Clean, Actionable Data Critical Success Factor for Population Health

Philips held its Connect2Care customer conference in late September in Aurora, Colorado. The event was primarily for North American customers and featured the range of Philips solutions and those of its key partners. I was fortunate to co-present with Cindy Gaines, Chief Nursing Officer for Philips Population Health for a session entitled "Scaling Care to Meet Patient Needs Across the Continuum."  The 90-minute session was designed to engage attendees in a discussion of the challenges to success and how organizations are overcoming those challenges in order to grow and mature their population health programs. 

 

The IDC Maturity Model for Population Health was presented. It describes the five levels of maturity ranging from "The Dabbler" all the way to "The Guru." The model was applied across six dimensions of population health: Vision, People, Process, Data, Technology and Finance. With each dimension discussed, the audience shared their experiences and strategies. While there were several lively discussions, one theme emerged: credible and actionable data as the cornerstone for gaining momentum in population health was raised by virtually all participants. 

 

Bringing together data from multiple Electronic Health Records (EHR) and integrating multiple claims feeds to create a comprehensive longitudinal patient record is the first and for many the greatest challenge. To create clean data, healthcare organizations or their third-party partner must have experience and an understanding of the nuances of both claims and structured clinical data.  They must understand issues of data harmonization to make sure values across EHRs are equivalent and that coding nuances have been identified and normalized. Ingesting claims data takes a similar set of skills and experience. Many a data and analytics effort has been sabotaged by exposing poor/inaccurate data. Physician participation in population health management relies on credible and actionable data.

The data analytics staff must be accountable for producing accurate data. This requires an intimate knowledge of the processes and steps that the data undergoes before it can be used for analysis. Validation of the steps and edits is critical as is the validation of results. Does the data balance with existing systems? If there are differences, why and can you explain them? Be sure to share early results with "friends." For example, are all of the patients identified as needing mammograms women over 40? If not, then why? Data anomalies, even if accurate, need to be presented, explained and documented. Get feedback early so that when you begin working with physicians directly you are prepared and know the data. Don't be impatient; this phase of the process always takes longer than expected and at times feels tedious. Virtually all attendees echoed the need for patience at this phase.

Physicians are naturally skeptical about the quality of data particularly when it is being used in performance management reporting. Don't surprise them with a report on their performance. Inform physicians early about what you are doing, and where appropriate include them in the process. Start with reports that are less threatening. For example, the issue of patient attribution is hugely challenging. You might want to start with an exploratory discussion with a small handful of physicians that reviews patients that are attributed to a physician. You may uncover data or methodology issues that are relevant to all. It helps if some of the early adopters can become evangelists. Delivering actionable data and insights is part of building credibility; hand physicians what amounts to a to-do list. Make it easy to answer the question: "What do I (the physician or the care management staff) do to reach my goals?"