by Lisa Sanders
She suddenly looked up in distress, and her body convulsed in a paroxysm of coughing. She grabbed a tissue and held it against her mouth under the mask. She gasped for air as the spasm tore through her upper body. Tears streamed down her face. Finally she was quiet. She wiped her mouth with the tissue, then showed me the dark bloody sputum. “I think. I’m dying,” she gasped, drying her face with the edge of the sheet. I tried to reassure her that she would be okay, but I worried that she might be right.
On exam, she had no fever but her heart was racing and she was breathing more rapidly than normal. And despite being on an oxygen mask getting 50 percent oxygen (normal air contains only 20 percent oxygen), she still wasn’t getting enough. The oxygen saturation of her blood was 90 percent (normal is 100 percent). Her neck was stiff. She couldn’t lower her chin to her chest, a sign suggestive of meningitis, an infection in the lining of the brain. When I listened to her chest, there were coarse crackling noises—like the noise of a crisp sheet of paper slowly being crumpled.
The blood work sent by the ER doctors showed an elevated white blood cell count. Her chest X-ray was dotted with white cloudlike masses a little smaller than golf balls.
At the nursing station Clark and I reviewed the data and tried to put the story together. She clearly had more than one infected organ system: she probably had pneumonia, and meningitis seemed likely too. As an intravenous drug user, Clark reminded me, she was at high risk for accidentally injecting bacteria from her skin directly into her bloodstream. From there, these aggressive bugs can go anywhere and infect almost any part of the body. It seemed likely that these bacteria had infected her lungs and possible that they had infected her heart and her brain as well. The emergency room docs had already started her on several broad-spectrum antibiotics. We needed to get a head CT and a lumbar puncture to look for an infection in her brain and an echo of her heart to look for infections there.
As I wrote the orders, Clark’s pager went off. It was the ER. There was another admission waiting for us downstairs. He looked toward the patient’s door, clearly torn about whether we were done thinking about this patient or not. When the beeper went off again he stood, reviewed what else had to be done, and left me to finish up as he ran down to the emergency room.
When I had finished my note, I put it in the chart and went in to see the patient once more. She was lying back on the bed now, but if anything she looked worse than she had earlier. Her hair was drenched with sweat and her chest heaved with every breath. I needed to go down to the ER but I couldn’t bring myself to leave her alone. Did she really look worse or was it simply the anxiety of a brand-new intern? I didn’t know, but what I did know was that I afraid to leave her room, afraid that she really was dying.
The respiratory therapist came in and gave the patient a breathing treatment with albuterol—a medicine to reduce wheezing. Desperate with uncertainty, I followed him out of the room and asked him how he thought she looked. “I’ve seen worse,” he told me before hurrying off as his beeper sounded.
I stood frozen at the doorway. I didn’t want to leave because she looked so sick and yet I couldn’t think of anything to do. Why was I more worried than the resident or the respiratory therapist? They had certainly seen more sick people than I had. And yet I couldn’t shake this concern. I pulled out the card on which I’d written Clark’s pager number. I had to talk with him to figure out what I should do. Before I could dial the number, David Roer, the attending physician, strode up. He was in his early forties and had dark hair and an open pleasant face. He greeted me with his usual cheer and asked me about the patient. I gave him a brief report and told him of my concern and then followed him into the room. He spoke with Jennifer briefly, then did a quick physical exam. I trailed him back to the nursing station, eager to hear his assessment. “This patient is on the verge of respiratory arrest.” His tone was kind, without a hint of reproach. “She really needs to be in the ICU. She’s probably going to need to be intubated.”
Hearing those words, shame flooded over me. And relief.
Of course this is what she needed. Why hadn’t I thought of this? My cheeks burned as I buried myself in the business of transferring the patient to the intensive care unit. Once she’d been moved into her new home, I ran down to the emergency room to see our next admission. The rest of that call day was a blur of admitting more new patients, following up on studies, and getting sign-out on the patients cared for by the other house staff as they headed home.
By the time I had finished all the tasks on my to-do list and trudged up to the sixth-floor call room, the predawn sky was beginning to lighten. I was tired but couldn’t sleep. I went through every step of what had happened with Jennifer and tried to figure out how I had gone so wrong in the plans I had so carefully put together for her—plans that didn’t take into account her most pressing and life-threatening issue, her breathing. It was right in front of me. And when her condition worsened so quickly—as I think it must have—I saw too that she was sick, dangerously sick, the kind of sick I’d heard so much about. The real surprise to me was that recognizing that she was sick had not helped me know what to do about it. I don’t think I figured this out that night but what I learned over the course of that month—and relearned many times over the years of my specialty training—is that, as important as it is, recognizing what “sick” looks like is only the first step.
In fact, several studies have demonstrated that the recognition of what “sick” is, while much touted by residents and many experienced physicians, has not been shown to be accurate or effective in guiding medical decision making. In one study done at Yale, John Mellors, then a fellow in infectious disease, followed 135 patients who came to the emergency room with a fever and no obvious source of infection. The decision that had to be made at that time was whether these patients had a virus—in which case they could safely be sent home for rest and TLC—or whether there was a chance that they had a bacterial infection that would require them to take antibiotics. All of the patients in the study had blood cultures and complete blood counts drawn, a chest X-ray and a urinalysis performed. The decision to either admit the patient or discharge him, with or without antibiotics, was made after all the results except for the blood cultures were reviewed.
All of the patients enrolled in the study were followed throughout the course of their illness. Then the researchers compared how sick the patients really were with how sick the physicians had thought they were when they were initially seen in the ER. The doctors were wrong far more often than they were right. Many of the patients who were judged to look quite ill and were admitted were discharged soon afterward with no medical interventions taken. And four patients, assessed as being “not toxic” and sent home without antibiotics, were ultimately found to have significant bacterial infections and had to be called back to the emergency room to get antibiotics. One of those patients died not long after being discharged from the emergency room, well before the doctors even had the chance to call him back in.
Other studies have also found that our instincts, our intuitive responses, to a “sick”-appearing patient are frequently wrong. Recognition that a patient appears sick is important, it turns out, but it’s not sufficient. As the Mellor study showed, patients can look extremely sick and not have a dangerous illness. Other patients, and this is particularly true of the elderly, can look remarkably well—at least for a while—despite a life-threatening infection. How sick a patient looks is just a clue, a single piece of data. Alone it is practically meaningless.
So what will help predict sickness? Concrete measures. Abnormal vital signs are key—a blood pressure that is too low or too high, a heart rate or respiratory rate that is too fast or too slow. Abnormal skin color or mental status. We are very good observers of abnormality. However, we often respond immediately and viscerally to a patient’s condition before we’ve even identified the abnormality that’s the cause of the concern. The fear I felt in Jennifer’s room was such a response. I reco
gnized sick but hadn’t gone the next essential step of identifying what was causing the fear and so I didn’t know what to treat.
When the attending first saw Jennifer, he immediately recognized that she was dangerously ill. He then noted the abnormal respiratory rate, the effort she was expending to breathe. She was using the muscles in her neck and shoulders to perform an act that is normally simple and effortless. Moreover, despite the hard work she was doing, she still was not getting enough oxygen into her bloodstream. These are ominous signals. As a medical student I had read about how patients working this hard to breathe can tire out and die. I knew it and yet that knowledge didn’t help me. I saw—it’s probably how I knew she was sick—but I didn’t recognize what I saw and so was unable to figure out what to do.
I followed Jennifer’s course over the next week. As predicted, she wasn’t able to sustain the effort it took to breathe and was intubated the next morning. Her blood cultures grew Staphylococcus aureus, an aggressive and destructive bacterium that lives on the skin. It is a disastrously common infection among intravenous drug users. Despite the powerful antibiotics, she continued to deteriorate. Her blood pressure dropped so that she needed medications to keep her blood circulating effectively. Then her kidneys failed. Her blood stopped clotting. After seven days in the ICU, Jennifer’s heart and lungs failed her and she died.
I don’t think the delay in getting Jennifer to the ICU had a major impact on her prognosis. I made important mistakes in my training—we all do—mistakes that hasten or even cause death in those at the boundary between life and death. But I don’t count Jennifer among my mistakes. She had a severe infection and precious little reserve. Nevertheless, I think of her often. Those minutes of terror and confusion I felt standing powerless in her room served as a visceral reminder throughout my training (and even now, occasionally) that the big picture isn’t enough in medicine; that the overall impression of a patient is worthless without looking further and paying attention to the specific measurements of health or sickness that were behind the impression in the first place.
Research into human perception reveals that we have developed a remarkable ability to quickly gather visual data and come to a conclusion without even noticing the steps we take to get there. Studies in perception show that this rapid automatic use of our eyes is by far the most efficient way to collect visual data. And most of the time, that’s good enough. Not so in medicine. Inexperienced doctors, like my intern self, need to learn to make themselves work backward from the conclusions they reach, attend to the details that got them there, and translate what they see into the language and numbers of medicine. Only then can we at least try to help the patient.
Noticing What You See
Sherlock Holmes perhaps expressed most succinctly the lesson I learned. “I have trained myself,” Holmes tells his amanuensis, Dr. John Watson, “to notice what I see.” It’s an important distinction.
“You have been in Afghanistan, I perceive.” With these first words Holmes initiated the quirky relationship with the man who would become his closest friend and most devoted follower. Watson, in London recovering from war wounds sustained in Afghanistan, is shocked by the man’s declaration. How could he possibly have known this? Had he been told? “Nothing of the sort. I knew you came from Afghanistan.” He retraces his reasoning. Watson’s military bearing suggested some time spent in the armed services, Holmes tells him. The deep tan indicated a recent return; his wasted physique, some kind of intestinal fever. And his injured arm pointed to a war zone.
Of course it is an easy enough trick to pull off in fiction. However, Arthur Conan Doyle based his most famous character on a Scottish surgeon named Joseph Bell, for whom he’d worked during his medical training. Like Holmes, Bell frequently wore a deerstalker cap, smoked a pipe, and was often observed using a magnifying glass. But the most important trait they shared was a keen eye for detail combined with remarkable deductive powers.
Stories about Bell sound like snippets straight out of a Holmes story. In a preface to one of his books, Doyle describes his debt to Bell in developing Holmes as a character and provides examples of Bell’s Holmes-like abilities. Seeing one patient, a young man in street clothes, Bell immediately asks the man if he was recently discharged from the military. He was. Was he a noncommissioned officer in the Highland Division? He was. Stationed in Barbados? Yes, how did he know all this? Like Holmes, Bell delighted in revealing his observations to the patient, the medical students, and the doctors observing him. Doyle quotes Bell’s response: “‘You see gentlemen,’ he explain[ed], ‘the man was a respectful man but did not remove his hat. They do not in the army, but he would have learned civilian ways had he been long discharged. He has an air of authority and he is obviously Scottish. As to Barbados, his complaint is elephantiasis, which is West Indian and not British.’ To his audience of Watsons it all seemed quite miraculous until it was explained and then it was simple enough. It is no wonder that after the study of such a character I used and amplified his method when in later life I tried to build up a scientific detective.”
Doyle clearly recognized that Bell’s powers of observation were extraordinary. He referred to himself and the other doctors who witnessed these remarkable instances of detection as “Watsons.” Yet Holmes and his model, Bell, firmly believed that this kind of close observation of significant details could be taught and sought to instruct those around them. “From close observation and deduction you can make a correct diagnosis of any and every case,” Bell wrote in a letter to his now famous student, Arthur Conan Doyle. With practice, he suggested, the power of observation can be sharpened, improved. Doctors, he seemed to suggest, can teach themselves to “notice what they see.”
Learning How to See
Medical schools across the country have recently joined ranks with the historic Joseph Bell in striving to teach medical students to be better observers. One of the first efforts came from Yale. Dr. Irwin Braverman, a professor of dermatology for over fifty years, had long been frustrated by the difficulty students had in describing findings of the skin. It might have been a knowledge deficit—easily remedied with books, pictures, and tests. But Braverman suspected that what his students principally lacked was the skill of close observation. Too often they wanted to cut straight to the answer without paying attention to the details that took them there.
“You teach students to memorize lots of facts,” he told me. “You say: ‘Look at this patient. Look at how he’s standing. Look at his facial features. That particular pattern represents one disease, and this pattern represents another.’ We teach those patterns so that the next time the doctor comes across it, he or she comes up with a diagnosis.” What’s missing, says Braverman, is how to think when an oddity appears. That requires careful and detailed observation. After years of teaching he still wasn’t certain he’d found the best way to communicate that complex set of skills.
In 1998 Braverman came up with a way to teach this skill. What if he taught these young medical students how to observe in a context where they wouldn’t need any specialized knowledge and so could focus on skills that couldn’t be learned from a book, where the teaching would force students to focus on process, not content? He realized that he had a perfect classroom right in his own backyard, in Yale’s Center for British Art. The course, now part of the curriculum, requires first-year medical students to hone their powers of observation on paintings rather than patients.
As I entered the cool soft light of the museum’s atrium, a dozen first-year students were standing around in small groups, waiting to enter the conference room to find out what they were doing in this unusual setting. Braverman, a round-faced man with a comb-over and an impish smile, sat at the head of a long table of lustrous dark wood like a folksy CEO of some big corporation. Their job that afternoon, he told them, was to look at the pictures they were assigned to and then just describe them. Not too hard, right? He looked around hopefully. A few students sitting near him smiled and nodded e
nthusiastically. The rest of the table was a harder sell. “It’s always like that,” Braverman told me as we followed the students up the stairs to the third floor, where most of the nineteenth-century paintings he liked to use were on exhibit. “A handful of students either get it right away or are just habitually enthusiastic. The rest of the students here need to be convinced. But, you watch, by the end of the afternoon, I’ll have some converts. Wait and see.”
Once stationed at their assigned pictures the students reviewed the rest of the rules. They were not to read the little labels next to the paintings. They’d have ten minutes to look at the pictures and then together the class would discuss the images, one by one. Each of the pictures would have a story to tell. It was the student’s job to figure out what that story was and relate it to the rest of the group, using only concrete descriptive terms. If you think a character looks sad, he told them, figure out what you are seeing that makes you think that and describe it. If you think that the picture suggests a certain place or class, describe the details that lead you to that conclusion.
A tall young man with a sweet face and a prominent Adam’s apple peered at the image of a slender man whose upper torso was hanging limply over the side of a bed, his right hand touching the floor. His eyes were closed. Was he asleep? asked Braverman.
“No,” he announced decisively to his fellow medical students gathered around the scene. “He could be drunk—he has a bottle in his hand—but he’s not asleep. I think he’s dead.” “How do you know that?” asked Braverman. “His coloring—it’s not right. He looks green,” he answered thoughtfully. “And there’s death all around him.” He described the sad scene. The young man lies in a small, unadorned garret apartment. An indifferent landscape of rooftops dark in the changing light of a setting sun is silhouetted outside the narrow dirt-encrusted windows. Petals of a dying rose ornament the windowsill, their color gray in the fading light. Shreds of torn papers are strewn across the floor. “I think he’s taken his own life,” he concluded triumphantly.