Early observations of S-myoglobin in the diagnosis of acute myocardial infarction. The influence of discrimination limit, analytical quality, patient's sex and prevalence of disease

Abstract

By means of a graphical method the influence of the analytical variation and the discrimination limit (DL) on the diagnostic power of the maximum serum myoglobin value observed from 4 to 12 h after onset of symptoms in 291 patients suspected for myocardial infarction (AMI) was examined. The prevalence of AMI was 0.45 and the male to female ratio 2:1. Serum myoglobin (S-myoglobin) was measured by a radioimmunoassy (RIA) with a coefficient of analytical variation (CV_A) of 9%. For the distributions of the log values of maximum S-myoglobin for AMI patients and non-AMI patients straight lines were obtained on a probit scale. A statistically significant difference was found between the distributions for females and males without AMI, whereas no difference was found between females and males with AMI. The distributions of patients with and without AMI overlapped markedly giving a high number of misclassifications. The minimum fraction of misclassifications among all patients admitted occurred at a DL of 325 μg/1 and was 0.16. When S-myoglobin is used for the purpose of early diagnosis of AMI the DL should be chosen so that the fraction of false negative patients is small. Consequently the fraction of false positive patients will be relatively high. At a DL of, for example, 175 μg/l, the false negative fraction was 0.06 of all patients with AMI (sensitivity 0.94), and the fraction of false positive patients was 0.35 (specificity 0.65). At a prevalence of AMI of 0.45 the predictive value of a negative result was 0.93 and the predictive value of a positive result 0.69. Increase of analytical variation broadened the distributions and consequently the fraction of misclassifications increased. However, most of the rapid RIA methods recently described in the literature have a CV_A below 15%, and a CV_A in this range will have no influence of significance on the number of misclassifications. The number of misclassifications was highly dependent on the prevalence of AMI among the patients, whereas the sex ratio was of minor importance.

Key Words:

• acute myocardial infarction
• analytical variation
• discrimination limit
• predictive values
• prevalence
• probit transformation
• serum myoglobin