The scope of diagnostic tests ranges from signs and symptoms elicited during clinical examination, imaging tests, to biochemical, pathologic, and psychological tests. Tests that are capable of fully discriminating between the presence or absence of a disease are uncommon. Diagnostic testing is generally performed to screen for, detect, and monitor diseases. To optimize the use of diagnostic testing, clinicians should be aware of how the results of testing will affect determination of the probability of the presence of disease. To be useful, diagnostic tests should have the potential to change the pretest probability of disease into a post-test probability that is more definitive. The process of diagnostic testing should be based on a logical sequence that arrives at a sufficiently high probability of disease to make the diagnosis or a sufficiently low probability to exclude the diagnosis. These thresholds will vary from disease to disease; when the consequences of missing the disease (false-negative) have high potential to be disastrous, the threshold of post-test probability should be very low, whereas when the consequences of falsely making a diagnosis of the disease have the potential to be disastrous, the threshold of post-test probability should be very high. For example, when a patient has a suspected myocardial infarction (MI), clinicians require a combination of test results that has a very low (<2%) post-test probability since the consequences of missing an MI and sending the patient home are potentially disastrous. On the other hand, when making a diagnosis of MI, clinicians need a high post-test probability because the consequences of treatment (thrombolytic therapy, invasive strategies) and on prognosis (life expectancy) can be serious.
- To optimize the use of diagnostic testing, clinicians should be aware of how the results of testing will affect determination of the probability of the presence of disease. To be useful, diagnostic tests should have the potential to change the pretest probability.
The process of considering the diagnosis of a disease is often triggered by components of the history and physical examination, which lead the clinician to consider the presence of the disease. Other key components in considering a diagnosis include the experience and knowledge base of the diagnostician, the frequency of the disease, and the clinical importance of making or refuting a diagnosis.
Experience and knowledge base not only affect whether a disease shows up on the “radar screen,” but can also influence the accuracy of assessment of the clinical pretest probability (PTP). Two ways of evaluating PTP are by using clinician's “gestalt” or by using validated clinical prediction rules. The former might be favored by clinicians experienced in the disease of interest (eg, cardiologists for diagnosing MI), particularly when there is some subjectivity in aspects of the diagnosis. For pulmonary embolism (PE), the challenge becomes the spectrum of disease (ranging from clinically silent to multiple clinical presentations often with nonspecific symptoms and signs to hemodynamic collapse), the limitations of the history and physical examination (“the great masquerader”), and the absence of a single effective noninvasive test. Despite this, clinicians can assign meaningful probabilities for acute PE. When PIOPED clinicians assigned a high probability of PE prior to V/Q scanning, 67% of patients had PE. When ...