To determine the relative importance of clinically recognized cardiac dysfunction and unrecognized cardiac injury to hospital mortality. Prospective, blinded, single-center study. Medical ICU of Barnes-Jewish Hospital, St. Louis, a university-affiliated teaching hospital. Two hundred sixty adult patients requiring admission to the medical ICU. Daily blood collection. The presence of cardiac dysfunction (myocardial infarction, unstable angina, cardiac arrest, or congestive heart failure) as determined by the physicians responsible for the care of the patient. Daily measurement of levels of cardiac troponin I, a sensitive, highly specific, and long-lived marker of myocardial injury. Fifty-five (21.2%) patients had clinical evidence of cardiac dysfunction, among whom 22 (40%) had an elevated level of cardiac troponin I. A total of 41 (15.8%) patients had evidence of acute myocardial injury based on elevated levels of cardiac troponin I. Patients with clinically recognized cardiac dysfunction had a significantly greater hospital mortality rate compared to patients without clinically recognized cardiac dysfunction (45.5% vs 10.2%; p < 0.001). Similarly, patients with elevated blood levels of cardiac troponin I had a greater hospital mortality rate compared to patients without elevated blood levels of cardiac troponin I (26.8% vs 16.0%; p = 0.095). Multiple logistic-regression analysis controlling for potential confounding variables demonstrated that the presence of clinically recognized cardiac dysfunction was independently associated with hospital mortality (adjusted odds ratio = 3.0; 95% confidence interval = 1.9 to 4.8; p = 0.016). However, having an elevated blood level of cardiac troponin I was not found to be an independent determinant of hospital mortality. Among critically ill medical patients, clinically recognized cardiac dysfunction is an independent determinant of hospital mortality. The identification of unrecognized cardiac injury, using serial measurements of cardiac troponin I, did not independently contribute to the prediction of hospital mortality.
Read full abstract