10), as were the characteristics of those patients who did and did not undergo CMR
(Supplemental Table 8). No patient had evidence of a myocardial infarction in two arterial
territories or myocardial scarring in a non-ischemic pattern. Left ventricular dimensions and
ejection fraction were similar between the two groups. Median infarct size was increased in the
oxygen group compared to the no oxygen group, (20.3 grams [interquartile range, 9.6 to 29.6] vs.
13.1 grams [interquartile range, 5.2 to 23.6]; P=0.04. When expressed as a proportion of left
ventricular mass, the difference in median infarct size was 12.6% (interquartile range, 6.7 to
19.2) in the oxygen group compared with 9.0% (interquartile range, 4.1 to 16.3) in the no oxygen
group (P=0.08), with the ratio of geometric means approaching significance: 1.38 (95% CI, 0.99
to 1.92; P=0.06). Troponin and CK measurements taken at the index admission were
significantly correlated with infarct size at six months (Supplemental Table 3).