Glossary of Key Terms
Affective Bias – A significant cause of diagnostic error, affective bias occurs when the clinician’s affective state (emotions, personal situations, feeling about a patient) impacts the clinical reasoning process.
Bayesian Analysis – A mathematical construct that helps determine the likelihood of a particular disease or disorder based on the interaction between pre-test probability (note: we use pre-test probability but the calculation is actually made using pre-test odds) and the characteristics of a test. A clinician determines a pre-test probability and then applies a specific test. The combination of the pre-test probability (expressed as a percentage likelihood of the disease in question) and test characteristics of the test (expressed as likelihood ratios) results in a post- test probability of the disease. This probability is then compared to the threshold to test and the threshold to treat to see if further testing or treatment is appropriate.
Differential Diagnosis – The list of potential causes of a symptom, sign, syndrome, or disease.
Dual Process Model – The construct of clinical reasoning based on an interaction between Type I thinking (pattern recognition) and Type II thinking (analytic reasoning).
FARCOLDER – A mnemonic used to help recall all of the key questions to ask a patient with an acute concern. Not all of the elements of FARCOLDER are appropriate for every presenting concern.
Frequency
Associated symptoms
Radiation
Character
Onset
Location
Duration
Exacerbating factors
Relieving factors
Heuristics – Mental short-cuts or rules of thumb used to guide thinking. Heuristics are unconscious and are used both in clinical reasoning and everyday life, often during the process of pattern recognition. Common examples include the availability heuristic and the representativeness heuristic. (See guide to heuristics at the end of the syllabus for a more exhaustive list).
Hypothetico-Deductive Reasoning – The process by which the clinician proposes a specific potential diagnostic hypothesis and then examines all of the evidence supporting and refuting that specific diagnosis.
Illness Scripts – Mental constructs of diseases that clinicians build over time with increasing experience. They include the key and salient aspects of the disease and serve as the basis for pattern recognition.
Likelihood Ratio (positive) – Derived from the sensitivity and specificity of a test [sensitivity/ (1-specificity)], the positive likelihood ratio is used in Bayesian analysis when a test is positive. It reflects the proportion (or likelihood) of positive tests among patients with the disorder (true positive rate) to those with positive tests without the disorder (false positive rate). It is expressed as a number from 1 to infinity with higher number reflecting a greater likelihood of disease with a positive test. The following gives you a sense of a strong versus weak likelihood ratio: LR positive 2 = weak test, 5 = moderately good test, > 10 = strong test for proving a diagnosis. On occasion, we may use a positive likelihood ratio of <1. Under these circumstances, the positive test (or symptom or physical finding) result is associated with a decrease in the post-test probability of the specific disease.
Likelihood Ratio (negative) – Derived from the sensitivity and specificity of a test [(1-sensitivity)/specificity], the negative likelihood ratio is used in Bayesian analysis when a test is negative. It reflects proportion (or likelihood) of negative tests among patients with the disorder to those with negative tests without the disorder. It is expressed as a number from 1 to 1/infinity with lower numbers reflecting a lesser likelihood of disease with a negative test. The following gives you a sense of a weak versus strong likelihood ratio: LR negative 0.5 = weak test, 0.2 = moderately good test, < 0.1 = strong test for refuting a diagnosis. On occasion, we may use a negative likelihood ratio of >1. Under these circumstances, the negative test (or symptom or physical finding) result is associated with an increase in the post-test probability of the specific disease.
Pre-Test Probability – The proportion of patients who have a specific disease or disorder as determined by the clinician on the basis of the information available at that time. The pre-test probability may be determined by clinical gestalt, clinical prediction rules or antecedent Bayesian analysis.
Post-Test Probability – The proportion of patients who have a specific disorder or disease given a specific test result and the previously available clinical information. It is the result of the interaction between the pre-test probability and the test result. The post-test probability may itself then become the pre-test probability if another test is to be done for the specific disorder or disease under consideration.
Sensitivity – The percentage of patients with a specific disease who have a positive test result for a specific test.
Specificity – The percentage of patients without a specific disease who have a negative test result for a specific test.
Type I Thinking – A component of the dual process model of clinical reasoning; the utilization of pattern recognition as the major means of determining the likelihood of a specific diagnosis.
Type II Thinking – A component of the dual process model of clinical reasoning; the utilization of analytic reasoning as the major means of determining the likelihood of a specific diagnosis. Examples include hypothetico-deductive reasoning, Bayesian analysis, algorithmic thinking and worst case scenario medicine.
Threshold to Test – Expressed as a percentage, this is the number above which the probability of disease is high enough that testing for the disease should be undertaken. The threshold to test varies by disease and may be affected by the seriousness of the disease, the adverse effects of testing, and patient situation. Coronary artery disease, for example, has a low threshold to test because the ramifications of missing the diagnosis are significant.
Threshold to Treat – Expressed as a percentage, this is the number above which the probability of disease is high enough that disease treatment should be undertaken. The threshold to treat varies by disease and may be affected by the seriousness of the disease, adverse effects of treatment, and patient situation. Pneumonia, for example, has a low threshold to treat because the treatment (antibiotics) is relatively benign and lack of treatment may be fatal.
VINDICATE – A mnemonic used to structure the potential causes of a symptom, sign, syndrome or disease. It is very useful for helping to build differential diagnoses.
Vascular
Infection
Neoplasm
Drugs
Inflammatory/Idiopathic/Iatrogenic
Congenital
Auto-immune/Allergic
Trauma/Mechanical
Endocrine/Toxic/Metabolic