Source Document: Our Study Analysis Review Spreadsheet
Narrative: Heart disease remains a major cause of death in western industrialized nations and around the world. Health care providers have traditionally found it difficult to reliably forecast which patients with vague or 'soft' symptoms of possible heart disease will actually have a heart problem. Chest pain that is not reproducible with exertion, and that is not occurring in the setting of electrocardiogram changes that are typical of active heart disease, is one such 'soft' symptom.
It is clear that a small proportion of these patients will suffer a heart attack or even a death, but reliably forecasting these outcomes without over-treating an unacceptable proportion has proved impossible with current technologies and decision aids. This review examines the highest quality data available from emergency department evaluation of patients age 40 and above who do not have classic cardiac pain, nor classic EKG abnormalities. When an emergency physician judged the patient unlikely to have a coronary syndrome and the above criteria were present, the chance of a heart attack or death was extremely small, roughly 1 in 80 (single biomarker) or 250 (two sets of biomarkers). These data are compiled from 16 studies, contributing a total of 5,792 patients. In these data there were no cardiovascular deaths that occurred within 30 days of the initial evaluation (there was one death due to motor vehicle accident and one due to other chronic disease).
*Cardiac enzymes used in all contributing studies were either conventional assays of cardiac troponin or CK-MB. Studies using only high-sensitivity troponin assays were excluded.
Caveats: These data rely on physician judgment which is generally unstructured, and it is likely that there is considerable variation in judgment patterns. The data are from multiple centers though for purposes of validity we generally chose studies of high quality with uncommonly complete data collection and 30-day follow-up success. This population may exhibit different characteristics than some others. Furthermore one American author group disproportionately contributed a large number (36 of 60) of the myocardial infarction patients (both those occurring on the second and third troponin (Hoffman et al.). The reasons for this are not clear, but because different patient groups exhibit different prevalence characteristics it may be useful to consider that with these data removed the rate of MI after an initial negative troponin is 0.66% (1 in ~150), and the rate of MI after two negative serial troponins is 0.15% (1 in ~660).
Importantly, because there is no evidence that percutaneous coronary intervention is beneficial for patients other than those having documented myocardial infarction, and only weak evidence of a long term benefit with bypass surgery, we do not count 'revascularization' as an important outcome. It seems particularly unlikely to us that these interventions could have an impact on short-term, 30-day outcomes. In general, patients who have had 'soft' chest pain, and in whom testing such as coronary angiography or stress testing demonstrates coronary narrowing (a relatively common finding even in asymptomatic persons), were treated as though the coronary narrowing was the cause of their chest pain. This is unlikely to be the case, but the conservative approach of most practitioners is to presume this. Therefore we do not consider revascularization procedures to be either of likely therapeutic benefit, nor of likely relevance to the two important outcomes that patients hope to avoid, death and heart attack.
It is also important to note that there was variation in the clinical criteria used in the contributing trials. In other words, the studies included in this analysis did not all use exactly the same criteria used in this review article. For example, many of the studies did not actually specify that the treating physician arbitrarily judge the patient as “low risk”, but instead used criteria such as TIMI or the NACP Rule, while other studies used no “low risk criteria” at all. If the reader feels that it is inappropriate to combine the studies in this way, we invite anyone to use our open source complete data set which explicitly lists each contributing trial and its particular low risk criterion, so that one may easily determine for one’s self the percentages for the criteria that apply best to a given patient population.
Author: Pendell Meyers and David Newman, MD
Published/Updated: September 21, 2010