Novel Protein Markers of Acute Coronary Syndrome Complications in Low-Risk Outpatients: A Systematic Review of Potential Use in the Emergency Department Alice M. Mitchell, 1 Michael D. Brown, 2 Ian B.A. Menown, 3 and Jeffery A. Kline 1* Background: Published literature was systematically reviewed to determine the diagnostic accuracy of new protein markers of acute coronary syndromes (ACS) in symptomatic outpatients at low risk of ACS and related complications, compared with patients evaluated in emergency department chest pain units. Methods: Studies were identified by a MEDLINE ® (1966 to May week 3, 2005) search. Abstracts were reviewed for relevance, and manuscripts were included by the independent consensus of 2 observers based on explicit criteria restricting the analysis to studies rele- vant to screening ambulatory patients with symptoms suggesting ACS. Publication bias was identified by a modified funnel plot analysis [study size (y) vs the inverse of the negative likelihood ratio (x)]. Results of individual markers were reported separately. When 3 or more eligible studies were identified, data were aggre- gated by use of the summary ROC (SROC) curve. Results: Twenty-two protein markers in 10 unique pop- ulations met the inclusion criteria. Data required for SROC analysis (true- and false-positive rates) were available for 17 markers, in 9 unique populations, from publications and personal communications. Of these, only C-reactive protein was published in more than 2 populations to allow aggregation (6 studies total). C- Reactive protein demonstrated poor diagnostic perfor- mance on SROC curve analysis, with an area under the curve of 0.61 and a pooled diagnostic odds ratio of 1.81 (95% confidence interval, 1.06 –3.07). Conclusion: Published evidence is not sufficient to support the routine use of new protein markers in screening for ACS in the emergency department setting. © 2005 American Association for Clinical Chemistry An accurate, noninvasive, and efficient method of detect- ing substantial coronary artery disease in otherwise ap- parently low-risk patients remains an unmet need. In the emergency department setting, the subpopulation of pa- tients who represent the greatest challenge in decision making are these low-risk patients. These patients gener- ally present with nonspecific chest discomfort, a popula- tion risk factor, and are at low risk of an acute coronary syndrome (ACS) 4 and related complications. However, reports by Goldman et al. (1) and Lee et al. (2) indicated that 2% to 5% of patients with myocardial infarctions are initially missed and are discharged from the emergency department. On the other hand, 5% of patients in this emergency department population category of “atypical chest pain” are ultimately diagnosed with ACS (3).A developing standard in care is to evaluate low-risk pa- tients with possible ACS by use of emergency department chest pain units that incorporate serial markers of myo- cardial necrosis (e.g., serum troponin, creatine kinase, and myoglobin concentrations), telemetry, and a provocative test. This strategy may increase the sensitivity for detect- ing incipient ACS, but at the cost of an overnight stay, possible false/positive testing, and the requirement of other resources. Even with a well-conceived chest-pain unit protocol approach, as many as 4% to 14% of patients 1 Department of Emergency Medicine, Carolinas Medical Center, Char- lotte, NC. 2 GRMERC/Michigan State University Program in Emergency Medicine, Grand Rapids, MI. 3 Craigavon Cardiac Centre, Craigavon Area Hospital, Belfast, Northern Ireland, United Kingdom. * Address correspondence to this author at: Emergency Medicine Re- search, Department of Emergency Medicine, Carolinas Medical Center, PO Box 32861, Charlotte, NC 28323-2861. Fax 704-355-7047; e-mail jeff.kline@ carolinashealthcare.org. Received April 12, 2005; accepted August 5, 2005. Previously published online at DOI: 10.1373/clinchem.2005.052456 4 Nonstandard abbreviations: ACS, acute coronary syndrome; LR, negative likelihood ratio; CI, confidence interval; SROC, summary ROC; TPR, true- positive rate; FPR, false-positive rate; and CI, confidence interval. Clinical Chemistry 51:11 000 – 000 (2005) Review 1 Papers in Press. First published September 15, 2005 as doi:10.1373/clinchem.2005.052456 Copyright © 2005 by The American Association for Clinical Chemistry