Bridging a gap

The U.S. medical community and the world watched in horror last month as the condition of Aimee Copeland spiraled downward.

Copeland's case became a media sensation as the 24-year-old, blue-eyed blonde graduate student from Georgia lost a leg, a foot, both hands—and nearly her life—to necrotizing fasciitis, commonly known as “flesh-eating bacteria.”

Proper diagnosis and prompt action are crucial with this rare condition, but three days lapsed between her May 1 initial encounter with emergency room physicians to treat a gash on her leg and the first of her surgeries May 4 for the subsequent infection, according to a recounting of events by her father, Andy Copeland.

Computer-assisted diagnostic support systems—a specialized subset of computerized clinical decision support systems—could have prompted Aimee Copeland's physicians early on to consider necrotizing fasciitis as a likely diagnosis, according to developers of the technology.

Given the intensity of his daughter's symptoms, a computerized diagnostic tool should not have been needed, Copeland's father said. But hearing the developers' claims, he said, “I wish the systems were around when Aimee showed up. It's something every hospital should have.”

Tanner Health System, where Copeland was initially treated, declined to comment on her father's account of events involving Aimee's care, citing patient confidentially, said spokeswoman Kelly Meigs.

Tanner, a three-hospital system in Carrollton, Ga., said in a written statement that its providers “have access to the latest, most current research and information available,” including “constantly updated informatics resources to aid in diagnosis and treatment.”

For a variety of reasons, diagnosis support systems are rarely used today, despite a lineage that goes back decades.

“This is how the field of informatics started, trying to build these systems,” said Dean Sittig, a professor at the School of Biomedical Informatics at the University of Texas, Health Sciences Center, Houston, and an expert in clinical decision support systems.

Yet only a few diagnostic support systems are on the market and installations are “minuscule,” said Jason Maude, cofounder of Isabel Healthcare, who developed one such system after his daughter nearly died from an infection similar to Copeland's in 1999.

Among the provider organizations that do make the systems available, “getting the staff to use them is the biggest challenge they face,” said Coray Tate, vice president of clinical research for KLAS Enterprises, a health IT product research firm in Orem, Utah. But once they do, “those that are using them think they are awesome,” Tate said.

The federal government, meanwhile, has paid hospitals, physicians and other providers more than $5 billion in incentives to adopt “certified” electronic health-record systems. And while the feds—including Congress—required EHRs to have e-prescribing functionality as a condition of certification, they didn't mandate diagnostic support tools in the Stage 1 meaningful-use criteria, won't require them in Stage 2 and probably won't even in Stage 3, which could take the program to 2018.

“It's a little too far out there,” said Dr. Paul Tang, vice president, chief information and technology officer at the Palo Alto (Calif.) Medical Foundation. Tang is vice chairman of the Health Information Technology Policy Committee, which advises the feds on the EHR incentive program. Diagnostic support technology has not been pilot tested sufficiently to ensure its efficacy, he said.

Broad acceptance of diagnosis support is at least five to eight years off, said Dr. Lee Jacobs of Atlanta, who worked on the rollout of the Kaiser Permanente EHR and continues as an IT consultant there.

Jacobs described the Kaiser EHR, based on software by Epic Systems, as “a powerful tool,” but it's such a “gigantic monster” of a task to collect and present the data at hand, “to go to the next level and do more thinking through the EMR, that's too big,” at least for now.

Healthcare visionaries “can make a business case for an accurate diagnosis, and I think the Kaisers will go down that road,” he said. “But even the good groups are still in the pioneering stage of bringing these things together.”

Developers and true believers in the technology, however, see shame in not having widespread adoption already.

Count among them Dr. Lawrence Weed, 88, whose problem-knowledge couplers, a diagnosis and care management tool introduced in 1984, are an extension of his

earlier work on problem-oriented medical records. Weed founded PKC, which markets couplers as diagnosis support tools. He left the firm in 2006.

“People have been saying to me since the 1960s, you're ahead of your time,” Weed said. “I say, my God, you want me to live to 160? How long are you going to take?”

According to Weed, physicians' minds no longer are capable of memorizing and recalling all the medical knowledge in a fast expanding universe of new clinical information and matching that to patients' needs. Physicians and medical education establishments are unwilling to acknowledge this reality, he said. “It's easy to see why you're having so much trouble getting over it. You've invested half a million dollars in your education, you're full of prestige,” Weed said.

Medical schools, Weed said, need to recruit students not based on their ability to memorize and regurgitate facts, which will reside in the computer systems, but on hands-on and interpersonal skills, and train them to be competent in discrete skills and procedures. But that, too, is threatening to the status quo, he said.

Thirteen years ago, Maude's daughter Isabel, then age 3, nearly died after a delayed diagnosis of necrotizing fasciitis, further exacerbated by toxic shock syndrome, both complications of chicken pox.

“Healthcare doesn't seem to regard diagnosis as important,” Maude said. “The fact that we have these cases that are absolutely tragic, the case of Aimee Copeland, they're happening all the time, even though the solutions are readily available and cheap and effective.”

Maude said he's confident his system, named for his daughter—had it been available and used by Copeland's physicians—would have red-flagged them about necrotizing fasciitis. “All they had to put in was leg cut and intense pain,” he said. “Necrotizing fasciitis comes up as the No. 1 suggestion.”

About 70 physicians at 118-bed Children's Hospital & Medical Center, Omaha, Neb., have been using the Web-based Isabel system for about three months. Though not fully integrated, a link to the tool is embedded in the hospitals' ambulatory EHR, where “it's literally a right click off the problem list,” said Dr. George Reynolds, the hospital's chief medical information officer. Isabel is also available to clinicians with iPhones, iPads and Android mobile devices, he said.

“Can I verify Jason Maude's statement, had the docs had a tool like this available, they would have used it and made things better, I can't say,” Reynolds said. “You can't guarantee it's going to change behavior every time, but I have evidence that we do change behavior with our CDS tools.”

Front office workers in a group practice in Bangor, Maine, use a “triage coupler” to direct patients who call in or arrive with a new problem, said Dr. Charles Burger, a family practitioner there.

The triage system “helps our front staff take patient complaints and sort them as to what needs to be done, how soon they need to be seen, or should they go to the emergency room or do they need lab work to be done before they come in,” he said.

Burger was one of Weed's students in medical school and has worked with him since the 1980s, helping Weed pilot test the problem-knowledge couplers Burger still uses. His practice is owned by Martin's Point Health Care system and is one of four practice sites with 30 physicians trained to use the tools.

Patients selected to fill out a full problem-knowledge coupler face a daunting list of questions, a task they either complete at home via the practice's Web portal, at a kiosk in the office, or, if help is needed, with the aid of an office assistant. Loading the patients' information into the coupler is done before they see the doctor, Burger said.

To adopt couplers, “you have to be a true believer and really understand the principles and philosophy behind it,” Burger said. That's becoming an easier sell, given the increasing complexity of medical practice where “it's more and more accepted at all levels, especially primary care, that we're getting overwhelmed with stuff and the complexity of things.”

In the 1960s, clinical computing pioneer Dr. G. Octo Barnett launched the Laboratory of Computer Science at Massachusetts General Hospital, which would two decades later develop its diagnostic support system, DXplain. Dr. Mitchell Feldman, a clinical assistant professor of pediatrics at Harvard Medical School and a staff physician at the lab, has worked with Barnett for more than 20 years on DXplain.

Feldman said he's seen necrotizing fasciitis only once in his medical career. Asked during a telephone interview whether his system could do what Jason Maude claimed his could do, flag a diagnosis for necrotizing fasciitis in just two symptoms, Feldman thought a moment to come up with a couple, then sat at his computer and typed into DXplain.

On entering skin tenderness, “one of the hallmarks of necrotizing fasciitis,” Feldman said, that diagnosis appeared as one of several possibilities. “It's not what we would call well-supported, but it would appear on the list,” he said.

Feldman typed again, adding the symptom, “extremity pain.” “It's first on the list.”

DXplain, now a Web-based tool, is queried by clinicians around the globe, “somewhere in the ball park of 10,000 to 12,000 users per year,” he said.

Like the migration of e-prescribing systems during the past decade from stand-alone tools to just another EHR function, “We've got to get these types of decision support tools in the background, working, so it's automatic,” Feldman said.

But Feldman said it was a toss up when or whether diagnostic support would be as commonplace as e-prescribing alerts. “I'd love to say five years, and I'd think five to 10 is feasible and fear that more than 10 was possible,” he said.

“It gets to the essence of what physicians do, and that can be a bit scary,” Feldman said. Hospital administrators, he said, may think, “We don't need systems like these because that's what docs do.”

Andy Copeland, Aimee's father, said if diagnostic technology “has a potential to save 100 lives, that's significant. If it saves one life, that's significant.”

“It's not just recognizing necrotizing fasciitis,” he said. “There are other conditions we've learned about. I would imagine that if the system can find necrotizing fasciitis, and can recognize these other rare things, it would be well worth it.”