Information technology, data, advanced statistical methods, and algorithms. All are important components of predictive analytics and machine learning, and all play supporting roles for improving healthcare outcomes. Indeed, these features enable faster and more accurate clinical decision-making while improving performance in areas such as reducing readmissions.
But Dr. David Wild, the Vice President of Lean Promotion at The University of Kansas Health System, would say people and process also play starring roles in reducing readmissions. At the 2017 Healthcare Analytics Summit, Dr. Wild will present a case study from his organization, “Using Predictive Analytics and Machine Learning to Lower Systemwide Readmissions.” This title may lead one to believe the future of healthcare is about the rise of technology. But equally responsible for the successes in this case study are process improvement, engaging clinicians, and educating leaders.
The University of Kansas Health System is a destination academic medical center with 773 beds and more than 5,000 employees. Leaders at the system believe to reduce readmissions and lower costs it “must provide patients efficient, value-added, effective, and patient-centered care during and after discharge from the hospital.” Some organizations working to improve readmissions target a single domain or clinical area. This organization, with over 30,000 admissions a year, has taken on all readmissions—and produced stellar results.
Dr. Wild, along with Chris Harper, the system’s Director of Business Architecture and Analytics, will demonstrate what was at stake before this journey began, and how the health system was able to achieve such remarkable results within a remarkable timeframe:
- Its target rate for all readmissions was 12 percent; typical rates hover around 25 percent.
- All-cause readmission rates dropped from a high of over 30 percent to a low of 8.5 percent in a nine-month period during the program.
- The heart failure readmission rate dropped from a high of almost 15 percent to a low of 3.8 percent in one year.
The wakeup call for this initiative came when The University of Kansas Health System realized that readmissions cost $17.4 million in 2014, or 17 percent of total hospital payments. It needed to improve its care management process, deploy predictive analytics to identify high-risk patients, and develop a new education process for leaders, all while engaging a team of clinicians from the entire continuum of care.
And with this particular system, the continuum was expansive, comprising centers for wellness, pharmacy, urgent care, primary care, specialty care, diagnostics, ambulatory surgery, acute care, rehabilitation, skilled nursing, assisted living, home infusion, home health, psychosocial services, and hospice.
The Continuum Advisory Team played a huge role in the improvement process, identifying gaps in the continuum, then drafting and operationalizing a strategic plan. But there were two other equally vital components. First, The University of Kansas Health System engaged and educated the right people in Lean techniques, from fundamentals to advanced concepts. Second, it promoted predictive analytics under an effective technology platform, to uncover insights about why some patients were being readmitted.
Most important in all of this were the connections made between predictive analytics and machine learning to surface the key clinical and operational improvements that kept people out of the hospital and reduced avoidable readmissions. Taking the broad view, this is the real solution to the healthcare crises: putting the tools and processes in place on the frontline and showing doctors and nurses the data they need to make decisions that improve outcomes.
Would you like to learn more about this topic? Here are some articles we suggest:
- Best Way to Run a Hospital Readmissions Reduction Program
- Effective Healthcare Data Governance: How One Hospital System is Managing its Data Assets to Improve Outcomes
- DKA Risk Prediction Tool Helps Reduce Hospitalizations
- Evidence-Based Care Process Model Reduces SSIs and Readmissions
- Reducing Heart Failure Readmission Rates