How Analytics Supports Medical Homes (Health Data Management)

At Stanford Health Care’s patient-centered medical home program, enrolling a patient in the home is a multi-hour process to gather data supporting complex care management by understanding the patient’s health history and present status, pharmacy history, socio-economic status and home life,

[Written by Joseph Goedert. View original HealthData Management article here]

The need for medical homes in the delivery system supported with dashboard-based data analytics was apparent, as the five percent most expensive patients account for 50 percent of medical costs, says Jorge Wilson, director of clinical and business analytics at Stanford Health Care. “We spend too much time right now addressing people in critical care or when they present themselves in the emergency department,” he noted during a presentation at Health Data Management’s Healthcare Analytics Symposium in Chicago.

Stanford’s own employees and dependents are among medical home patients. Stanford moved the top 10 percent of the university’s and hospital’s health spenders into capitated care, supported with such services as 24×7 access to a provider, same day appointments, medication reconciliation, a care management team and patient education. “Care coordinators and medical assistants have their own patient panels and work to the limits of their medical credential,” Wilson explained.

Data to be assessed in Stanford’s analytics program must meet one of four criteria: enable clinicians to understand population risks, identify workflow issues that result in care gaps, enable patients to understand their risks, and measure outcomes from quality improvement. What the analytics team doesn’t do is generate predefined reports, Wilson said. It focuses on no more than six large and prioritized initiatives a year, using only the most useful data to make sure they are done well.

Sources of the data come from financial and administrative systems, electronic health records, patient satisfaction scores and claims. To use the data effectively meant standardizing certain terms and processes across the enterprise, such as defining who a doctor’s patients are and what a diabetic is. That took two months. Stanford also had to decide on an enterprise level low, medium and high risk measures for A1C, LDL cholesterol, blood pressure and smoking status, among other factors, a process that took four months. “These questions are among the first issues you have to settle when developing a population health management plan,” Wilson advised.

The initiative further included gathering self-reported data, such as blood pressure and weight checks. Patient-reported blood pressures are scanned into the EHR. “This is basic stuff, but you can’t do advanced analytics without the basic stuff,” he added. To monitor weight when congestive heart failure patients are discharged, Stanford sends them home with a Wi-Fi scale that sends the daily data.

Getting all this data from all the different sources opens new opportunities for improving patient care and outcomes, but some problems as well, Wilson said. “We’re still trying to handle the signal-to-noise issue to keep doctors from being overloaded with data.”