Every Patient Tells a Story (36 page)

Mark Graber and a colleague tested the system with fifty case studies drawn from the pages of the
New England Journal of Medicine
. Since Isabel accepts information two ways, the researchers tested it in both modes. In one, Graber manually typed in three to six key findings from each case study. On average this took less than a minute. The correct diagnosis was included in the list of possible diagnoses generated by Isabel in forty-eight of the fifty cases (96 percent). When the text of entire case studies was cut-and-pasted into Isabel (an artificial, but easy, approach) accuracy declined dramatically, with the correct diagnosis appearing in only thirty-seven of the fifty cases (74 percent).

The authors note that this performance shows that diagnostic decision support systems have evolved significantly since the first-generation systems developed in previous decades. Still, there are many of the same barriers to wide acceptance of the system. Because Isabel and other systems like it are not fully integrated with other medical information systems, data has to be entered into the system by the physician. This is time-consuming and tedious, although Isabel seems to have worked hard to minimize the work involved. Using this system, doctors can describe the patient’s symptoms in everyday language. And the machine is smarter, so the amount of detailed information required is much smaller.

But more important, doctors must decide when to use the system. By far the most common diagnostic error in medicine is premature closure—when a physician stops seeking a diagnosis after finding one that explains most or even all the key findings, without asking himself that essential question: what else could this be? If a doctor is satisfied with his diagnosis, he is unlikely to turn to a digital brain at all, and thus the potential value of the system is lost.

So even this new generation of clinical decision-making systems such as Isabel, improved as they are over older programs, is still not widely used. Even Dr. Bergsagel, whose use of Isabel so vividly illustrates that system’s power, says he uses it only a few times a month.

“The systems available today are still cumbersome to use,” says Jerome Kassirer. “Doctors still have to input all sorts of stuff into these programs … and nobody has the time to type it all in. Besides, most of the time you don’t need the system. Most of the day-to-day issues a doctor sees are amenable to the traditional kinds of diagnostic approaches that we’ve used for years. In fact, it’s easier these days because we’ve got echos and CT scans and MRIs.”

One final impediment exists for Isabel and its competitors: price. Isabel is made available to hospitals on a per-bed cost basis, which works out to about $80,000 for a typical hospital. Individual doctors can buy the service for an annual fee of $750.

Although hardly unaffordable by either institutions or individual doctors,
the cost of commercial diagnostic decision support systems means that such programs are vulnerable to competition from what might seem like an unlikely quarter: Google.

Googling a Diagnosis

Patients, friends, and family have periodically confessed to me that they regularly use Google to investigate their own symptoms. My adolescent daughter does it whenever she is baffled by one of her own body’s new and peculiar ways. They are not alone in this. According to a 2005 survey done by the Pew Center, 95 million Americans looked for health information on the Internet. I’ll bet that most of those people somewhere along the line in their search used Google.

I got an e-mail several years ago from a reader who had managed to diagnose herself using Google when she developed fever and a rash. She didn’t start with Google. She started with a man she had always trusted—her doctor.

“I always heard that when your palms itched it meant you were coming into money,” she told her doctor when he entered the exam room. “No money so far,” she continued, “but lots of fever.” Dr. Davis Sprague eyed her attentively. They’d known each other for years, and despite her playful tone he thought she looked pretty sick.

She’d been well until a few days earlier, she told him. She had a little pain when she went to the bathroom, which made her think she had a urinary tract infection, and so she’d increased her fluids. That didn’t work, so the next day she came in and saw a different doctor, who started her on an antibiotic and a painkiller. She didn’t get better; in fact, that’s when she first noticed the itchy palms. The next morning she was so achy she could barely get out of bed. That night, she had shaking chills and a fever of 102°.

The rash appeared the following day. It started on her arms, her face, and her chest. She stopped taking the painkiller, thinking the rash could be an allergic reaction to it, she told him. But the rash just kept spreading.

Now Sprague was worried. The patient was fifty-seven years old, and
other than a back injury a few years ago and some well-controlled high blood pressure, she had always been healthy. Not today. He was glad she was the last patient of the day because he could tell this was going to take some time.

On examination she looked tired, and her face was flushed and sweaty. Her short, dark hair lay plastered to her scalp. She had no fever, but her blood pressure was quite low, and her heart was beating unnaturally fast. The rash that now covered her body was made up of hundreds of small, flat red marks. The newest ones, those on her legs, were like red-colored freckles. The ones on her arms and chest were larger—the size of nickels—and less well defined. The rash didn’t itch or hurt. But the palms of her hands, though rash-free,
were
red and irritated. A urine sample showed no evidence of an infection, but was positive for blood. That might have been a result of the fever, or it could indicate kidney damage.

“You need to go to the emergency room,” Sprague instructed the patient. “You may even need to be admitted to the hospital. I’m not sure what you’ve got, but I am pretty sure that this is serious.”

She might have developed an allergy to one of the medicines she was taking, he explained, which could be dangerous and might even require other medications. What he was really worried about, though, was that she had some sort of infection that was spreading throughout her body. In a hospital they could test her blood and get a better sense of what was going on.

The ER doctor ordered what seemed like an endless stream of blood tests as well as a chest X-ray. When all the tests came back normal, he decided she was well enough to go home. It probably was an allergic reaction, he told her, and gave her a different antibiotic. She should follow up with her doctor in a couple of days, he said.

Two days later, she was back at Sprague’s office. She did feel a little better, she said, but she was still having fevers, and now she felt short of breath with even minimal effort. “What do you think is going on?” she asked.

Sprague wasn’t sure. Maybe the ER doctor had been right, and it really was an allergy—she was a little better since they’d changed the antibiotics. But the shortness of breath started after that. He was still worried about infection.
Fever and rash were common symptoms. It could be a viral illness—Coxsackie? West Nile? Or was it bacterial? These symptoms, he told her, were so nonspecific that they could be found in everything from garden-variety Lyme disease to something really exotic like Rocky Mountain spotted fever. “We may never figure it out,” he confessed. But since she was getting better, he was willing to give her a few more days. If she was still spiking fevers then, he’d send off some blood work to try to find an answer.

At home, though, the patient continued to worry. That night she sat down at the computer to do a little research of her own. “Rash, adult, fever,” she Googled.

When you Google a set of symptoms, you don’t necessarily get the most common or the most likely diseases; you get the diseases with the greatest number of links from other Web sites. Her Google search brought up dozens of fairly unusual, but well-linked, illnesses: coccidioidomycosis—a fungal infection most common on the West Coast; dengue fever—endemic to the tropics and near tropics; measles; scarlet fever.

But the patient immediately focused on the first result: Rocky Mountain spotted fever, which her doctor had mentioned. As she read about the disease, she began to feel a little panicky. The description of the symptoms, she said, fit her perfectly: the rash, the fever, the muscle aches. The rash, she read, can involve the palms of the hands, which is pretty unusual. She didn’t have a rash there, but her palms were red and itchy. Also, the disease is transmitted by dog ticks—she had a dog. It’s most common in the summer—it was August. Though it’s rare, it is more commonly seen on the East Coast than in the Rockies, and she was in upstate New York. People can die from this disease, she read. It’s the deadliest of all the tick-borne illnesses.

She called the emergency room where she had been seen. Had they tested her for Rocky Mountain spotted fever? No, she was told, why would they? They had never seen a single case in the area. She hung up feeling somewhat relieved. They didn’t think it was Rocky Mountain spotted fever; Dr. Sprague didn’t think it was. Chances are that it wasn’t.

Over the next few days, the patient started to feel almost normal again. The rash was fading—though now it itched like crazy—and her energy was
coming back. But she continued to have fevers at night and still occasionally felt short of breath. She returned to Sprague’s office one more time. “I’m glad to hear you’re feeling better, but these fevers worry me,” he said. “I want to send off some tests.”

“What about Rocky Mountain spotted fever?” the patient asked. She confessed that she had looked it up on the Internet and thought the symptoms were close to what she had. The doctor thought for a moment. “I don’t think that’s what you have, but let’s add it.” He had heard doctors complain about their patients surfing the Web for diagnoses, but he didn’t mind. He had never seen Rocky Mountain spotted fever—maybe she was right.

The results came back a few days later. “You’re an internist’s dream,” the doctor said with a smile as he entered the exam room. “It really
is
Rocky Mountain spotted fever, and I would have completely missed it if I hadn’t listened to you.” He started the patient on doxycycline—the antibiotic of choice for this bacterium. Her body seemed to be fighting off the illness without it, but he wasn’t taking any chances. Within a few days her fever was gone, the rash was fading, and her palms were beginning to feel normal.

I asked the patient how she felt about her doctor, who had come so close to missing this diagnosis. “But he didn’t miss it. He was the first to think of it. And he sent off the test—even though it could prove him wrong. He just wanted to figure out what was going on.”

This case illustrates a real and growing trend—patients who either diagnose themselves by using the Internet or follow up on their doctor’s diagnosis in that manner. But it’s not just patients using the power of Google and other search engines these days. A doctor wrote to the
New England Journal of Medicine
about an amazing diagnosis made at his institution. The case involved an infant with diarrhea, an unusual rash, and multiple immunological abnormalities. The patient was discussed at length in a case conference with residents, attending physicians, and a visiting professor. No consensus was reached. The letter continues:

Finally, the visiting professor asked the fellow if she had made a diagnosis, and she reported that she had indeed and mentioned a rare syndrome known as IPEX (immunodeficiency, polyendocrinopathy, enteropathy, X-linked). It appeared to fit the case, and everyone seemed satisfied …

“How did you make that diagnosis?” asked the professor. Came the reply, “Well, I had the skin biopsy report, and I had a chart of the immunologic tests. So I entered the salient features into Google, and it popped right up.”

This story and their own experiences with patients who had consulted the Internet for information about their own symptoms prompted a pair of Australian researchers to test Google’s diagnostic accuracy.

Like Graber, they used the medical case studies published in the
New England Journal of Medicine
, selecting three to five keywords from each article, and entered them into Google before they, themselves, read the actual diagnosis. The doctors selected and recorded the three most prominent diagnoses that Google came up with for each case. Then they compared the Google findings with the real diagnosis.

The result? Google flunked. Google found the right diagnosis for only fifteen out of twenty-six cases (58 percent). Of course, Google isn’t designed to provide diagnostic support for doctors, so any right answers provided by the powerful search engine are bonuses. One interesting observation was made by the authors: Google was most accurate for diseases that had unique signs and symptoms or rare presentations. This isn’t surprising to any of us who use Google, but it’s interesting. As anybody who has used a search engine knows, the more unusual your target is, the easier it is to find. For example, if you want to Google two friends, you are much more likely to find the one named Ionia Khammouane than the one named Ann Jones. Information on Ionia is going to pop right up, just like the diagnosis of the case of the child with leukemia and the brown marked rash.

What’s interesting is that it’s precisely the unusual disorders—the ones with peculiar symptoms that doctors rarely see—that can be most baffling to both doctors and patients. In the case I presented in an earlier chapter, a resident in our program was able to diagnose a patient with intermittent
nausea and vomiting because of an unusual symptom—her nausea was improved by hot showers. By Googling that, Amy Hsia was able to identify an unusual and recently described disease called cannabinoid hyperemesis.

Because Google is so universally available, simple, fast, and free, it may become the go-to diagnostic aid for oddball cases. Even the august
New England Journal of Medicine
finds Google “helpful in diagnosing difficult and rare cases.” Google gives users ready access to more than three billion articles on the Web and is far more frequently used than PubMed for retrieving medical articles.