To Get Population Health Management, Providers Need Big Data Chops

There are good reasons why population health management, or PHM, is such an important trend in healthcare. By viewing patients in groups with related issues and conditions, it’s possible to anticipate problems and find more cost-effective ways to address them, improve outcomes, and save money at the same time. PHM programs typically pay for themselves inside four years.

But a provider can’t just jump into the practice. It takes specific knowledge and skills — data analytics being one of the biggest needs. PHM only works if you can analyze data about apparently similar people and then find the commonalities and connections between environment and culture on one hand and health outcomes on the other. That’s a big data problem, both in the sense of “big data,” using information from inside and outside the organization, and a big problem to address.

For example, one of the difficult steps is to determine which healthcare data is important for the goal. An old precept of strategic planning is the importance of knowing what data to consider and what to throw away. For example, if you’re about to move into a new market or area of activity, historic numbers may mean nothing because they’re about something entirely different.

HealthIT Analytics suggests five types of data that will likely be useful:

  • Claims data will, or should, have demographics, diagnosis codes (for treatments histories), dates of service, and costs, but may be old.
  • Electronic healthcare record data has details of care beyond diagnosis codes, includes important medical history information, however, much of the material may be unstructured data in text blocks that make analysis difficult.
  • Environmental and socioeconomic data can be critical, but rarely appears in EHRs. System modification may be necessary to begin collecting the information.
  • Patients may provide information as well through portals or devices, whether a smartphone or actual medical device. Again, not all systems are structured to use such information, or to accept direct data transfers, in which case some custom development might be needed.
  • Medication adherence data, to see how thoroughly chronic disease patients follow a regimen, is likely found in some of the above data types, possibly with the addition of external data sources, but also will need work to integrate.

One of the first analyses of data should include identification of patients in the highest treatment spending brackets. A report from the Health Care Transformation Task Force notes that the top five percent of patients by spending will, in total probably represent half of money spent. Although eventually more extensive analysis will be necessary, focusing on high spending will help identify areas that can provide a solid return on investment.

Are you interested in finding a rewarding and lucrative healthcare career that fits your individual strengths and interests? Find out how education can help you adapt to the changing healthcare landscape. American Sentinel University is an innovative, accredited provider of healthcare management degrees, including an MBA Healthcare and Master of Science Business Intelligence and Analytics.