3. How do you risk stratify patients with chest pain? Do you use any clinical decision rules?
It’s a daily ED conundrum: how concerning is this patient’s chest pain? This most common, yet challenging, clinical quandary is not an insignificant one as there is a high cost (in lives and dollars) associated with missed diagnoses and over-admission of patients found not to have ACS. Despite the efforts of many, there is still no (validated) answer as to how to identify the patient who is safe to discharge or who is sure to have ischemic disease. Sadly, we will not be providing you with one here, either. But, we’d like to leave you with some pearls to use when risk stratifying ED chest pain patients.
Clinical impression: Several studies have looked at how good a clinician’s gestalt is in risk stratifying patients. Outcomes of these suggest that while our clinical estimation of a patient’s risk does have some prognostic value in that it correlates with adverse outcomes in patients, our intuition together with an ECG are not sufficient to identify patients safe for discharge (Chandra, 2009). Even when we think we have a “clear cut” non-cardiac explanation for chest pain, it turns out we may not be right: although these patients have a reduced risk of adverse events, they have been found nonetheless to have a 4% thirty day event rate (death, MI, revascularization) (Hollander, 2007). These findings are made all the more concerning by the results of a recent study that found the presence of “atypical” chest pain to be an independent predictor of in-hospital mortality (El-Menyar, 2011).
Bottom line: Your clinical impression is important, but not sufficient.
Risk factors: Ah, risk factors. We all find comfort in asking questions about hypertension, hyperlipidemia, smoking, diabetes, and family history. The problem is that these “risk factors” are risk factors for developing coronary artery disease–not for acute cardiac ischemia. Established through the Framingham study, these risk factors have been found to be poor predictors of AMI or ACS, and are considered much less useful than symptoms, EKG, or cardiac biomarkers (AHA/ACC NSTEMI Guidelines 2007). One study found that risk factors had no predictive value in women, and that the only risk factors that increased risk of acute ischemia in men (with a very low relative risk) were diabetes and family history (Jayes, 1992). Another, more recent study that analyzed over ten thousand patients using the Internet Tracking Registry of Acute Coronary Syndromes (i*trACS) registry, found that while the absence of risk factors in patients under 40 has a strong negative predictive value, the presence (or absence) of them in patients over 40 has limited clinical significance (Han, 2007).
Bottom line: Risk factors are for coronary artery disease, not for acute ischemia. They have limited utility in assessing the presence (or absence) of ACS.
History & Clinical features: Among the gamut of questions we ask patients about the quality, location, severity, and alleviating/exacerbating factors of their pain, few have been shown to be independent positive or negative predictors of ACS. Several studies and reviews have evaluated various symptoms and chest pain descriptors. The one, common, positive predictor of ACS/AMI is radiation of pain to the shoulders, especially if it is to both shoulders (LRs ranging from 4.1-7.1) (Panju, 1988; Swap 2005; Goodacre 2002). Other potential predictors are not as clear. Nausea, vomiting and diaphoresis has been found predictive by some (Panju, 1988), while others found exertional pain to be predictive (Goodacre, 2002). With respect to characteristics that may help rule out ACS, Panju, et al., found that pleuritic, positional, reproducible (with palpation) or sharp/stabbing chest pain has a negative predictive value. However, Goodacre’s study found only a non-significant trend towards an unlikely ACS diagnosis for pleuritic pain, and found no association at all for the other characteristics. Severity, location, character (e.g. pressure, ache, burning) were all found to be poor predictors.
Bottom line: Chest pain radiating to the shoulders is concerning for AMI/ACS. Other findings are not independently reliable; however, they may be more useful in aggregate (See clinical decision rules below).
Response to treatment: While we might hope that response to nitroglycerin might suggest angina, or that improvement of symptoms with a GI cocktail might reassure us that the pain is indeed related to GERD, studies have suggested otherwise. It seems that nitroglycerin can also relieve esophageal causes of chest pain and has been found to equally improve pain in patients with and without ACS (Henrikson, 2003). Similarly, studies and case reports suggest caution in using a GI cocktails to differentiate ischemia from gastrointestinal pain (Wrenn, 1995, Servi, 1985).
Bottom line: Exercise caution in attributing diagnostic weight to the relief (or absence of relief) of symptoms by nitroglycerin or GI cocktails.
Clinical decision rules: Because of all the disheartening studies that have failed to identify the single Golden Key to separating the sick hearts from the well ones, multiple clinical decision rules have tried to amalgamate predictive factors to increase the ability to do just that. Two of the most commonly used and compared are the TIMI Risk Score (for UA/NSTEMI and STEMI) and the GRACE score. Both have been validated as risk stratification tools for patients presenting to the ED, and hence can help guide treatment, according to the ACC/AHA guidelines (Lyon, 2007). However, even these tools do not help determine whether a patient is safe to be discharged. One clinical decision rule that may help with this is the Vancouver Chest Pain rule. This rule was found to have 98.8% sensitivity for AMI/ACS, and defined a group of patients who were safe for discharge after a few hours in the ED, but has not yet been validated (Christenson 2006). Unfortunately, this rule, plus seven others, were systematically reviewed by Hess et al., who found none of the rules yet acceptable for clinical practice (Hess, 2008).
Bottom line: Clinical prediction rules can be used to help risk stratify, but are not helpful in identifying patients safe for discharge.
Is there anyone I can discharge? The ACC/AHA guidelines allow for discharge of patients established to have non-cardiac causes of chest pain, but require at least a “rule-out” work-up (initial and repeat cardiac enzymes, ECGs and preferably an in-house stress test) for any patient for whom ischemic disease is considered a possibility, no matter how low the risk. Of course, establishing whether a patient’s pain is ischemic or not in origin is the crux of the problem we face daily. TIMI and GRACE scores may help. Others have used data from a few studies including that by Han (2007) and Marsan (2005) to support discharging patients under 40 years of age with a normal ECG, no established CAD, and none of the traditional Framingham risk factors, as they had only a 0.2% thirty-day rate of death or MI (Newman, 2011). But, again, this approach has not been validated.
In summary, establishing whether someone’s chest pain is ischemic in origin can be difficult, and one should be wary of placing too much emphasis on individual clinical findings in making that determination. Ultimately, if ischemia is at all a concern, the patient should be risk stratified (TIMI and GRACE scores together with history and clinical findings help with this) and treated according to the ACC/AHA guidelines based on that risk stratification, which at the very least may include admission to an observation or chest pain unit. There may be some patient groups who are safe for discharge, but rules defining such groups have yet to be validated. Here’s to hoping for a validated rule soon.