Trusted Data Is the Foundation for Advanced Analytics
By Vicky Mahn-DiNicola RN
Much has been said about using advanced predictive analytics to improve the quality of healthcare. But one thing not receiving the attention it deserves is the pre-requisite of trusted data being sewn into the fabric of the healthcare organization. Every organization has data at its fingertips, but full value of that data can only be actualized if it is properly understood and trusted.
Take a relatively straightforward data element like a patient’s weight. While it is a simple, basic element, it can create havoc for analytics teams who discover there are upwards of 17 different places in their HIT systems where weight is captured. Weight is recorded in the emergency department flow sheets, nursing assessment intake forms, pharmacy profiles, ambulatory clinic records, and daily critical care flow sheets, just to name a few. Determining which weight field is the most reliable and appropriate to use is a difficult, lengthy process and one that is multiplied by hundreds of data variables required in advanced analytics projects.
Healthcare organizations are excited by the brilliant technology coming our way in the form of genomics, mobile health, and telemedicine. But too often, the cart is put before the horse. Just as bad ingredients guarantee a bad meal for even the best of chefs, unreliable data in healthcare will inform inaccurate, even dangerous decisions.
Effective use of analytics is not something you can buy off the shelf from a vendor. Rather it is an organizational strategy, structure, and culture that have to be developed over time. While the technical and tactical execution is delegated to others, the chief executive in a healthcare organization is responsible for determining and overseeing this direction and progress.
The executive also needs to align the organization with data cooperatives and national groups that promote data standardization. National standards have historically been ambiguous, so it is important for providers to ensure they are not working in a vacuum, but have a common understanding of national guidance.
Diversity of systems and processes breeds confusion. Because there are many ways to express any given concept, there is a need for robust crosswalk, data mapping, and standardization to ensure data integrity within, between, and across organizations. This body of work is the responsibility of a designated data governance body within an organization.
Data governance implies far more than the maintenance of documents that describe measurement plans and reporting outputs. It is a comprehensive process of data stewardship that is adopted by all data stakeholders across the organization, from the board room to the bedside. Data governance is critical in order to standardize data entry procedures, reporting outputs, clinical alerts, or virtually any information that is used in clinical and business decision-making. In the era of pay-for-performance and risk-based care, data standardization is mission critical for a true, accurate comparison to take place when evaluating an organization’s performance against external benchmarks and determining reimbursement based on value.
Another final step toward creating robust data governance structures is to create a data validation process. Data cleansing and maintenance should be automated, centralized, and transparent across the organization and should be designed to accommodate the needs of both clinical and business stakeholders.
A “data librarian” should be appointed to catalogue and oversee data elements across the healthcare system. The most mature organizations will implement a master data hub that is fully integrated into their application system environments so that changes are made simultaneously to all systems that need the same data. By doing so, a simple element like a patient’s weight will always be consistent in HIT systems.
Organizations need to recognize that the advanced analytics of tomorrow will only be achieved if the data we have today can be trusted. Those who succeed in establishing proper data governance will unlock the full value data can provide in our industry, beyond regulatory reporting and retrospective benchmarking initiatives to the more exciting prospects of predictive and prescriptive analytics.
Vicky Mahn-DiNicola RN, MS, CPHQ is VP of research and market insights with Midas+ Solutions, A Xerox Company.