Intro to Clinical Reasoning

Question 1

T or F. Because they are characteristics of the test (not the population), Sens and Spec are of no direct use in estimating the probability of disease in an individual patient.

Answer 1

T.

Don’t be fooled by this! If asked, “What is the probability that a patient who has a positive result on a test with 99% sensitivity and 95% specificity for disease X has the disease?” your answer should be, “It depends on the pre-test probability.” Sens and spec alone cannot provide an answer to this question.

Question 2

What is Specificity?

Answer 2

proportion of subjects who do not have the disease who have a negative test result

Question 3

Tests with high specificity are most helpful in “ruling in”* the presence of disease

 Specific tests, if Positive, rule IN disease (mnemonic: “SpPIN”)

Answer 3

Tests with high specificity are most helpful in “ruling in”* the presence of disease

 Specific tests, if Positive, rule IN disease (mnemonic: “SpPIN”)

Question 4

What is Sensitivity?

Answer 4

= proportion of subjects with the disease who have a positive test result (mnemonic: Positive In Disease [“PID”])

Question 5

Tests with high sensitivity are most helpful in “ruling out”* the presence of disease

 Sensitive tests, if Negative, rule OUT disease (mnemonic: “SnNOUT”)

Answer 5

Tests with high sensitivity are most helpful in “ruling out”* the presence of disease

 Sensitive tests, if Negative, rule OUT disease (mnemonic: “SnNOUT”)

Question 6

Note that “ruling in” and “ruling out” are relative terms. Only tests with 100% specificity and 100% sensitivity could definitively determine the presence or absence of disease, respectively.

Answer 6

Note that “ruling in” and “ruling out” are relative terms. Only tests with 100% specificity and 100% sensitivity could definitively determine the presence or absence of disease, respectively.

Question 7

What are predictive values?

Answer 7

Predictive values answer the more clinically relevant question, “What is the probability that this patient has this disease, given this test result?”

Question 8

What do predictive values depend on?

Answer 8

Predictive values depend upon characteristics of the test (i.e., Sens and Spec) and the prevalence (pre-test probability) of the disease in the population (patient) tested.

Don’t forget the latter! The predictive value of a test may vary dramatically depending on the pre-test probability of disease.

Question 9

What are the most useful single indicators of a test’s diagnostic strength, i.e. the degree to which it can modify pretest probability and facilitate clinical decision-making?

Answer 9

likelihood ratios

Question 10

What is a likelihood ratio?

Answer 10

the ratio between the probability of observing that result in patients with the disease in question, and the probability of that result in patients without the disease

Question 11

Eqn for LR+?

Answer 11

proportion of people with disease who test positive/ proportion of people without disease who test positive

Question 12

Eqn for LR-?

Answer 12

proportion of people with disease who test negative/ proportion of people without disease who test negative

Question 13

The diagnostic value of a ‘test’ (of any type) depends on what?

Answer 13

how much its result changes the probability of the disease(s) it tests for.