Observations on the Probability of Error in Medical Diagnosis

Publisher Website
Google Scholar
Abstract
Measurements of total serum cholesterol used to demonstrate how diagnostic error can be measured or controlled. Two populations, 1 with coronary heart disease and myocardial infarction and the other population without, were studied. The probability of a wrong provisional diagnosis is calculated and criteria for a definitive diagnosis when the separate probabilities of error are specified in advance are established. Techniques of combining indices that are not highly correlated are suggested, as is the practice of averaging 2 or more measurements on each patient.