270- Sensitivity & Specificity

A test's _________ will tend to be higher in a sample of severely affected patients. A. Specificity B. Sensitivity

B.

An ideal cut-off point for a diagnostic test is in the ______________ portion of the receiver operator characteristic curve. A. Upper right B. Upper left C. Lower right D. Lower left

B.

In a 2x2 table, true negatives are found in cell: A. a B. b C. c D. d

D.

True or False: Reported values of sensitivity and specificity should be taken too literally if a small number of patients is studied.

True.

Define the validity of a diagnostic test. Define the reliability of a diagnostic test.

Validity (accuracy): the extent to which the observations correspond to the "truth" of the actual state of affairs. Reliability (precision, reproducibility): the extent to which a series of observations of the same thing match one another.

Diagnostic tests with low specificity have: A. Many false positives B. Many false negatives

A.

Diagnostic tests with many false negatives have: A. Low sensitivity B. Low specificity

A.

In a 2x2 table, true positives are found in cell: A. a B. b C. c D. d

A.

Negative predictive value goes down when a diagnostic test is used in a population where the disease is: A. Highly prevalent B. Low prevalence

A.

Performing multiple tests in sequence is termed: A. Series Testing B. Parallel Testing C. Confounding Testing D. Sequence Testing E. Perpendicular Testing

A.

Sensitive tests are most useful to clinicians when the results are: A. Negative B. Positive

A.

_________ tests are especially useful in "ruling out" diagnoses during the early stages of work up. A. Sensitive B. Specific

A.

__________ have good negative predictive value. A. Sensitive B. Specific

A.

__________ tests are especially useful when the penalty of missing disease is great. A. Sensitive B. Specific

A.

___________ tests are especially useful in "ruling in" diagnoses that have already been suggested by other data. A. Specific B. Sensitive

A.

______________ maximizes negative predictive value and sensitivity. A. Parallel testing B. Series Testing

A. A positive result on any of the tests warrants diagnosis, so few diseased patients will be missed (↑ sensitivity and NPV).

False positive diagnoses are more likely to be made in: A. Parallel testing B. Series testing

A. A positive result on any of the tests warrants diagnosis, so few diseased patients will be missed but people without the disease may be misclassified as diseased (false positive).

Compared to series testing, parallel testing is more: A. Sensitive B. Specific

A. A positive result on any of the tests warrants diagnosis, so few diseased patients will be missed.

When diagnostic tests are used for screening asymptomatic patients they have _____________ compared to when the same test is used in patients suspected of having the disease. A. Higher specificity, lower sensitivity B. Higher specificity and sensitivity C. Lower specificity, higher sensitivity D. Lower sensitivity and specificity

A. When the disease is in its early stages, the test is less likely to pick it up but will do well at identifying those without the disease. Therefore, there will be fewer false positives (higher specificity) and more false negatives (lower sensitivity).

As a result of natural reluctance to initiate aggressive workups, data on the number of true and false __________ generated by a test (in cells ________ of a typical 2x2 table) tend to be much less complete in medical literature. A. Negatives, a+b B. Negatives, c+d C. Positives, a+b D. Positives, c+d

B.

Diagnostic tests with low sensitivity have: A. Many false positives B. Many false negatives

B.

Diagnostic tests with many false positives have: A. Low sensitivity B. Low specificity

B.

In a 2x2 table, false positives are found in cell: A. a B. b C. c D. d

B.

Performing multiple tests at one time is termed: A. Series Testing B. Parallel Testing C. Confounding Testing D. Cross-sectional Testing E. Perpendicular Testing

B.

Positive predictive value goes down when a diagnostic test is used in a population where the disease is: A. High prevalence B. Low prevalence

B.

Specific tests are most useful to clinicians when the results are: A. Negative B. Positive

B.

When a test's sensitivity and specificity are initially evaluated by testing on medical student volunteers, the _________ will be high. A. Sensitivity B. Specificity

B.

_________ results, even when ________ is high, will likely be false when the prevalence of disease is low (i.e. patient has a low likelihood of having the disease). A. Positive, sensitivity B. Positive, specificity C. Negative, sensitivity D. Negative, specificity

B.

__________ tests are especially needed when the consequences of further work up on false positives is potentially harmful. A. Sensitive B. Specific

B.

___________ have good positive predictive value. A. Sensitive B. Specific

B.

The y-axis of the receiver-operator characteristic (ROC) curve conveys the: A. % specificity B. % sensitivity C. False positive rate D. False negative rate E. 1-% specificity

B. % sensitivity = true positive rate

______________ maximizes positive predictive value. A. Parallel testing B. Series Testing

B. A diagnosis requires positive results on all tests, so few patients without the disease will be misclassified as diseased (↑specificity and PPV).

Compared to parallel testing, serial testing is more: A. Sensitive B. Specific

B. A diagnosis requires positive results on all tests, so few patients without the disease will be misclassified as diseased.

When some tests required for diagnosis are risky or expensive, clinicians are more likely to use: A. Parallel testing B. Series testing

B. Subsequent tests are only performed if the test before was positive.

If you decrease your test's cut-off point for the diagnosis of hypertension, you will effectively ________________. A. ↑ sensitivity, ↑ specificity B. ↑ sensitivity, ↓ specificity C. ↓ sensitivity, ↑ specificity D. ↓ sensitivity, ↓ specificity

B. • You are less likely to miss disease • Your are more likely to misclassify someone without disease as being diseased

You screen 200 patients from a geriatric hospital for ARMD. If you then screened 200 patients from the general hospital, which of the following is true? A. Positive predictive value will ↑ B. Positive predictive value will ↓ C. Positive predictive value is unaffected

B. ↓ prevalence in the younger population will ↓ PPV

In a 2x2 table, false negatives are found in cell: A. a B. b C. c D. d

C.

The overall accuracy of a diagnostic test can be described as: A. The slope of the ROC curve B. The area above the ROC curve C. The area under the ROC curve D. The maximum value on the ROC curve

C.

_________ results, even when ________ is high, will likely be false when the prevalence of disease is high (i.e. Patient has a high likelihood of having the disease). A. Positive, sensitivity B. Positive, specificity C. Negative, sensitivity D. Negative, specificity

C.

___________ is also called "posttest probability" A. Sensitivity B. Specificity C. Predictive value D. Incidence E. Prevalence

C.

If you increase your test's cut-off point for the diagnosis of hypertension, you will effectively ________________. A. ↑ sensitivity, ↑ specificity B. ↑ sensitivity, ↓ specificity C. ↓ sensitivity, ↑ specificity D. ↓ sensitivity, ↓ specificity

C. • You are more likely to miss disease • Your are less likely to misclassify someone without disease as being diseased

You screen 200 patients for a particular disease. If you screened 20,000 patients, which of the following is true? A. Positive predictive value will ↑ B. Positive predictive value will ↓ C. Positive predictive value is unaffected

C. ↑ prevalence of the disease will ↑ PPV ↑ the sample # does not affect the prevalence or PPV

The x-axis of the receiver-operator characteristic (ROC) curve conveys the: A. % specificity B. % sensitivity C. True positive rate D. False negative rate E. 1-% specificity

E. 1-% specificity = false positive rate

___________ is also called "pretest probability" A. Sensitivity B. Specificity C. Positive predictive value D. Negative predictive value E. Incidence F. Prevalence

F.

Provide an example of a situation in which you would not wish to maximize the sensitivity of a test. Provide an example of a situation in which you would not wish to maximize the specificity of a test.

If false-positive labeling were harmful (e.g. expensive or dangerous follow-up testing or treatment). If false negative labeling were harmful (e.g. dangerous disease).

Define likelihood ratio. What is the formula for positive LR using the cells from a typical 2x2 table? What is the formula for negative LR using the cells from a typical 2x2 table?

The probability of a particular test result (i.e. positive or negative) in people with the disease divided by the probability of that result in people without the disease. +LR = [a/(a+c)] / [b/(b+d)] - LR = [c/(c+a)] / [d/(d+b)]

As you ↓ the cutoff value for a diagnostic test, how is this reflected by the position on the ROC curve?

• Higher on the curve (↑ sensitivity) • Farther to the right (↓ specificity

As you ↑ the cutoff value for a diagnostic test, how is this reflected by the position on the ROC curve?

• Lower on the curve (↓ sensitivity) • Farther to the left (↑specificity

Define the sensitivity of a diagnostic test. What is the formula using a 2 x 2 table? What is the formula using false/true positives/negatives?

• Proportion of patients with the disease that have a positive test. A sensitive test rarely misses people with the disease. • Sensitivity = a/(a+c) • Sensitivity = true positives/(true positives + false negatives)

Define the specificity of a diagnostic test. What is the formula using a 2 x 2 table? What is the formula using false/true positives/negatives?

• Specificity: proportion of patients without the disease that have a negative test. A specific test rarely misclassifies people without the disease as diseased. • Specificity = d/(d+b) • Specificity = true negatives/(true negatives + false positives)