"The strength is that it pulls data sources from different parts of the record, thus giving clinicians a more complete picture of the patient's condition," says Tranquillo.
"It doesn't matter if you are talking retrospective or real-time," Haughom says. "It takes a lot of work and effort to build the infrastructure, to get care providers used to using it, and to figure out how to do it effectively so clinicians will use it to improve decisions."
Decision support 3.0
Eventually decision-support tools will include genomics data that will enable care providers to tailor care for individual patients. For example, a patient with a certain genetic profile may respond to a drug differently than other patients and require a lower dose. Before healthcare providers can realize the potential of decision support, however, there are myriad challenges that need to be addressed—and technology is just one element.
There is limited availability of structured data. "How do you put a code on a patient that comes in and says, 'I don't feel well'?" says Vaughn. "The main limiting factor right now is getting enough data on the patient that allows us to create decision-support rules."
The lack of interoperable systems also poses a problem. The decision-support tools available in SSM's version of Epic, for instance, are not easily translatable to Cerner, says Vaughn, who is a proponent for decision-support "SOA [service-oriented architecture] software." Healthcare organizations should be able to share decision-support rules, so that resources aren't being wasted by reinventing the wheel in different vendor systems, he says.
How the data should be presented hasn't been figured out yet either. Physicians are already being bombarded with data and suffer from alert fatigue, says McNichol. But fatigue is not just about the number of alerts, says Tranquillo. Physicians are also being asked to navigate through various systems. That is why TJUH integrates as much information as possible either through a single view or by using a consistent presentation, Tranquillo explains.
Haughom is not sure what the "ideal" presentation of the data should be so that clinicians can easily and effectively use it, but he can tell you what it won't be—tables of data.
"We already see in our analytical strategy that tables don't work," he says. "Trending tools and dashboards [are better], because people can look at a tool with graphs and trends more quickly."