I held my wife Veronica's hand as the technician applied cool gel to her chest. At first, the ultrasound images were the fuzzy black-and-whites I remembered from before our daughters Rebecca and Hannah were born. After a few touches to the LCD screen, a breathtaking three-dimensional movie began to run. It featured Veronica's heart, its thick walls beating yellow against a black background.
The technician maneuvered a trackball to reveal the various parts undulating in unison. Colored regions displayed blood velocity and turbulence through the different chambers. Suspended in virtual space, Veronica's heart looked every millimeter the impregnable pump I had always assumed it was.
Veronica is 46, does four hard workouts every week on the stepping machine, eats sensibly, and has a resting pulse of 60. So when she woke me at 2 A.M. and calmly reported funny chest pains radiating to her shoulder blades and down her arms, the obvious came to mind, but it was hard to really believe. Veronica and Rebecca had been coughing and feverish for a week. The three of us had embarrassing cold sores. Acid reflux, a sore diaphragm -- anything seemed more likely than a heart attack.
You need a hard head and a soft heart to manage a loved one's medical emergency. It's surprisingly easy for smart people to be nudged by circumstance and human frailty into doing careless or foolish things. We had two sleeping daughters across the hall. The thought of them waking up to flashing ambulance lights was daunting. We worried about leaving them or dragging them to an emergency room. Still, Veronica had never felt anything like this. We had to do something. So we threw on some clothes, and drove to the 24-hour urgent-care center a half-mile from our house.
Several people made mistakes in Veronica's care. The worst and most deadly mistake was ours: going to this urgent-care center. Veronica's symptoms demanded a 911 call. I knew better -- or I certainly should have. I am a certified expert, director of the University of Chicago Center for Health Administration Studies. I've served on expert panels of the Institute of Medicine, no less.
I was swayed to discount what was happening --Veronica, a clinical nurse specialist, was, too -- by disbelief, by her recent illness, and by her general fitness. We were also swayed by the expected hassle and expense of an ER visit. We envisioned paying a large bill to be prescribed some Tums. Last year, Veronica went out-of-network for urgent care. That cost $700.
In part, we hesitated because that was exactly what the modern health-insurance system is designed to make us do. A quarter-century ago, the RAND Health Insurance Experiment (HIE) established the basic argument for deductibles and co-payments in insurance. HIE remains the most important policy experiment in American history. Its most potent finding was that people who got free care used 40 percent more services than did others assigned to cost-sharing plans. Yet the free care produced little measurable additional benefit for the average patient. These results are often cited in support of co-payments and deductibles designed to discourage inappropriate care. Policy-makers and payers are particularly concerned about the real and alleged over-use of emergency care. Charging higher co-payments is one obvious response.
It seems counterintuitive that demand for ER services would be sensitive to price. If you slice off your finger with a steak knife, you won't be thinking about the money. Yet it turns out that many ailments -- Veronica's included -- are ambiguous, and so price matters. RAND investigators found that individuals in cost-sharing plans reduced ER use by one-third when compared with the free-care group.
Co-payments did discourage wasteful use among HIE participants. ER visits in relatively non-urgent categories such as sprains and back pain were 47 percent less frequent in cost-sharing plans. Unfortunately, co-payments also discouraged appropriate use. Participants enrolled in the cost-sharing plans were 23 percent less likely to seek ER care for "more urgent" problems, including fractures and asthma.
Most patients cannot reliably distinguish appropriate from inappropriate ER use. In many cases, even experts find the distinction fuzzy. I once co-wrote a study of a managed behavioral health plan that imposed a 50 percent co-payment on psychiatric ER visits. Do we really want to impose these barriers? When someone feels that funny chest pain, how long do we want her to dither before seeking help?
Veronica and I made a critical decision in choosing the urgent-care clinic. Your first medical provider in an emergency determines who will frame the initial hypotheses of your illness, who will coordinate your care, and, often, the person who hears the cleanest direct account of what is wrong. I had never been inside this imposing structure, which advertises and charges as an emergency-department affiliate of a local hospital. We arrived to find it nearly empty. The staff promptly took an electrocardiogram (EKG) that looked normal and administered aspirin and nitroglycerin. Veronica took a gastrointestinal cocktail of antacid and lidocaine in case this was acid reflux. It seemed to help, which I found reassuring. They administered a chest X-ray. After bumpy preliminaries, they administered the standard cardiac-enzyme tests.
Key enzyme levels were very high, indicating that heart cells had died and had released their hidden proteins. Yet the staff remained unsure that the test equipment was working. As the tests were rerun, the staff tried to administer a CT scan, but the intravenous dye infiltrated into Veronica's forearm, causing excruciating pain.
I remained convinced this was all an annoying set of benign, if painful, screw-ups.
I cannot say why I was not more forceful in getting Veronica out of there. Throughout, she seemed fine, talking normally, except that her chest, and then her arm, really hurt. My alarm steadily increased as the realization sank in that something could be genuinely amiss. An amazing four hours after arriving, we received the repeated enzyme tests. That's when the ambulance was called to transport Veronica to a real hospital. I gingerly asked the doctor about taking her to the big university hospital one hour away. He replied, quite reasonably, that there was no time. I raced home and drove the kids to a friend's house.
At the hospital, an emergency-room doctor stated without preliminaries: "Bottom line -- you've had a heart attack." The enzyme tests were definitive. Fortunately there was no other detectable damage. He explained that this was the kind of heart attack, more common than one would suppose, that can leave no obvious damage. A tiny piece of plaque becomes dislodged, initiating clotting. Such an attack can be essentially self-healing once it runs its course. I gave the gruff but comfortably authoritative cardiologist the business card of Veronica's internist and asked him to call.
Veronica needed cardiac catheterization. This is a delicate procedure. Cardiologists and their surgical teams differ substantially in skill and in post-operative mortality. For 25 years, health-services researchers have documented that it's good to have an operation in the right hospital by the right people. Many jurisdictions have begun to publish hospital-specific and surgeon-specific rankings of observed and expected mortality rates for these procedures.
As you might imagine, ranking is a complicated subject. Hospitals complain they are penalized because they serve high-risk, complex patients. Hospitals may also game things. There is suggestive evidence that cardiac report cards encourage physicians to provide less-aggressive treatment to minority patients and others who tend to have worse outcomes. Risk-adjustment methods developed to address these concerns have spurred needed changes. A striking number of surgeons in the highest mortality categories retired or moved away when New York implemented report-card systems. A 2006 Health Affairs paper by Ashish Jha and Arnold Epstein reports: "With the release of each report card, approximately one in five bottom-quartile surgeons relocated or ceased practicing within two years." New York's post-operative mortality rates sharply declined after ratings were published. Rankings were not the only reasons for improvement, but they helped.
Not surprisingly, high-volume facilities perform better. Surgeons get better with practice. Care teams get better at minimizing post-operative infections. Some hospitals become popular because they are good; others become good because they are popular. Which came first? If you're a patient, you don't care. There are ongoing debates over whether cardiac catheterization and other delicate services should be provided by a small number of high-volume regional centers. Probably they should, though this is hard to pull off in our decentralized and competitive system. The data also reveal surprising disparities, sometimes between adjoining hospitals or those we might otherwise consider peers.
New York state publishes risk-adjusted 30-day mortality rankings. Based on 2003–2005 data (released last February), where would you want your ambulance to go in the New York area? You might not guess that Bellevue Hospital and the Long Island Jewish Hospital performed markedly better than many more famous hospitals. You might not suspect that Montefiore-Einstein Heart Center ranked poorly in both mortality and post-operative complication.
I have presented this information to hundreds of students at Yale, the University of Michigan, and the University of Chicago. I could cite a wealth of data on many topics. Yet when Veronica got sick, my personal databank included nothing on the hospitals near my own home. You don't comparison shop alongside a loved one's hospital gurney.
As the bedside conversation proceeded, I wondered whether to sell our house. I wasn't thinking about the sub-prime mess. I just wanted to live near a great cardiac facility. A classic analysis by Mark McLellan, Barbara McNeil, and Joseph Newhouse showed that people who happen to live near these hospitals were more likely to survive cardiac emergencies. I wish I had taken that paper to heart.
The community hospital that treated Veronica is, by reputation, probably the best within 10 miles of us. The attending cardiologist is well respected and projected an infectious certainty about what was wrong, how to fix it, and who was in charge. I found his decisiveness reassuring. Still, I would rather have had this performed at a major academic medical center or at least done by someone I had vetted. I again rather awkwardly asked the emergency-room docs whether Veronica should be moved. I called a friend who is a good internist who said they seemed to be doing sensible things, and there was no time to screw around moving her. Given the situation, there was nothing else to do.
The team whisked Veronica upstairs for the angiogram. They threaded a catheter into her groin area and ran it up near the heart to examine arteries that might have been blocked. I sat pensively in the waiting area. The cardiologist shortly emerged to report that the angiogram had gone well. There was no observable tissue damage. There were no blockages. Her arteries were squeaky clean.
Days later, I looked up the local rankings. Our hospital wasn't ranked badly. Its cardiac catheterization is 40 percent cheaper than the fancy university hospital I preferred. The bad news: Its post-operative mortality rate was 40 percent higher than that of another community hospital I never held in much regard three miles from our home.
The various waiting rooms were especially sobering. Dozens of tight-lipped people filled them, worried, first and foremost, about their loved ones. The hospital is located in a gritty South Chicago suburb. Many of the people sitting with me were surely wondering, how will I pay for this?
I wasn't worrying about money. I remember thinking: Thank God we have good insurance. At least I think we do. Six months later, I still don't know how much this episode will ultimately cost. I am confident we will not go medically bankrupt, as many patients do with limited or no insurance. Jonathan Cohn's book Sick describes Chicagoans' struggles with medical debt, including a poor, semi-retired nun sued by a Catholic health system. Sitting in that waiting room, I was also struck by the responsibility each of us has to care for our mind and body. We are vulnerable to genetics and bad luck. Still, the intensive care unit brutally displays the consequences of poor health behaviors. Surprising numbers of young people are there, suffering and sometimes dying when this doesn't have to be.
It was hard not to notice something else. That waiting room, like so many others I have frequented in my 15 years in public health, was filled with people of color. Public perceptions of racial and ethnic disparities are shaped by headlines about homicide, substance abuse, infant mortality, AIDS. Mundane cardiovascular diseases exact a far heavier toll in minority communities, within which child and adult obesity have markedly worsened. I fear that waiting rooms may need more chairs.
Within a few hours after the angiogram, Veronica was in intensive care, and we began to digest the bizarre news. Once the anesthesia wore off, she felt real chest pain but was otherwise amazingly normal. Wired up to the monitors, she was soon sitting up doing her cross-stitch, joking with my sister, asking about the kids. An infectious-disease specialist came through and treated her cold sores. Things became boring.
Veronica stayed in that ICU for three days. A pneumatic messaging tube thwonked loudly and randomly throughout the night. Various machines would beep if Veronica moved her arm and impinged on some tubing. On top of that, Veronica was in pain, which the cardiologist explained later was a normal reaction to blood returning to the damaged heart areas. The effect is grueling. Sleep disruption is a prominent cause of what is charmingly labeled "ICU psychosis." Despite that, the staff provided much wonderful care. A community-hospital ICU resembles what hospital care often used to be: kind nurses in an unhurried environment where they could pay close attention to patients.
Veronica spent her last 24 hours in that hospital on a regular floor. Fewer nurses were responsible for more sick patients. Veronica was in pretty good shape by then. She saw her nurse one or two times, not much more. The cardiologist and the local attending shook our hands, assured Veronica she would be fine, and sent us packing.
I was nervous but happy to bring Veronica home. Forty-eight hours earlier, she had been wired up in a cardiac ICU; now no medical provider seemed all that interested in seeing her. We made an appointment to see the cardiologist nearly one month later. We called Veronica's young university internist. I would have thought the words: "I had a heart attack" would provide some scheduling advantage -- apparently not. The medical center is de-emphasizing primary care. It's hard to make money on these services in a tertiary-care setting. During the 10 days before we saw the internist, Veronica dutifully took her medications and set about recovering from her illness and from the grueling days in the hospital. Recovery was slow. She had trouble climbing stairs, got winded a lot, and needed a lot of sleep.
Internists have taken some hits in recent years. A New York Times story in March noted that dermatologists earn twice as much and work 10 fewer hours per week. The Times quotes an aspiring dermatologist as saying that internal medicine is "viewed as easy because anyone can get into it." Since preventive medical care cases can be "humdrum," he said there is a "lack of respect for what they do."
Although that student doesn't know it, internists are the linchpin of our medical system. As described in Jerome Groopman's beautiful book How Doctors Think, physicians make sense of a disorganized jumble of data, recognize latent signs of trouble, chase down patterns when things don't look right, and help patients form a coordinated care plan. Veronica's internist started the 30-minute appointment with a jaw-dropper: "I want to hear what happened straight from you. I should say at the outset that I don't think you've had a heart attack."
Before the appointment, he had mastered Veronica's hospital record. That already put him miles ahead of most other doctors. It just didn't look right that a healthy gym rat would have a sudden heart attack with no warning and no detectable damage. He had a hunch, which he checked out with five or six senior colleagues. They agreed that a viral infection of the heart, viral myocarditis, was more likely.
He took an EKG, which revealed Veronica's resting pulse of 47. She had previously been so fit that her normal heart rate was already quite low. The beta-blocker Veronica had been prescribed was too potent, and nobody was monitoring it--making her one of many people who become sick from their medication. Mercifully, the internist tapered the beta-blocker. He also arranged for an echocardiogram in order to make a more definitive diagnosis. That echocardiogram is where this article began.
Two days after the echo, we sat in an examining room with a university cardiologist, a wonderfully effervescent, small man with a flowing gray beard and an Irish brogue. My heart initially sank when he said, "I have not read your chart. I want to hear from you." He proceeded to ask Veronica in detail about everything that had happened. Veronica tried to be efficient and precise to fit the confines of our visit. "Slow down," he said. "We have plenty of time. Did the cardiologist say your arteries look 'clean,' or 'squeaky clean'?"
After 15 or 30 minutes of questions, he said, "OK. I am going to stop the conversation now, and I am going to read your records." He methodically reviewed what had been written. "Your internist has written a Bible about you," he happily noted. He went through all the lab values and commented almost flirtatiously: "You have the kidneys of a young girl."
After more back-and-forth, he noted the competing hypotheses. He then looked over the echocardiogram results and said, "This is a classic presentation of viral myocarditis." He noted that a damaging heart attack would have shown a dead or damaged region, too weakened to support the heart's syncopated beat. I cannot imagine what cardiac patients experience when they watch live movies of their own hearts in visibly damaged condition.
My own heart skipped when he said to Veronica: "Your echo clearly shows a heart pumping poorly from the myocarditis." It wasn't just the beta-blockers that were making her winded. Her right atrium was enlarged.
As this article goes to print, Veronica is doing well but is facing a nine-month recovery. We have one loose end. Veronica's university-hospital record says that she is on aspirin and a blood thinner and that she is recovering from viral myocarditis. Yet if she falls ill tonight, an ambulance will deliver her to that community hospital, whose records indicate that she is a recovering heart-attack patient taking a potent dose of beta-blockers. Nothing in our health-care system reliably reconciles these different versions of reality. Everyone involved seems skittish to close this loop. What will we tell her original cardiologist? Will he worry that we will sue? Will he argue with us or with the other guy?
People draw their own lessons from intense experiences. Perhaps most frightening is the ease with which smart people make bad mistakes and never look back. Cognitive psychologists have documented the impact of imperfect heuristics and biases on medical decisions. It is hard to overstate the power of getting stuck in a groove, particularly when psychological crosswinds or workplace pressures distort our thinking. A wealth of data confirms this observation when we are driving a car, buying a home, or diagnosing a seriously ill patient. Such findings provide a human frame through which to view many mistakes in Veronica's care, including mine.
Our community hospital did a great job that first day. The cardiologist performed an expert angiogram. We are grateful, even knowing that they overlooked the myocarditis when Veronica's presentation cried out for this diagnosis. She had recently experienced a bad viral infection. She had no sign of artery or heart tissue damage consistent with a heart attack. Every doctor I know has said: Yup, of course, viral infection.
Medical errors seem more egregious in hindsight than they actually are. Groopman's How Doctors Think recounts many serious mistakes but also several heroic diagnoses made when doctors spot things others have missed. But many of these cases just don't seem that hard: the chronic anorexia that turns out to be celiac disease, the ER patient with chest pain who turns out to have unstable angina, the overlooked infected abscess. These examples are frightening because they reveal how skilled professionals go astray.
I can't say why Veronica's doctors missed her heart infection, but I have some clues. For one thing, Veronica's doctors never performed an echocardiogram. Such missed opportunities are common. Tejal Gandhi of Brigham and Women's Hospital and colleagues recently examined closed malpractice cases involving missed or delayed diagnoses. More than half included some failure to order an appropriate diagnostic test. This pattern may be hard to generalize. Only a tiny proportion of medical mistakes and injuries result in malpractice claims. Moreover, a missed diagnostic test is an especially provable form of malpractice.
Emergency physicians face disconcerting challenges that make them especially vulnerable to cognitive error. They must act decisively based on what is currently suspected or known. Doctors and patients both want certainty in an anxious situation. No one is reassured when the doctor says, "I'm not sure what's wrong." Yet those same doctors must remember that their provisional hypotheses might be wrong.
That openness is hard to sustain over the hours and days in which everyone's thinking becomes anchored in a specific diagnosis. The possibility of heart attack was on everyone's mind based on Veronica's dramatic cardiac-enzyme numbers. Had we gone to the hospital first rather than to the urgent-care center, the staff might have conducted a more reflective conversation with Veronica about the specific history of her illness. Given her urgent-care admission, Veronica needed an immediate angiogram before that conversation could really be had. In those first few hours, heart attack was the most reasonable working hypothesis. This deadly possibility needed immediate attention. Yet as economists and psychologists could readily predict, confirmation bias distorted subsequent judgments.
Later, things became murkier. Veronica's arteries and heart tissue looked fine. Her only symptoms were the bad enzyme results and continued chest and arm pain. These were consistent with a heart attack but also with other things. Healthy, 46-year-old women rarely have heart attacks that refuse to leave a trace. That pattern would later pique the curiosity of Veronica's internist. In the moment, hospital staff seemed stuck in a groove created by their own initial treatment plan. Their real mistake was to be incurious once the immediate crisis had passed.
Here's where the need for systemic thinking becomes apparent. When a tired doctor writes an extra zero on his prescription pad or makes a bad initial call, the result can be catastrophic, but it doesn't have to be. Hospitals can be organized to acknowledge the reality that doctors make mistakes and have messy handwriting, and that busy nurses make mistakes, too. As Jerome Groopman knows as a doctor, these mistakes are part of the landscape of medical care.
But Groopman's perspective shows its limitations. He focuses on how clinicians can avoid predictable errors and cognitive distortions. Yet as pediatric cardiologist Darshak Sanghavi notes, diagnostic errors reflect faulty systems as much as they reflect faulty thinking by any one specific person. When we consider how a decent community hospital can improve care, it may be most useful to ask not how doctors think but how systems think.
Writing in The New Yorker, another physician/journalist, Atul Gawande, has noted the value of simple checklists in matters such as controlling hospital infection. Standardization helps individual clinicians to avoid errors. It also forces hospitals and health-care systems to scrutinize their procedures and habits when elements of that checklist are frequently left undone. It's not glamorous, but this is how large organizations improve their performance. One can also create practices and protocols that reduce the likelihood and the probable consequences of common diagnostic errors. Suppose a hospital established a simple rule: Every cardiac patient who reports a recent infection should receive an echocardiogram. Such a rule or a more refined alternative would probably have saved us much time and trouble.
Some things can't be easily replicated. Our internist brought a fresh perspective, distanced from the initial emergency. Equally important, he operated in a hallway culture that encourages questioning and provides backup when things don't add up. He could ask several smart colleagues about what might have been missed. That's a key advantage of academic medicine.
Given my health-policy credentials, I'm embarrassed that I navigated this emergency relatively badly and generally felt no less bewildered than anyone else. I guess the final lessons are more personal. We must forgive ourselves, and others, for our near-misses. Then we must learn from these experiences.