Cutting it Fine: How Granular Data Can Help the Bottom Line

Making assumptions is fundamental. It can also be dangerous. Assumptions are at the heart of financial and program planning in healthcare organizations, underpinning revenue projections, business acquisitions and new program design. We use them when data isn’t available in the form in which we need it. There’s always a risk, however, that using an assumption instead of the detailed data will lead us astray. “Averaging in” small differences can mean the difference between profit and loss, especially given the slim margins of many healthcare organizations.

THE DEVIL IN THE DETAILS

DGA has seen hospitals ready to accept a new payment system based on their analysis of average payment and fees, only to find that analysis of detailed patient-level data revealed they would forego significant revenue if they accept the new system. Healthcare Financial Management recently featured a story on a hospital that initially couldn’t improve collection performance because its existing data systems didn’t provide key indicators at the payer level. The value of highly-developed analytical capabilities is becoming more widely recognized with the recent publication of books like Competing on Analytics and Super Crunchers. Fortunately, hospitals and health systems do have a better and relatively simple way of generating more detailed internal data to support analyses that may impact the bottom line.

PLANNING AHEAD FOR SIMPLIFIED ACCESS

The solution is a system that provides easy access to very “granular” data – data at the lowest level possible. By creating such a system in advance, you gain the capacity to ask and answer questions quickly as they arise, selecting and structuring the data elements (e.g. service date, procedure code, payer, department) required for each specific analysis. When the answers received generate new questions, you can drill down into the data for rapid response. Hospital data systems rarely accommodate these analytical needs. Turf struggles, lack of data user skills, software limitations and “language barriers” between IT and users can all be obstacles to obtaining detailed data. Users may believe that current hardware and software can’t handle the required volume of data. Where needed data can be accessed, it is typically a time consuming process, and answering the next question is just as onerous. A well-designed data warehouse populated with commonly-used data eliminates these barriers. It is a tool created specifically to facilitate rapid analysis of granular data, through easy, consistent and well-defined data access, including the capacity to “drill down” into data as each answer generates its own questions. While the data warehouses of a retail giant may handle terabytes of data uploaded daily, health system needs are more modest, requiring gigabytes of data in relatively static data sets.

A DATA WAREHOUSE THAT WORKS

The design of a data warehouse is critical. It must consider both the proper handling of the underlying data sources and the needs of the data analysts. It must present the data to the analyst in a usable format, yet also provide for the multitude of IT processes (e.g. matching header and detail claim lines; handling reversals and voids) that are necessary for accurate data. All data should be labeled in an “analyst-friendly” format (business terminology, not IT system “field names”). It is vital that the warehouse be designed by people who are intimately familiar with both raw healthcare data and the requirements of healthcare financial and business analysts.

CHOOSING THE TOOLS

A healthcare data warehouse may be:

Part of a comprehensive healthcare IT system
An add-on product designed to extract data from existing systems (e.g. billing, laboratory, medical records) and provide analytical capabilities
A custom-designed solution highly tailored to a particular user’s needs

Each approach may be appropriate under various circumstances. The custom approach may offer high value, as it presents a comprehensive solution while making use of off-the-shelf hardware and software that are easily integrated into the organization’s operations. For example, DGA Partners utilizes Microsoft SQL Server and other related products in designing our data solutions. Surprisingly, a custom solution may also be lower in cost than a vendor’s “bolt-on” system. One advantage of a customized solution is that it is easy for analysts to adopt. Many vendor-based systems utilize proprietary interfaces that must be learned by analysts. DGA has found that using more standard tools (like Microsoft Excel pivot tables), allows analysts to seamlessly integrate data analytics into their other analyses.

PRE-PROCESSING FOR FASTER RESPONSE

It’s important that a data warehouse utilize On-Line Analytical Processing, or OLAP technology. While standard relational databases perform calculations when they are requested by the analysis, OLAP databases pre-aggregate data beforehand, making answers available even before the questions are asked. This generally reduces the response time for queries from minutes to seconds and greatly facilitates the "stream of consciousness analysis” that is so important in getting to the right answers. An OLAP-based data warehouse that lets end-user analysts work with familiar tools like Excel is simple to implement. It is an approach that lets any hospital or health center have the competitive advantage of using real, detailed data instead of assumptions.
¹ Healthcare Financial Management, ”UPMC’s metric-driven revenue cycle”, Hammer, Langford and Riefner; September 2007 ² Competing on Analytics: The New Science of Winning, Thomas H. Davenport and Jeanne G. Harris, Harvard Business School Press, March 2007 ³ Super Crunchers: Why Thinking-by-Numbers Is the New Way to Be Smart, Ian Ayres, Bantam, August 2007