As an anesthesiologist helping Boston's 900-bed Massachusetts General Hospital create the ‘Operating Room of the Future’ in 2002, Julian Goldman, M.D., did a lot of thinking about how to integrate the information coming from all the medical devices in the operating room. “I realized the absence of plug-and-play interoperability was really a barrier,” he says.
Despite the need — from both clinical and IT efficiency standpoints — to merge medical device data and hospital enterprise networks, that barrier still exists today. Not so much in radiology and cardiology, where standards have emerged, but in point-of-care systems, which were designed to run as stand-alone networks.
The fundamental problem with devices such as vital sign monitors and infusion pumps is that there is no common format for exchanging data. Because each device vendor has its own proprietary protocol, there is no obvious cost-effective route to connectivity.
“This is just plumbing, and hospitals are not too excited about spending $2 to $5 million on plumbing,” says Tim Gee, principal at Medical Connectivity Consulting (Beaverton, Ore.). They want to spend that money on patient safety efforts and other meaningful clinical projects, he says. “So the lack of interoperability just adds cost and complexity to the situation,” he adds, “and means that progress is slower than it ought to be.”
Yet hospitals with no connectivity between devices and electronic health records have clinicians jotting down vital sign information on scraps of paper and manually entering data into a patient's chart. They get delayed information and face a greater risk of errors, typos or transposed numbers. “These are all weaknesses of manual entry and it's what you're trying to overcome by paperless charting,” Gee says.
Goldman, who now heads up the Medical Device Plug-and-Play Interoperability Program at Mass General, which is working to develop industry-wide standards, says individual hospital data integration efforts “are like tape and bailing wire.” They each have to recreate the wheel and there's a chance it won't be done right.
Goldman also stresses that more important than efficient data transfer is the added functionality related to patient safety. “For instance, an infusion pump shouldn't be able to give a patient a drug that's on their allergy list,” he says. “Well, a clinical information system could stop a smart pump from giving that medication. If systems are smartly integrated, they can flag suspicious data.”
Some CIOs plunge ahead
Most EMR vendors have not addressed medical device connectivity from their end. One exception is Cerner Corp. (Kansas City, Mo.), whose CareAware architecture offers an enterprise approach to connectivity. Gee notes that if Cerner does well in this niche, other large EMR vendors may follow suit. In the meantime, despite the lack of standards and the considerable expense, some CIOs have made progress by turning to third-party integrators.
For such a small hospital, the 108-bed St. John's Medical Center, located in the resort town of Jackson Hole, Wyo., is quite technologically sophisticated, and has been working for several years to integrate its medical devices with its clinical information system, San Francisco-based McKesson's Paragon.
Traditionally, the devices have been on separate, single-purpose, physically protected networks with their own protocols. As it became clear that they might want to send data to other applications, some medical device companies wanted hospitals to purchase middleware solutions to link their proprietary network to the hospital's clinical information system. But that would only solve a portion of the problem. For instance, a small hospital like St. John's has eight different device vendors, explains David Witton, director of information systems. Purchasing a middleware solution from each of those vendors would be cost-prohibitive.
“We realized we needed some glue layer in the middle,” he says. St. John's got involved in a development project with McKesson and Andover, Mass.-based Capsule Tech Inc.'s DataCaptor software, which provides an interface engine to hundreds of devices and helps automate getting data from both mobile and bedside vital signs monitors into the McKesson Paragon system. (Before information is fed into the system, a nurse looks over the data for approval.) The automation has proven highly popular with the nursing staff, according to Witton.
“This application is the only time I have seen a nurse hugging an information systems employee,” he says.
For Hays Medical Center CIO Bill Overbey, the initial drive to integrate device data came from the nursing department. In the past, nurses in the intensive care unit would read vital signs every 15 minutes and write them on paper charts. When the 194-bed Kansas hospital adopted a Meditech (Westwood, Mass.) system in 2004, the vital sign entering became much more labor-intensive because it involved clicking through screens. The clamor for greater efficiency came from nurses who felt data entry was taking too much of their time.