The expected value of clinical information

When and how should a physician determine whether to obtain additional clinical information, or whether to start a treatment in view of the information available to him? The concept of the expected value of clinical information can assist the physician in arriving at this decision. Expected value of clinical information is defined as "the difference between the outcome rate (e.g. death rate or survival rate) when the treatment decision must be made without the test information and the outcome rate when the decision can be made contingent upon the test result." We will utilize the decision problem already presented by H.V. Fineberg for patients with chronic progressive liver failure to illustrate this concept, its calculation, and its relationship to factors such as risk associated with the test procedure, sensitivity and specificity of the test, and prevalence of the disease(s) in the population. Usually, "performing a test to gain additional information is worthwhile if at least one decision would change given some test results, and the intrinsic risk of using the test is less than the expected benefit gained from the subsequent change in decision." We will than briefly introduce the subject of sensitivity analysis to illustrate the interrelationship of these different factors upon the decision.