Share this article on:

Emergentology: Risk Stratification and the Unsweet Spot

Walker, Graham MD

doi: 10.1097/01.EEM.0000405492.01170.cb


When I was applying to residency four years ago, I wrote in my personal statement that I was drawn to emergency medicine for its dichotomies: sick/well, admit/discharge, CT/observe. But now that I've graduated from residency, I realize that emergency medicine isn't a dichotomy; it's a trichotomy. And it's that somewhere-in-the-middle patient, that patient in the gray zone — the unsweet spot — that I find the most challenging part of this job. I never realized that it would be risk stratification keeping me up at night.

Everyone knows what to do when the chart says “chest pain, history of three stents, ran out of Plavix.” Or what care a febrile, hypotensive 80-year-old needs. And we know how to discharge the 6-year-old with fever who's terrorizing everyone from the moment he steps foot in the department. But it's everyone else's stories that make me anxious. There frequently is not a right answer to the patient with vague pains in the chest and a couple of risk factors, or the cellulitis that probably will start to recede, as long as the patient fills his prescription. (In the medicolegal world, the “right” answer is to admit him. In the ideal world, however, what's best for the patient is frequently to discharge him with close follow-up. But when we're held to a standard of not missing anything, ruling out everything, and a magical retrospectoscope, the 80-year-old with reproducible chest wall pain can certainly be in that five percent of acute coronary syndromes with tenderness to palpation.)

I wish patients read the textbooks, mostly because I studied them so well. I can rattle off algorithms from the United States Medical Licensing Examination study guides and mnemonics from board review books, but nowhere in them do they mention that an extremely tender epigastrum that you're working up for GI stuff can be a STEMI, or that the nonfluctuant abscess can be hiding gallons of pus beneath it. Patients will continue to surprise us until the end of time because their physical complaints are subject to their brain's interpretation of them, with a layer of cultural and language barriers that can make things even fuzzier.

We're always told to maintain a high index of suspicion, but we're all huddled on that fine line between ridiculous workups and missing something subtle. In the gray zone, there's wide practice variation and possibly no right answer (but a bunch of definite wrong ones). We can't always accurately assess these patients' risk, and when we do, we frequently disagree about it.

So what's an emergency physician to do? We could test them all for everything under the sun. But doing CT angiograms on all patients with shortness of breath so we never miss anything in the chest will only irradiate our community and find more false positives, contrast extravasations, and anaphylaxes than pulmonary emboli.

Or we can admit and observe them. But no other physician besides us apparently feels comfortable dealing with undifferentiated patients, and there are certainly risks to being admitted to a hospital when you don't need to be there.

Then there are scores and risk factors and algorithms (oh, my!), and they have certainly helped us to quantify risk, and given us a number to go along with our clinical judgment. But if a patient doesn't fit one of the equations or criteria — or one doesn't yet exist for the patient and the potential disease on the stretcher in front of you — we're all out of luck. We need more and better of these decision aids. Yesterday.

I'd like to suggest two other options. The first is cheap and easy, but forces you to actually interact with the patient in front of you, to talk to them about this stuff. It's a radical notion frequently espoused by my radical colleague, David Newman, MD. He will literally — get ready for this — discuss with patients what he thinks their risk is of “badness” based on his knowledge of the disease, the scientific literature, and the patient. And then — with careful documentation, of course — he'll have the patient try to make a decision with his guidance: “If we keep you, you'll probably stay in the hospital for a few days, and they'll do a stress test on you. If you go home, we'd have you follow up tomorrow with your doctor, and come back here if you get worse. We can't say 100% this isn't a heart attack, but your EKG looks good, your blood tests are normal, and your story isn't very typical for one. Less than five percent is good odds for some people, but for others, 95 percent sure isn't enough. Which are you?”

This technique isn't for every patient obviously, just like a CT angiogram isn't for everyone with pleuritic chest pain. You pick the patient that has the faculty to understand these choices and who can come up with an educated decision about his risk, preferably someone who is reliable and can and will follow up.

Another option is becoming more common, and maybe it is the right answer for the patient that you're leaning toward admitting but need some tincture of time: the observation unit. Patients are coming to the emergency department so frequently and quickly that their diseases haven't come to a head just yet, and you need to see which way they're going. Sit on the asthmatic who's kind of, sort of starting to turn around and who definitely won't stay for admission; rule out infarction in the low-risk-but-could-be-ACS guy; make sure the cellulitis moving up the patient's leg starts to retreat. These observation units are mostly being staffed and run by emergency physicians, which is the right move: no one else really sees these unsweet spot patients (and if they do, they're sent from the clinic to the ED with that lovely, detailed referral note: “For evaluation.”)

Who knows, maybe in 20 years, we'll have the data and the computing power to come up with clinical decision rules we can use at the bedside to rule out “badness,” as Dr. Newman puts it, but I don't think so. One person's dizziness is another person's weakness is another person's “it hurts everywhere.” With increasing numbers of ED visits and an emphasis on quality and cost control, we've got to come up with ways to figure out where and how we should approach these gray-zone patients, and we've got to agree that there's not always a right answer and that a bad outcome isn't necessarily because of a bad decision. Perhaps simply being reasonable and discussing risk and options with the patient is the solution for some; perhaps simply letting the patient declare themselves over time is another.

The fact is, in emergency medicine, we don't spend most of our days saving lives, as most people think. Mostly we try to predict risk, especially in these gray-zone patients. It's why we admit some and discharge others. But I'm not going to start correcting anyone. “I'm an emergency physician, I do risk stratification for a living” just doesn't sound as sexy.



Dr. Walker is a fellow in simulation medicine in the Stanford/Kaiser Emergency Medicine Residency Program. He has been blogging since medical school, first at Over My Med Body, and now as the developer and co– creator of MDCalc, a medical calculator for clinical scores, equations, and risk stratifications (, and The NNT, a number-needed-to-treat tool to communicate benefit and harm (

Back to Top | Article Outline

Only Online

Read all of Dr. Walker's past columns in the EM-News.comarchive.

© 2011 Lippincott Williams & Wilkins, Inc.