Exchanging health information in an accountable care organization can include dozens of different electronic health record systems with data from thousands of physicians.
It is one thing to have the systems talk to each other. It is another to make sense of what they are saying.
Michael Sills is a cardiologist and ACO vice president of informatics technology for Baylor Quality Alliance, said the ultimate IT goal is to manage patient populations and create one patient record from multiple sites of care.
BQA purchased Covisent, AT&T’s health information exchange tool, to link the independent EHRs. The ACO also bought Humedica, a population-health analytics software system that will link inpatient EHR, physician records for Baylor’s North Texas Physician Network, the HIE, claims data from the system’s contracted commercial insurers and the system’s financial records. It will be the first time Baylor will have cost and quality information in the same data tool.
Humedica represents the “big data” piece of the project. Big data draws on large repositories of information from multiple sources to create evidence-based protocols.
Baylor wants to be able to predict who is at risk of hospital readmission in the next three months and proactively manage their care.
According to McKinsey & Co., big data in healthcare could translate into $300-$450 billion in savings annually. The consultant cited Kaiser Permanente, which used big data to identify ways to improve outcomes in cardiovascular disease. The discoveries resulted in fewer lab tests and office visits, saving $1 billion annually.
A Lavastorm Analytics survey of more than 600 healthcare technology professionals found that more than half planned to increase analytic investments in 2013 because their existing tools were inadequate to handle complex analysis and predictive modeling.
In a recent blog post, Obama administration chief technology officer Todd Park said the U.S. was undergoing a “data-powered revolution in healthcare” that is improving the industry “from the ground up … These (big data) tools are … enabling clinicians to analyze their patient population, understand who needs help … and proactively reach out and give those patients the care they need.”
Other Dallas-Fort Worth healthcare organizations have been using big data for predictive analytics as well.
Texas Health Resources (THR), three other U.S. health systems and IBM announced Thursday they have joined Premier healthcare alliance in a data analytics venture aimed at lowering healthcare costs and improving quality. The consortium initially will target preventable hospital readmissions and lack of medication compliance.
Melissa Gerdes, MD, assistant vice president and chief medical officer of outpatient services and accountable care organization strategy at Methodist Health System, said, “We are really just beginning to unlock the potential of analytics to improve patient care and outcomes.”
One such tool is helping Methodist’s Patient Centered ACO manage population health. The tool analyzes claims data on the 14,000 Medicare beneficiaries attributed to Methodist’s ACO, as well as 6,000 Methodist employees to predict who in the population is most likely to utilize high cost health care.
Initially, Methodist will use the predicted list of people for its ACO beneficiary nurse navigators to help coordinate care, provide education, service referrals, and identify and overcome barriers to achieving better health for these patients. A workflow manager will house the documentation for the beneficiary nurse navigators and track interventions.
Eventually, Methodist plans to add clinical data from labs, radiology, and electronic medical records to enhance the predictive accuracy. The tool, recently launched, is expected to enhance communication among all patient care team members with an overall goal of improving patient experience, quality, and cost/utilization.
To analyze population health and trends further, Methodist participates in the Dallas Fort Worth Hospital Council Foundation MYIQ Quikview database. The database tracks patient readmissions to all area hospitals to help Methodist manage readmissions.
“Historically, physicians have not had access to credible data that resonates with their clinical goals.” says Sam Bagchi, MD, Methodist Health System vice president and chief medical informatics officer. “The need to reliably measure physician performance in advance of CMS public reporting of physician profiles has elevated the importance of data analytics at Methodist Health System.”
Parkland and THR have been using a software application to identify high-risk patients. An algorithm digitally analyzes the patient’s clinical and social factors in patients’ records and uses advanced predictive modeling to predict likely outcomes without extra intervention. Parkland has cut 30-day readmissions for Medicare patients with heart failure, including readmissions to all hospitals, by 31 percent, with an estimated savings to the hospital of $500,000 with no increase in staffing.
Baylor’s Sills said one of healthcare’s biggest problems has been data aggregation to create actionable predictive modeling. He said big data takes “chaotic and disparate” information and puts it in an easy-to-use format.
“We have to identify the sickest people in a proactive way. If you were in the neonatal intensive care unit or had a car accident last year, you won’t be expensive this year. But about half of last year’s costly patients will be again this year. We need to change their cost trajectory going forward. For people who are readmitted to the hospital, most never see a primary-care doctor or have one. We can flag that,” Sills said.
Cliff Fullerton, family physician and BQA chief medical officer, calls risk stratification the key to healthcare’s future. He said Humedica creates information based on category, such as age, overdue appointment or lack of A1C control.
Sills said most large U.S. healthcare systems either are considering using big data or are at the same stage as Baylor. He said it requires a robust IT infrastructure and “enormous” costs, making it unworkable for smaller organizations. Despite its complexity, Baylor did not hire additional employees dedicated specifically to the project.
BQA has performed analytics on the health system’s employees and learned some interesting lessons:
- Generic prescribing can improve. Sills said every 1 percent increase in generic prescribing could save the system $600,000 a year. “That’s for a population of 30,000. What if we could do that for millions,” he said.
- Five percent of Baylor employees account for 60 percent of the system’s health costs, and about half of those are unmanageable cases such as automobile accidents or cancer. However, the remainder is manageable chronic cases. “That’s the magic,” Sills said.
- The system is managing costs for lower back pain more successfully than it expected, based on protocols it has implemented to minimize surgery and imaging costs.
- Hospital readmissions have not budged much despite efforts to reduce them. Fullerton cited a recent New England Journal of Medicine article that identified “post-hospital syndrome.” In addition to attempting to recover from an acute illness or surgery, patients have to overcome the fact that they are deconditioned from extended bed rest, placing them at greater risk of accidents and falls. Decreased stamina can hamper patient resolve to engage in their recovery. Fullerton said Baylor is using Pulse360 to drill deeper into the causes of readmission potentially based on care given in the hospital.
The goal is to share data with the 13 Texas hospital systems that comprise the Healthcare Coalition of Texas, which includes Baylor, Scott & White in Temple and Memorial Hermann Health System in Houston. BQA officials hope to have initial data by the end of 2013.
Steve Jacob is editor of D Healthcare Daily and author of the book Health Care in 2020: Where Uncertain Reform, Bad Habits, Too Few Doctors and Skyrocketing Costs Are Taking Us. He can be reached at [email protected]