By Sandra G. Boodman

This KHN story was produced in collaboration with The Washington Post

Doctors at a Northern California hospital, concerned that a 40-year-old woman with sky-high blood pressure and confusion might have a blood clot, order a CT scan of her lungs. To their surprise, the scan reveals not a clot but large cancers in both breasts that have spread throughout her body. 

Had they done a simple physical exam of the woman's chest, they would have been able to feel the tumors. So would the doctors who saw her during several hospitalizations over the previous two years, when the cancer might have been more easily treated.

A middle-aged man admitted to a Seattle emergency room for the third time in six weeks displays the classic signs of liver cirrhosis for which he has been repeatedly treated, including swollen legs and a distended abdomen. 

But a veteran doctor spots a telltale indicator of a different disease: rapid inward pulsations just beneath the man's right ear. The patient's problem is not his liver but his heart: he has constrictive pericarditis, a serious condition that requires surgery. 

Both cases reflect a phenomenon that some prominent medical educators say has become increasingly commonplace as medicine becomes more technology-driven: the waning ability of doctors to use a physical exam to make an accurate diagnosis. Information gleaned from inspecting blood vessels at the back of the eye, observing a patient's walk, feeling the liver or checking fingernails can provide valuable clues to underlying diseases or incipient problems, they say.

But over the past few decades the physical diagnosis skills that were once the cornerstone of doctoring have withered, supplanted by a dizzying array of sophisticated, expensive tests. 

"A lot of people downplay the physical exam and [wrongly] say it's fluff," said Salvatore Mangione, associate director of the internal medicine residency at Philadelphia's Jefferson Medical College and director of its physical diagnosis curriculum.

In a 2012 article in the Cleveland Clinic Journal of Medicine, Mangione wrote that he has seen "many cases in which technology, unguided by bedside skills, took physicians down a path where tests begot tests and where, at the end, there was usually a surgeon, and often a lawyer. Sometimes even an undertaker."

To address the problem, programs to revive and teach physical diagnosis - also known as bedside medicine - are underway at some medical schools, including Stanford, Jefferson and Johns Hopkins. The programs are predicated on a belief that these skills are an essential adjunct to technology and can boost diagnostic accuracy, curb unnecessary and expensive testing and foster a greater connection between patients and doctors, many of whom spend increasing amounts of their day staring at their computers rather than looking at the patients they are treating.

At Hopkins, a Web-based program called Murmurlab.org seeks to improve young doctors' ability to use a stethoscope -- a tricky skill that studies have shown is lacking - to distinguish serious cardiac problems from far more common benign heart murmurs.

The goal is to reduce unnecessary and costly echocardiograms.

"There are two reasons it remains crucial to do this [physical diagnosis] at least as well as doctors did 100 years ago," said internist and best-selling author Abraham Verghese, senior associate chairman of Stanford's program on the theory and practice of medicine. Verghese was instrumental in creating the six-year-old Stanford Medicine 25 program: 25 physical exam skills that students are required to learn, demonstrate and teach. These include assessing enlarged lymph nodes, measuring ankle reflexes and performing a knee exam.

"We can pick off the low-hanging fruit - the obvious diagnosis that one can miss at great cost to the patient," such as the woman whose metastatic breast cancer was repeatedly missed, Verghese said. In his view, the physical exam also represents an "important transactional moment" between doctor and patient - a laying-on of hands that helps foster trust. An increasingly common complaint from patients, he said, is that "the doctor never touched me."

Overreliance on technology, he said, has produced perverse results. "If you come to our hospital missing a finger," he quipped, "no one will believe you until we get a CT scan, an MRI and an orthopedic consult."

Differentiating heart murmurs

But some experts are skeptical that reviving the physical exam is the best approach in the 21st century. Robert Wachter, former chairman of the American Board of Internal Medicine, said he shares Verghese's concerns about declining clinical skills. But Wachter said he isn't sure that "restoring the physical exam of yore" is a solution.

"Taking time and energy to train doctors in the physical exam may be less valuable than teaching them how to communicate or to analyze . . . data," said Wachter, associate chairman of medicine at the University of California at San Francisco. "You've got to make some choices."

There is general agreement that the technological explosion that began in the 1980s led to the decline of bedside skills.

Insurance that pays for tests but gives short shrift to a careful and time-consuming history and physical exam accelerated the trend, as has the growing paperwork burden doctors face. The generation of influential mentors who taught physical diagnosis has largely retired. Even bedside rounds - where such knowledge was often imparted to impressionable neophyte physicians -- are mostly a thing of the past, migrating from a patient's hospital bed to a conference room down the hall where test results and the chart -- not the actual patient -- are examined.

Too often, physical exam skills are dismissed as inferior relics of the past when compared with "the glitter and perceived objectiveness of modern technology," said Steven McGee, a professor of medicine at the University of Washington and the author of a recent textbook on evidence-based physical diagnosis.

McGee said that studies have found that physical exam findings can be as accurate as their technological counterparts. Case in point: A pair of studies involving 185 acutely dizzy patients found that the presence of certain abnormal eye movements were more accurate than an initial MRI scan in distinguishing a serious stroke from a benign inner ear problem.

The enormous amount of technology that doctors now must master has crowded out physical diagnosis, he said. But, he noted, "there is a giant chunk of diagnosis that still depends on what we see and detect" through observation and a physical exam.

For a surprising number of diseases, McGee added, diagnosis is based on observation and examination, not a test. Among them are Parkinson's disease, shingles, drug rashes and constrictive pericarditis.

These days, medical students often train on actors who are only pretending to have medical problems, notes Poonam Hosamani, a newly minted hospital-based internist who joined the Stanford team last year.

Hosamani said that she recently enlisted her husband, who has a bad knee, as a featured patient. Many students told her they had never seen a patient with a knee problem. "When we bring in patients with real pathologies, the students are very excited about that," she said. "We have to show them that this is worth their time and demonstrate how much information you can gain" through a good exam, which is not intended to replace technology but to guide its use.

Internist John Kugler, an assistant professor of medicine at Stanford, said that typically medical students learn diagnosis skills before they have seen patients. "They are taught where to put their hands, but these techniques are taught in isolation and there is little to no reinforcement," he said.

W. Reid Thompson, a pediatric cardiologist at Hopkins, launched Murmurlab, a website containing the normal and abnormal heart sounds of more than 1,300 people, in part to curb unnecessary referrals for echocardiograms, which cost up to $900 apiece.

Heart murmurs in children, Thompson said, are common -- between 60 and 70 percent of children have them -- but only about 1 percent are problematic. Distinguishing "innocent" murmurs from serious ones, he said, is an essential skill for physicians, not just cardiologists. But studies have repeatedly found that many doctors do a poor job with auscultation, or listening to the heart and lungs with a stethoscope.

Despite doctors' reliance on a plethora of sophisticated tests, auscultation remains "a fundamental clinical skill," Thompson says. "Every day . . . I walk up to a patient and the first thing I do is listen" to the heart. "People walk around with a stethoscope not just because it looks good or is expected, but because there is information to be learned."

But Thompson said it is not yet clear whether Murmurlab has improved doctors' skills. Stanford officials say they are attempting to devise ways to measure the impact of their program as well.

Lots of data, little interaction

In a recent essay, Arnold Relman, a former editor of the New England Journal of Medicine, described the months he spent last summer at Massachusetts General Hospital after he broke his neck in a near-fatal fall. "Doctors now spend more time with their computers than at the bedside," wrote Relman, an emeritus professor of medicine at Harvard. Reviewing records of his hospital stay, Relman "found only brief descriptions of how I felt and looked" but "copious reports of the data from tests and monitoring devices." Conversations with his doctors were "infrequent, brief and hardly ever reported."

McGee said that he once saw a nurse tell a resident that a patient had spiked a fever and watched as the young doctor frantically scrolled through the electronic medical record searching for a cause, instead of walking down the hall to the patient's room to discover the reason: an inflamed IV site.

"In most hospitals today, the average amount of time a busy intern spends with a patient is four minutes," said Brendan Reilly, who until recently was the executive vice chairman of medicine at New York-Presbyterian Hospital. No longer are tests ordered based on the results of a careful physical exam and history, Reilly said, but the "technological tests become the primary source of information on the patient. It's backward now," and the process is driving up health-care costs and subjecting patients to the risks posed by sometimes unnecessary, risky procedures.

"Doctors trained outside the U.S. are much better clinically than young American doctors," said Reilly, the author of "One Doctor," an unsparing 2013 account of his medical career. They are trained -- or forced by circumstance -- to rely less on technology and more on physical diagnosis skills.

The Stanford Medicine 25 program reflects Verghese's medical training in Ethiopia in the 1980s. Doctors were required to hone their clinical skills because technology was largely nonexistent.

"In some ways," Reilly said, "what Verghese is doing is opening people's eyes and showing that medicine can be a lot of fun."

Reilly said he hopes the accountable care organizations that are part of the new health law - groups of doctors that band together with hospitals to improve the quality of care for patients and share in cost savings - might boost the effort to revive bedside medicine. "The current system is so ridiculous and inefficient and expensive that we're going to have to go back to doing some of the old stuff."