Showing heart patients the way home

August 2007
Cover Story

Karen Titus

Emergency department physicians are sorry hosts. No sooner does a patient arrive than they start calculating, "How soon can I send this person home?"

Some patients, like some guests, are particularly unobliging. Those would be the cardiac cases—expensive, confusing, and tough to show to the door.

With guests, a simple yawn might do the trick. With cardiac patients, the markers are not as simple, as any laboratorian will attest. There’s no shortage of markers, but figuring out what, exactly, they mean, and how, exactly, to use them is an endless safari, especially in emergency departments, the mosh pits of medicine.

ED specialists weighed in with their views at an American Association for Clinical Chemistry cardiac conference in Chicago in May, poring over data and sharing their hard-won knowledge about what works and what might work.

When a patient with suspected heart failure arrives in the ED, it’s as if a giant jigsaw puzzle has been dumped open. It needs to be put together quickly—but how? ECGs are nondiagnostic 98 percent of the time, says W. Frank Peacock, MD, vice chief, emergency medicine, Cleveland Clinic. Physical exams are diagnostic in only one-third of cases. The numbers can run even lower for chest x-rays, which miss 20 percent of echo-proven cardiomegaly. If patients are x-rayed in supine position—typical for sicker patients—one-third of pleural effusions will be missed. “And if you do a portable, which is how you really take x-rays if people are sick, their numbers are even worse.” History—the best tool, says Dr. Peacock—has its own snags. Patients may be too sick to give a history, and those who aren’t may not speak English, may be drunk, may be poor historians, or may simply show up at the ED looking for a warm place to sleep, Dr. Peacock says wryly, evidently speaking from experience.

Yet the ED remains the major source of heart failure admissions, and it doesn’t come cheaply. ADHERE (Acute Decompensated Heart Failure National Registry) data show that from the ED, 66 percent of such patients end up in a telemetry bed, with 13 percent in the ICU/CCU—among the most expensive stops in the hospital. Citing another study, Dr. Peacock notes the economic impact of missed heart failure at admissions is considerable. “It costs you a grand to be wrong—and that’s not even dealing with mortality and other problems,” he says. Missed diagnoses also mean lost reimbursement and 0.9 days longer in the hospital.

The clock ticks loudly in the ED. Dr. Peacock, not one to mince words, says it’s not a hard concept: “Treat early or die.” It’s true for hypoglycemia, hypoxia, ventricular tachycardia, myocardial infarction, cerebral infarction, and pneumonia, and appears to be true for heart failure as well.

A 46,599-patient study done by Dr. Peacock and his colleagues (Ann Emerg Med. 2003;42:S26) found delayed treatment meant higher mortality rates (253 percent higher), more ICU transfers (500 percent), longer lengths of stay in the hospital (150 percent) and the ICU (155 percent), and more invasive procedures (142 percent).

Since quick is good, why not treat every possible case as if it were indeed heart failure? Because “getting it wrong kills people,” Dr. Peacock says. Citing one example of how quick can turn to quack, he notes that appropriate treatment of patients with chronic obstructive pulmonary disease carries a 3.8 percent mortality rate. If they’re mistakenly treated for heart failure, the mortality rate jumps to 13.6 percent.

All of which leaves ED and laboratory personnel at the mercy of markers.

Dr. Peacock is a fan of the heart hormone BNP. The Breathing Not Properly trial (Maisel AS, et al. N Engl J Med. 2002;347:161–167) asked two blinded cardiologists to assess 1,586 ED patients with dyspnea and suspected heart failure, estimating their probability of what they thought the diagnosis was. Relying on BNP as well as clinical judgment, they were right four out of five times. “That’s actually pretty good,” says Dr. Peacock. “It doesn’t sound great, but it’s as good as the state of the art is.” Without BNP, the number was three in four.

Just as important to Dr. Peacock is how BNP affects clinician confidence. When not using BNP, clinicians expressed uncertainty with their diagnoses in 43 percent of the patients; with BNP, indecision dropped to 11 percent. “That’s actionable. I can do something and be pretty certain that I’m doing the right thing.”

Using the ADHERE registry, Dr. Peacock and colleagues (article in press, J Am Coll Cardiol) looked at initial BNPs in hospitalized patients. If the level was under 500 pg/mL, death rate was two percent; if the level was above approximately 1,700 pg/mL, death rate was six percent. That threefold increase in death means more decisions for ER physicians. The patient won’t go home—but where will he go next? The ICU? “It’s the most expensive resource you’ve got,” Dr. Peacock points out. A telebed? Some other destination? For an ED physician like Dr. Peacock, BNP results mean, “I am the card dealer at the table, and I have to make those decisions.”

BNP is a good card, and makes a nice hand alongside other risk stratifiers. If BUN is low, blood pressure is high, troponin is negative, and BNP is under 500 pg/mL, “that person is going to do very well.” If BUN and creatinine are high, blood pressure is low, and BNP is very high (greater than 1,700 pg/mL), that patient is headed straight for the ICU. Moreover, according to the BASEL Study Algorithm, a BNP test up front, in the ER, reduces ICU admissions, length of hospital stay, and dollars spent (by approximately $1,800).

In what Dr. Peacock says is the first such study, he and his colleagues also used ADHERE data from about 15,000 patients to show an important link between BNP level and diuretic time. In the two lower cohorts (BNPs of less than 449 pg/mL or between 450–864 pg/mL), delay in receiving diuretics did not significantly affect mortality. Patients with higher BNPs, especially with those above 1,738 pg/mL, had higher mortality as time to diuretics increased. Delayed BNP, he says bluntly, delays diagnosis and treatment and increases patient death rate.

How valuable is troponin in the ED? Dr. Peacock says it’s popular to believe troponins mean nothing in heart failure. “I’ve been told this many times. I’ve had very famous cardiologists tell me that.” Dr. Peacock discards the idea, ranking it with a host of other myths, among them: “You can’t drown a witch,” “Night air causes pneumonia,” and, he adds with a laugh, “Stop that or you’ll go blind.”

Turning again to the ADHERE registry, Dr. Peacock reports on an analysis of 70,000 patients that showed a 6.3 percent positive troponin rate, which, in turn, translated to a 300 percent increased risk of CABG, IABCP, or intubation; a one-day increase in hospital stay; an ICU length of stay increase of one-half day; and 250 percent increased risk of mortality. “The idea that troponins don’t mean anything in heart failure is just bogus. Troponins have a tremendous outcome impact on what I’m going to do, and who I’m going to do it to.”

Other fallacies include the idea that troponins don’t predict outcomes if creatinine is elevated, and that troponin is always elevated in cases of renal insufficiency.

Dr. Peacock referred to a study he and a colleague did nearly a decade ago (J Am Coll Cardiol. 1999;33:471– 478) that he says has been misquoted repeatedly. The two looked at 51 patients with creatinine levels greater than 2 mg/dL, versus 102 matched controls. Troponin sensitivity for adverse cardiovascular events was set at 87 percent. The patients with higher creatinines had lower rates of adverse events than those with lower creatinines, acutely and at six months.

The common interpretation, which Dr. Peacock terms technically correct, is that prediction of adverse cardio events and acute coronary syndrome is reduced. “But if we look at the actual numbers, 35 percent of the patients in the renal group had an adverse event, acutely, and 45 percent in the non-renal group.

“There’s no difference as far as I’m concerned,” he says. “If I send this person home, that’s a one in three chance of badness.” At six months, there was no difference in outcomes.

In another ADHERE-derived assessment involving 16,000 patients with creatinines greater than 2 mg/dL, those with positive troponins had twice the rate of mechanical ventilation, longer hospital length of stay, and double the mortality of those with negative Tn. “So this idea that if the creatinine is up, ignore the troponins—yeah go ahead,” he says sarcastically.

IMA—ischemia modified albumin—is a solid means for appraising chest pain. Negative predictive value is critical in the environment in which you want to send patients home, Dr. Peacock notes. “If you don’t get this as an emergency doctor, you’re going to see Judge Judy,” he says. “If you still don’t get it after that, you’re going to be unemployed and looking for a job at McDonald’s. This is our No. 1 lawsuit. When we’re wrong and you go home and die, I’m going to pay a million dollars.”

The routine is a familiar one: chest pain, which can have any number of possible explanations—Tietze’s disease? Mondor’s syndrome? Anxiety? Breast abscess? Acute coronary syndrome? —assessed by risk factors, history, physical exam, ECG, and markers. Ideally, a marker will rise early and, like Gustav Mahler’s forehead, manifestly. It also needs to do better than the current clinical practices of emergency physicians, who have been shown to be 94.7 percent sensitive and 74 percent specific in identifying patients who were subsequently diagnosed with acute coronary syndrome within 30 days of an ED visit (Christenson J, et al. Can Med Assoc J. 2004;170:1803–1807).

IMA, an altered form of albumin, no longer binds to cobalt when it undergoes ischemic events. It is a relatively straightforward test since it rises within minutes—unlike CK-MB, myoglobin, or troponins—and remains elevated for six to 12 hours. However, if symptoms are too old, false-negatives can occur. It’s also elevated in patients with some cancers, sepsis, liver disease, end-stage renal disease, and brain ischemia. It decreases with endogenous lactate production. And demand ischemia does not cause it to rise. “So if you go out and run a marathon, it’s not going up.” When Dr. Peacock and a colleague did a meta-analysis of IMA in the context of a triple prediction test (alongside ECG and troponin), the negative predictive value was in the ballpark of the current gold standard for acute diagnosis of ACS (Am Heart J. 2006;152: 253–262).

It’s promising, says Dr. Peacock, but will require more study. (He says the 1,200-patient IMAGINE [Ischemia Modified Albumin to Diagnose New Events] study may provide some answers in the next year or so.) Further scrutiny will also be needed with two potential markers of plaque rupture: MMPs (matrix metalloproteinases) and MPO (myeloperoxidase). Both are secreted by macrophages that, along with neutrophils, infiltrate coronary artery plaque, and both degrade the collagen layer that prevents atheroma erosion or rupture.

MPO is stored in granules of polymorphonuclear neutrophil leukocytes and macrophages and released into circulation during inflammatory conditions. Since neutrophil activation is not induced by ischemia, that makes MPO a marker of plaque instability, not myocyte damage, says Dr. Peacock. It’s not specific to cardiac diseases, since it’s present in any infectious, inflammatory, or infiltrative process. More likely, it will serve as a marker of inflammation. “The question really becomes, how sensitive can this be? Am I going to be able to send people home?” he asks.

A study looking at 100 patients who were partially or completely deficient in MPO, compared with 118 controls, found the MPO-deficient patients had more severe infections and more chronic inflammatory processes. Does that mean MPO is a cause, and not merely a marker, of coronary disease? Could it be both? There are data suggesting that MPO in the subendothelium of atheromata may participate in ACS by promoting superficial erosion and increasing thrombogenicity (Sugiyama S, et al. Arterioscler Thromb Vasc Biol. 2004; 24:1309–1314), as well as a host of other studies indicating its link to heart disease.

Dr. Peacock has seen some promise using MPO, citing a study that looked at plasma MPO levels in 604 patients presenting to the ED with chest pain and followed up at 30 days and six months (Brennan M, et al. N Engl J Med. 2003;349:1595–1604). In a nutshell, the higher the MPO, the worse the risk of major cardiac events. It’s predictive even when troponin is negative. “Having a high MPO is not a good thing,” says Dr. Peacock, summing up several other studies.

Perhaps the most basic form of assessment—cardiac risk factors such as age, LDL/HDL, blood pressure, smoking—are noticeably absent when ED physicians talk shop.

“We hear about cardiac risk factors since Day One of medical school,” says Judd E. Hollander, MD, professor and clinical research director, Department of Emergency Medicine, University of Pennsylvania, Philadelphia. The reason is simple: They predict long-term risk. Anyone who shows up in the ED with chest pain has symptoms now.

When emergency physicians have to risk stratify for myocardial ischemia, they need tools with negative predictive values. Missed AMI rate is inversely proportional to the admission rate for ED chest pain patients, Dr. Hollander notes, referring to several landmark clinical trials.

The goal is to miss less than one percent, though Dr. Hollander concedes this is, for all intents and purposes, a mirage—no one has found a way to get below the current, higher miss rates. “You’re going to miss a decent number, even in patients with the most classically obvious symptoms of musculoskeletal pain,” he says. Partial pleuritic pain carries a six percent miss rate, for example; positional pain has a four to nine percent miss rate. Pain reproduced by chest-wall palpation isn’t much better. “This is the most difficult thing we deal with in the ER,” Dr. Hollander says. “I can’t tell you the number of times a resident walks out of the room and says, ‘I can reproduce their chest pain.’ Still, five or six percent of those patients have MI.”

“So what we really want to do is find the ‘real patients,’ and send home the patients without serious disease,” Dr. Hollander says.

From the ED trenches, the choices are distressingly limited.

ECG? “Doesn’t help,” he says. The negative predictive value remains steady no matter how long since symptom onset: 93 percent at zero to three hours, three to six hours, and six to nine hours; at nine to 12 hours, it increases to 94 percent (Singer AJ, et al. Ann Emerg Med. 1997;29:575–579). “That doesn’t let us send people home.”

ED physicians aren’t the only ones frustrated by mystery chest pain. Dr. Hollander points to one study with interesting origins (Lee TH, et al. N Engl J Med. 1991;324:1239–1246): “They did this study because they [cardiologists] got sick of the ER admitting bogus chest patients,” Dr. Hollander says. Yet the cardiologists themselves were unable to identify patients who could be released safely. As Dr. Hollander puts it, “That’s from the group who’s tired of these patients being admitted, when they don’t have real disease.”

At Penn, where Dr. Hollander practices, the ED department chair, William Baxt, MD, developed a neural networks approach, which Dr. Hollander terms a great black box. The algorithm contains nearly two dozen input variables under four categories—history, PMH, exam, and ECG—including intensity and radiation of pain, AMI, angina, Rales, 2- and 1-mm ST elevation, and ST depression. Based on initial studies, the algorithm appeared worth trying in real time. So Dr. Hollander and his colleagues did just that, in what he said was a very complicated protocol involving some 500 patients. It worked well, according to Dr. Hollander, with emergency physicians receiving a card listing each patient’s risk at the time of initial evaluation, as well as a report of what ultimately happened to the patient at followup. “The database generated an e-mail that basically said, ‘Your patient who we told you didn’t have an MI in fact didn’t have an MI, and here’s the test results you had, and here’s all the additional time you wasted by not listening to us.’ Turns out that doctors still don’t listen to it,” he says. “So even though it works well, I don’t think it has as lot of long-term potential.”

That means turning to cardiac markers. “A single marker doesn’t cut it,” says Dr. Hollander. “Single markers don’t get you home. I’m guessing they’ll never get you home. You can come up with any marker you want to—it ain’t getting you home.” He’s betting on multimarker strategies, which might combine a marker of necrosis with a marker of volume overload, and possibly markers of inflammation and platelet aggregation.

The latter have been studied extensively, says Dr. Hollander, though it’s yet to be shown platelet aggregation markers work in an ED-based study in a broad population. In studies comparing patients with MI to normal volunteers—and there have been plenty such studies, says Dr. Hollander—the markers “work great. But we don’t actually see normal volunteers in the emergency department. We see people with diabetes and hypertension and stroke and other problems, who are likely to have some platelet aggregation at baseline.”

Beyond markers, Dr. Hollander is thinking long and hard about imaging modalities.

Coronary CT angiography, specifically. “I’ve been searching for a strategy that can send you home for 15 years. I think coronary CT angiography is finally the holy grail where I can send patients home.” In fact, he suggests, its real strength may emerge when it’s coupled with markers.

Stress tests, as it turns out, are of limited value. Patients with negative stress tests are just as likely as patients who don’t receive the tests to return to the emergency department (39 versus 40 percent), be admitted to the hospital (29 versus 32 percent), and receive cardiac catheterization (12.5 versus 10.4 percent), says Dr. Hollander, citing a study published in Academic Emergency Medicine in 2004. “In the ED, a stress test is a permission slip to go home. That’s it. Emergency physicians never use it again.”

Coronary angiography appears to be a different story. Low-risk patients evaluated by CA, compared to stress tests, have fewer repeat ED visits and fewer hospitalizations.

Coronary CT angiography may be a different story, too. One randomized control trial comparing CCTA versus myocardial perfusion imaging has shown CCTA shortens length of stay (3.4 versus 15 hours) and saves money ($1,586 versus $1,872, despite the expense of the test) (Goldstein JA, et al. J Am Coll Cardiol. 2007;49:863– 871). Other studies have been similarly promising, with one showing 100 percent negative predictive value at five-month followup. In another, 100 cardiac referral patients with normal coronaries on CCTA had 100 percent survival one year later.

Dr. Hollander leaves no doubt he’s a fan of CCTA. In his own recent study, published in Academic Emergency Medicine (2007;14: 112–116), patients were given immediate CCTA in the ED (following a creatinine test), with disposition based on results. “We’re able to send 85 percent home right away,” he says.

In the emergency department, that makes him a perfectly good host.