and would like to access portions of our content for 30 days, please fill out the following.

First Name:

Last Name:


   Become an HFMA member instead