John Showalter, MD, MSIS

Dr. John Showalter is an influential thought leader on the innovative use of health information technology, devices, and data to drive improvements in healthcare delivery. Dr. Showalter serves as the Chief Health Information Officer at the University of Mississippi Medical Center. He received his B.S. in biomedical engineering from Columbia University and his medical degree and a master of information systems in health care delivery and management from Penn State University. In addition to his board certification in internal medicine, Dr. Showalter became one of our nation’s first physicians to become board certified in clinical informatics in 2013.

Those who try to transition to value-based reimbursement by just working harder are very likely to fail. Changes in the demographics and disease burden are already causing health systems to struggle financially. The population is aging, chronic disease is increasing and the promise of genetics-driven personalized medicine is at least 5 to 10 years away—but value-based reimbursement is here now and expanding.

Providers will need to provide better care at lower costs by using resources more efficiently and effectively. This means that they will need to get the right care to the right patients at the right time. Fortunately, we don’t need to wait for genetics-driven personalized medicine to accomplish these goals. To tailor healthcare for target populations and individuals today, we can use data-driven methods such as:

  • Predictive analytics to identify those most at risk for becoming ill or having a significant decline in their health
  • Geospatial analytics to determine “hot spots” that indicate socioeconomic and environmental factors impacting the delivery of care and development or progression of disease.
  • Patient cohorts to facilitate population management.

Predictive analytics create the most clearly actionable knowledge and need to be adopted by all organizations that are transitioning to value-based models of reimbursement. They can be used to identify patients at greatest risk for readmissions, pressure ulcers, major declines in health, and many other important clinical events. They also can be used to determine which patients are likely to engage in an intervention or even pay their bills. These predictions aren’t tomorrow’s technology. They are available today and are being used by innovative health systems across the country.

The three greatest challenges of using predictive analytics are: convincing clinicians and administrators that predictions are accurate and not science fiction, getting clinicians to trust the prediction when an established risk stratification exists, and enabling action to be taken on the prediction.

Convincing end users that predictions are not science fiction can be time-consuming, but it isn’t complicated. You can simply prospectively validate the predictions. Run the predictions in the background while a patient receives care and track the outcomes of interest. It will be clear if the prediction is accurate and actionable. Once you have the data on your population, you can prove that it isn’t science fiction. That said, getting clinicians to incorporate predicative analytics when there are validated models for risk stratification can be complicated.

The primary difference between established risk-stratification methods and predictive analytics is the source of data. Most risk-stratification methods use 5 to 15 clinical or demographic parameters to determine risk. Predictive analytic platforms use a layered deep-machine learning approach that incorporates thousands of variables including socioeconomic determinants of care and environmental factors. By enhancing the clinical data set with thousands of variables that help tell the more complete story of the patient’s life, analytics platforms have been able to demonstrate superior performance to classic risk models. But the algorithms are not published and/or endorsed by physician organizations. Transitioning clinicians to use these new approaches requires changing culture, getting them to trust the approach, and guiding them to develop appropriate interventions based on the predictions.

Once the interventions are determined, the challenge is that clinicians need to be able to take action on the predictions. Few organizations have fully integrated predictive analytics to enable consistent action. Ideally, predictions are available within a clinician’s workflows with minimal barriers to acting. For example:

  • Multiple orders should be organized in an order set or order panel.
  • Communications should happen without slowing down the clinicians.
  • Individuals receiving the communication should know what to do.

Significantly impacting outcomes requires a straight line from prediction to action.

Given the challenges of adopting predictive analytics, some people question whether implementation of them is worth the effort. The clear answer is yes. The integration of predictive analytics into your value-based initiatives can drive outstanding returns for your patients and your revenue that otherwise could not be identified.

For example, we integrated predictive analytics for pressure ulcers at the University of Mississippi Medical Center by partnering with a cloud-based deep-machine learning vendor who produced pressure ulcer predictions significantly superior to the Braden score (a standard clinical score for skin integrity). To provide the line of site to action, we incorporated the predictions into the workflow for our wound-care nurses. We have seen a 66 percent decrease in stage III and stage IV pressure ulcers since we began sharing the prediction. The 66 percent reduction resulted in over $300,000 in savings in just six months and improved our value-based purchasing score for pressure ulcers. We plan to reduce the incidence of pressure ulcers even further by integrating the predictions into the workflow for all nurses and to develop alerts to make taking action easier.

The benefits of predictive analytics for value-based reimbursement greatly outweigh the challenges. Organizations need to begin the process of adopting this new technology for the benefit it brings their patients and their organizations. It is important that we make the improvements we can today while we wait for the next great idea. Predictive analytics is available now with impactful results being proven out at health systems across the nation.

About the Author

Dr. John Showalter is an influential thought leader on the innovative use of health information technology, devices, and data to drive improvements in healthcare delivery. Dr. Showalter serves as the Chief Health Information Officer at the University of Mississippi Medical Center. He received his B.S. in biomedical engineering from Columbia University and his medical degree and a master of information systems in health care delivery and management from Penn State University. In addition to his board certification in internal medicine, Dr. Showalter became one of our nation’s first physicians to become board certified in clinical informatics in 2013.

Contact the Author

This email address is being protected from spambots. You need JavaScript enabled to view it.