Assessment tools for unrecognized myocardial infarction: a cross-sectional analysis of the REasons for geographic and racial differences in stroke population
1 Department of Epidemiology, University of Alabama at Birmingham, Birmingham, AL, USA
2 Division of Preventive Medicine, University of Alabama at Birmingham, Birmingham, AL, USA
3 Department of Healthcare Organization and Policy, University of Alabama at Birmingham, Birmingham, AL, USA
4 Epidemiological Cardiology Research Center, Wake Forest University School of Medicine, Winston Salem, NC, USA
5 Department of Biostatistics, University of Alabama at Birmingham, Birmingham, AL, USA
BMC Cardiovascular Disorders 2013, 13:23 doi:10.1186/1471-2261-13-23Published: 26 March 2013
Routine electrocardiograms (ECGs) are not recommended for asymptomatic patients because the potential harms are thought to outweigh any benefits. Assessment tools to identify high risk individuals may improve the harm versus benefit profile of screening ECGs. In particular, people with unrecognized myocardial infarction (UMI) have elevated risk for cardiovascular events and death.
Using logistic regression, we developed a basic assessment tool among 16,653 participants in the REasons for Geographic and Racial Differences in Stroke (REGARDS) study using demographics, self-reported medical history, blood pressure, and body mass index and an expanded assessment tool using information on 51 potential variables. UMI was defined as electrocardiogram evidence of myocardial infarction without a self-reported history (n = 740).
The basic assessment tool had a c-statistic of 0.638 (95% confidence interval 0.617 - 0.659) and included age, race, smoking status, body mass index, systolic blood pressure, and self-reported history of transient ischemic attack, deep vein thrombosis, falls, diabetes, and hypertension. A predicted probability of UMI > 3% provided a sensitivity of 80% and a specificity of 30%. The expanded assessment tool had a c-statistic of 0.654 (95% confidence interval 0.634-0.674). Because of the poor performance of these assessment tools, external validation was not pursued.
Despite examining a large number of potential correlates of UMI, the assessment tools did not provide a high level of discrimination. These data suggest defining groups with high prevalence of UMI for targeted screening will be difficult.