The fact that many clinical performance measures are now being designed with EHRs as their data source may lead to systemic improvements in patient safety. Healthcare Informatics looks at four healthcare providers whose IT infrastructures already underpin their performance improvement strategies.
Most healthcare organizations focusing on performance improvement have traditionally faced some trouble getting buy-in from clinicians. In many cases this was because reports were based on 6-month-old data from chart abstractions, which were partial, random, and sometimes inaccurate.
As more healthcare data becomes electronic, clinical informatics teams have begun extracting data from transactional electronic health records (EHRs) into data warehouses, where it can be normalized and reported on much more rapidly. “In a lot of settings, quality improvement has been embedded for 20 years, but by pulling information from EHRs, you can get a much broader population view and more timely data,” says Carole B. Black, M.D., who oversees clinical development for Chicago-based Valence Health, a solutions and consulting firm that helps healthcare organizations measure quality care and outcomes.
EHRs tend not to do a good job of this reporting, says Elizabeth Simpkin, Valence Health's vice president of consulting services. “EHR vendors are working on improving in that area, so I would argue that the time you might need help with a data warehouse is now. Even if hospitals have ways to get data from their own systems, they now need to extract it from multiple sources.”
Where organizations are gaining the most benefit is using EHRs for population-level reporting, Simkin adds, by extracting data from a variety of EHRs in use and then synthesizing it to provide a comprehensive view of a patient across service providers. “The larger integrated delivery networks and multispecialty groups have spent more time thinking about the population level,” Black says, “but so far other providers have spent more thinking about the patient right in front of them. It is a different mindset.”
Healthcare Informatics recently spoke with leaders of four healthcare system initiatives about how their IT infrastructure and strategy underpin their performance improvement strategies and about some of the improvement gains they are already seeing.
EHRs DESIGNED FOR MEASUREMENT
Typically, quality measure reporting tools have been bolted on to existing EHR systems, most of which haven't had strong reporting tools baked in. But one organization, the Alliance of Chicago, had a vision early on that among key uses of its system by more than 25 safety net health centers were quality, safety, and research, says Erin Kaleba, director of research initiatives. “The Alliance selected a system before rollout that it could customize with structured fields to make it easy to pull data out of the back end,” she adds.
WE DECIDED TO LET THE CLINICIANS THEMSELVES ASK QUESTIONS ABOUT PATIENT SAFETY MORE TARGETED TO THEIR SPECIALTY, AND THEY ARE LEARNING TO ASK INNOVATIVE QUESTIONS.-JEFFREY FERRANTI, M.D.
The Alliance's centrally hosted GE Centricity system was chosen because of its flexibility to allow for the creation of evidence-based dashboards, adds Tim Long, M.D., chief clinical officer. “We are getting accurate, population-level data in front of providers for the first time, and the power of that is amazing,” Long says. “We are able to show them 100 percent of their population. They can see it is tied into national measures and American Diabetes Association guidelines. That is powerful. A lot of people think they are adhering to guidelines when they are not.”
The Alliance provides its member health centers monthly dashboard reports of performance on diabetes, coronary artery disease, preventive care and screening, HIV/AIDS, and hypertension.
Kaleba says the Alliance is making some modifications to meet the 44 meaningful use clinical measures, but by and large those are things it is already doing. “Where we will have to do more is in patient engagement with a portal and having visit summaries ready for patients,” she adds.
Another next step is getting clinicians more comfortable with using clinical decision support tools at the point of care and sharing data with patients, Long says. “Physicians are missing the boat if they see this whole new tool only as a way to do what they used to do on paper but do it electronically.”
BOTTOM-UP REPORTING AT DUKE
Enterprise data warehouse efforts at the three-hospital Duke University Health System in North Carolina have evolved to take a bottom-up approach to performance improvement. Duke clinical informatics leaders have layered business intelligence tools on top of the data warehouse and then opened them up to its own clinicians on the Intranet.
“We could have analysts in a centralized environment write reports. There is value in that,” says Jeffrey Ferranti, M.D., M.S., acting chief medical informatics officer and vice president of clinical informatics. “But we decided to let the clinicians themselves ask questions about patient safety more targeted to their specialty, and they are learning to ask innovative questions.”
Duke University Health System teams are developing tools such as the Adverse Drug Event Surveillance system. Every day a rules engine scans the clinical information systems of inpatients at Duke hospitals to look for patient demographics, lab results, and medication orders that may be indicative of potential harm. Another project involves computerized physician order entry (CPOE) metrics. A partnership between CPOE, health analytics, and IT patient safety team members provides users metrics related to clinical order sets and decision support. A CPOE survey provides feedback on clinical decision support, workloads, and safety concerns.
A tool called DEDUCE (Duke Enterprise Data Unified Content Explorer) supports quality monitoring and improvement efforts. “We could ask it to show all the diabetes patients in Durham County and what medications they are on,” Ferranti explains. It can identify outliers such as diabetic patients who haven't had their A1c hemoglobin levels tested recently.
“These tools also can help us see differences in outcomes in our three different hospitals, such as rates of C. difficile colitis,” he adds. “The difference between 0.5 percent and 1 percent is very hard for doctors to discern, but with the data warehouse we can identify problems and intervene.”
TRIHEALTH: PULLING DATA FROM MULTIPLE EHRs
Setting up a performance measurement scheme for ambulatory physicians using a multitude of EHRs is a challenge, but TriHealth, an integrated healthcare system formed as a partnership between Good Samaritan Hospital and Bethesda Hospital Inc. in Cincinnati, was undaunted. Beginning in late 2008, TriHealth, which has 250 employed physicians, created a clinical data warehouse that enables practices to generate on-demand reports about treatment recommendations, patient vital statistics and treatment compliance.
Led by Georges Feghali, M.D., chief medical officer for TriHealth, a quality committee came up with 12 priority protocols. “Then a clinical data warehouse committee made up of IT and physician leaders defined the components to extract, how we are going to build it, and if there are any missing data elements, how we can get that data,” explains John Ward, director of health system IT integration.
“One of the biggest challenges is working with the different vendors' applications in terms of trying to find where to extract data elements so we can normalize it,” says Ward. “So in terms of recording smoking status, five applications might put it in five different places in the chart, such as in social history or in physician notes. We had to go in and find all the places that might be, where to get it and in some cases change the workflows of physicians to standardize that. That was one of the most difficult aspects.”
TriHealth has a strong focus on the medical home model and has added nurse coordinators and diabetes educators for patient follow-up. Ward says it has already seen some practices make vast improvements since it started offering them more immediate and complete performance data.
For instance, one practice started in 2009 with only 16 percent of its diabetic patients meeting the definition of having their condition under control. That number changed to 32 percent in 2010. “That is a huge improvement,” Ward says. “I think the physicians have previously been starved for data, and are now thinking in a different way,” he adds. “They now think in terms of their populations with chronic conditions. That's how you get the dial to move.”
TIMELY DATA FOR CHRISTIANA CARE'S CLINICIANS
Digging deeper into clinical results in its own data warehouse has led to several performance improvements at Christiana Care Health System in Wilmington, Del.
Two years ago, the two-hospital system sought to better understand several performance issues, including extended hospital stays. Using a tool called CareDiscovery from ThomsonReuters, Christiana Care analysts and physician teams studied clinical performance and utilization management in specific diagnosis-related groups such as heart disease.
Donna Mahoney, director of data acquisition and measurement, leads a team of eight analysts working in the Center for Quality and Safety. She says her job is made much easier by the fact that Christiana Care has had a data warehouse in place for 15 years. “We can go to the data warehouse team with any issue we have. They are the ones preventing us from having to do this through chart abstraction, which would be quite difficult.”
Christiana Care used a value-score methodology, which combines quality metrics (mortality, morbidity, readmission, patient satisfaction, and evidence-based guideline compliance) with expected-cost and length-of-stay metrics to give an overall value grade for high-volume populations. Value improvement teams review monthly dashboards to identify opportunities to improve the value of care for patients. “An example would be better compliance with evidence-based practices, such as patients getting the antibiotics they are supposed to,” Mahoney says.
From sharing actionable data with clinicians, she says Christiana Care has seen improvement across the board in the areas it worked on. For instance, readmission rates for heart failure patients have dropped 35 percent over a two-year period. “Many of the metrics in our data warehouse are almost real-time, and it is clear that the timelier the data, the more engaged physicians are,” Mahoney adds. “They don't want data that is six months old. They are not interested.”
STILL IN THE MINORITY
Despite the inspiring work being done to mine clinical information systems for performance and patient safety efforts, the organizations that have made that kind of progress are still in the minority, says William Fera, M.D., executive director of Ernst and Young's Advisory Services Group practice. “We are reaching a point where more health systems are working on these reporting efforts, but it is a question of how you design it and how robust it is. It can't just be about aggregating data,” he says.
One danger Fera sees is organizations focusing too narrowly on the meaningful use requirements. “We tell clients to put their focus on the longer term,” he says. “If you look at Stages 2 and 3 of meaningful use, there are many more reporting requirements. Just cobbling together a few reports isn't going to cut it. And if you are looking at accountable care organizations, that will require even more sophistication, including marrying cost information with clinical outcomes and that requires robust data warehousing.”
Daniel J. Marino is president and CEO of Health Directions, LLC, Oakbrook Terrace, Ill., a national consulting group that provides business solutions for healthcare organizations. He can be reached at (312) 396-5414 or firstname.lastname@example.org. Healthcare Informatics 2011 October;28(10):22-44