Software Aggregates Medical Knowledge In Hopes Of Averting Misdiagnoses

The Isabel pediatric diagnostic system--based on pattern-recognition technology--prompts physicians to consider a range of different diagnoses, some of which might not be immediately apparent.
In April, 1999, 3-year-old Isabel Maude might have died. Her family physician and the doctors in her local hospital's emergency room in the United Kingdom failed to recognize that she had developed potentially fatal complications from chicken pox, necrotizing fasciitis among them.

"It didn't strike her doctors that what she had was this life-threatening condition and they should be admitting her and treating her," explains Dr. Joseph Britto, then pediatric intensive-care consultant at St. Mary's Hospital in London. "Instead, they sent her home. I looked after her in a children's intensive-care unit in London and her parents, rather than be angry and litigious, said, 'Why don't we do something about this?'"

During Isabel's two-month convalescence following multiple organ failure and cardiac arrest, the Maudes and Britto came up with the idea for a software-based clinical-decision-support system to reduce the chance of misdiagnoses for future patients.

A 1999 report by the Institute of Medicine, "To Err Is Human," found that an estimated 98,000 preventable deaths occurred each year due to medical errors. And the situation hasn't improved since then, according to a July 2004 report by HealthGrades Inc., a health-care ratings, information, and advisory services company, noting that nearly half a million lives have been lost needlessly in the five years since the Institute of Medicine report appeared.

According to the HealthGrades report, "The United States loses more American lives to patient safety incidents every six months than it did during the entire Vietnam War. This also equates to three fully loaded jumbo jets crashing every other day for the last five years."

Thus was born the Isabel pediatric diagnostic system. Launched in 2002 with funding from The Isabel Medical Charity, and support from the U.K. Department of Health, the U.K. National Health Service, and other medical institutions, Isabel is a hosted Web service that relies on pattern-recognition technology from Autonomy Corp., a maker of information-retrieval software. With minimal input from clinicians, the system prompts physicians to consider a range of different diagnoses, some of which might not be immediately apparent, particularly to less experienced professionals.

Only a few months after launch, the system had 20,000 users from more than 100 countries, according to Isabel Healthcare Ltd., a commercial subsidiary owned by the Isabel charity that has taken over development and commercialization of the diagnostic system. As of Deccember 2004, 11 hospitals in the United States and the United Kingdom are customers.

"There are several reasons why we make mistakes," says Britto, now director of Isabel Healthcare Ltd. "One is because biomedical knowledge is increasing exponentially. Every 20 years, it doubles. I think my colleagues and I in medicine are different in that as knowledge workers, we tend to rely almost exclusively in the knowledge that we carry around in our minds, particularly in the clinical workflow at the bedside. And that's why I think we make mistakes."

Dr. Neal Thomas, assistant professor of Pediatrics and Health Evaluation Sciences at Pennsylvania State University College of Medicine, gives the system high marks. "Coming from the intensive-care unit, a lot of times, things aren't straightforward," he says. "The value of Isabel is the fact that they have amassed such a large differential diagnosis for things that it makes you think about abnormal diagnoses much more."

Particularly in communities without substantial medical expertise, he says, it should lead to fewer misdiagnoses, if people take the time to use it.

While Britto acknowledges that some physicians initially resisted the technology, he says that's changing. "There's increasing emphasis on the quality of care, and increasing litigation," he says. "One diagnostic error like little Isabel's could wind up costing a hospital a million dollars in a lawsuit."

He says it's also a workflow issue. "The introduction of electronic medical records is facilitating the use of systems like this," he explains.

Although Thomas says his hospital is evaluating whether it wants to bear the cost of becoming a subscriber (the service used to be free), he believes tools like this have their place in medicine. "Physicians can't ignore the fact that we can't remember everything," he says. "And we have to learn how to take advantage of tools such as this."