A new survey from the American Health Information Management Association finds that 95 percent of the more than a thousand healthcare industry professionals queried believe that "high-value information" is essential for improving patient safety and care quality.
"I'm reassured," says Deborah Green, AHIMA's executive VP of operations and COO, and a contributor to the study. "That's what it is about."
This bit of news comes as analytics is slowly making its way from business intelligence – revenue-cycle management and financial performance – to the clinical side of healthcare. "Predictive analytics can help with readmissions prevention," notes Charles Christian, CIO at St. Francis Hospital in Columbus, Ga.
His and other community hospitals are just starting to make small strides in this direction, but larger organizations are seeing results. As the Wall Street Journal reported this month, the University of Pittsburgh Medical Center has put $105 million toward a massive data analytics program, which is employed to analyze the success of a patient-centered medical home pilot. UPMC found that those with medical homes had substantially better health outcomes after six months in the program, and the medical home reduced health expenditures by $15 million in the first year.
[See also: Clinical data analytics next big thing.]
In a pilot at the Ohio State University's Wexner Medical Center, analytics have deployed cardiologists and oncologists to an ambulatory clinic in a part of Columbus, Ohio, with high rates of heart disease and cancer. With data algorithms, the university has been able to identify patients in need of intervention and personalize care for those individuals in order to reduce initial hospitalizations as well as readmissions, according to Burroughs Healthcare Consulting Network.
Jim Adams, executive director for research and insights at The Advisory Board Company, a Washington, D.C.-based consultancy and research firm, notes that analytics can help health systems add psychosocial factors to their decision-making. For example, if a hospital knows that a patient lives alone, the discharge plan can include arranging transportation for follow-up care.
Adams says that it's possible to do "simple analytics" just by getting people together in teams and brainstorming, which may be the best some can do at the moment, though he has seen many organizations applying business intelligence to stratify clinical risk and prioritize patient interventions. "Ideally they're doing it at admission time, not just at discharge," he says.
[See also: Mayo Clinic launches bedside analytics.]
As the AHIMA survey suggests, healthcare organizations may get the importance of having good data, but many apparently still have a long way to go in building a foundation for high-functioning analytics programs.
The Advisory Board has developed a four-phase maturity model for business intelligence, starting at "fragmented," then progressing to analytics from an enterprise perspective, "advanced" analytics and, ultimately, big data.
Following this model, organizations cannot say they have achieved big data until they have developed "an engrained understanding of BI capabilities and limitations." They must be stewards of both internal and external data, employ "sophisticated delivery models" and be able to apply such things as natural-language processing and genomics, according to Advisory.
Based on a survey of Advisory Board members, Adams sees three major challenges to business intelligence: data governance; culture; and organizational structure and resources. "It's really about usage and usability of the data," Adams says. Factors include quality of data, literacy of people using the data and, of course, privacy and security.
"Just because you can go out and buy a predictive modeling tool doesn't mean you can do predictive modeling," according to Adams. "It's kind of at the phase of hardwiring some of this now," both technically and culturally, he says.
AHIMA is urging its members to place particular emphasis on the importance of data governance. Information governance is particularly important when pulling from multiple data sources in order to ensure data integrity – the classic "garbage in, garbage out" philosophy.
While 65 percent of respondents to the AHIMA survey said their organizations understand the need to formalize information governance practices, just 43 percent of this segment has done so, and 13 percent has reaped no benefits so far.
Green says health information managers tend to pay close attention to coding and demographic data, but that is not good enough for analytics based on clinical records. "I don't think we have taken a holistic view," she says. Nor do many organizations fully understand that they need to maintain the integrity of data throughout the entire lifecycle in which that information is used.
Information governance is not all that different than any other form of governance in terms of principles to follow, according to Green. Health systems should set rules for data accountability, transparency, integrity, security, regulatory and standards compliance, availability, retention and disposition, she says.
"It's critical to have an accurate picture of your business performance," Green says.
This might be a daunting task, but it is critical to success of any BI program. As Adams points out, the more people that get involved with analytics, the more chance for error. Some don't know that they're dealing with garbage so they make decisions based on bad data, or they will figure out that it's bad only after wasting a lot of time. "Then you lose a lot of credibility within your organization," Adams says.
"Healthcare data is so complex that you can't govern every piece of data all the time," he says, but it is possible to simplify the task. Adams notes that in developing its own information governance strategy, Mayo Clinic was able to whittle its list of prioritized data points from an unwieldy 400 down to a more manageable 40 or so.
"One of the rookie mistakes," according to Adams "is trying to take on the world."
