Diagnostic tests
Judging a test: sensitivity and specificity, positive and negative predictive values and how they depend on prevalence, false positives and false negatives, choosing the threshold and the ROC curve, likelihood ratios, with calculations on real cases.
- 34 cards
- 3 topics
- Free
The outline, topic by topic
Each topic is weighted the way the exam weighs it.
- Concepts
- Scenarios
- Understand
A few cards from the deck
Sensitivity: what is it?
Show the answer
The share of sick people the test detects (positives among the sick).
A highly sensitive test has few false negatives: negative, it rules the disease out.
Out of 100 sick people, a test detects 90. What is its sensitivity?
- 90 %
- 10 %
- 100 %
- It cannot be calculated without the healthy
Show the answer
90 %
Sensitivity = true positives / sick = 90 / 100.
Specificity: what is it?
Show the answer
The share of healthy people the test classes as negative.
A highly specific test has few false positives: positive, it rules the disease in.
Why does sensitivity not depend on prevalence?
- It depends on prevalence
- It is calculated among the healthy
- It is always 100 %
- It is calculated only among the sick
Show the answer
It is calculated only among the sick
Likewise, specificity is calculated only among the healthy.
Positive predictive value (PPV): what is it?
Show the answer
The probability of being sick when the test is positive.
It depends on prevalence: low for a rare disease.
Out of 1,000 healthy people, 950 test negative. What is the specificity?
- 5 %
- 95 %
- 50 %
- 99.5 %
Show the answer
95 %
Specificity = true negatives / healthy = 950 / 1,000.