Game 1 · Diagnostic tests
Where should a test draw the line between positive and negative?
Sensitivity, specificity, PPV and NPV, and why PPV falls when a disease is rare.
Read with: Diagnostic tests, Bayes' rule and combining tests49 DNB questions
Play
Move the cut-off, meet the target
Each round gives you a job for the test and a target. Drag the cut-off, watch the two-by-two table change, then lock it in. Every fourth round asks you to predict the PPV in a new population. You have three lives.
Loading the first round…
The idea
Healthy and diseased people overlap
A test result such as a blood level is not a clean yes or no. Healthy people have a spread of values and diseased people have a higher spread, and the two spreads overlap. The cut-off is a line you draw across that overlap. Everyone to the right of the line is called positive.
Wherever you draw the line, some diseased people fall to its left (false negatives) and some healthy people fall to its right (false positives). Moving the line left catches more diseased people but calls more healthy people positive. Moving it right does the opposite. That is the trade-off you played with: sensitivity rises only when specificity falls.
SnNout: a highly Sensitive test, when Negative, rules the disease out. SpPin: a highly Specific test, when Positive, rules the disease in.
Formulas
The four measures from one table
Sensitivity = TP / (TP + FN)
Specificity = TN / (TN + FP)
PPV = TP / (TP + FP)
NPV = TN / (TN + FN)
- TP
- true positives: diseased, test positive
- FN
- false negatives: diseased, test negative
- FP
- false positives: healthy, test positive
- TN
- true negatives: healthy, test negative
| Feature | Sensitivity and specificity | PPV and NPV |
|---|---|---|
| Question answered | How good is the test in people whose status is known? | What does this result mean for my patient? |
| Read the table | Down the columns (disease present, disease absent) | Across the rows (test positive, test negative) |
| Change with prevalence? | No | Yes |
| Who uses it | The person choosing a test | The clinician reading a result |
Why PPV falls
The same test means less when the disease is rare
Take a test with 90% sensitivity and 90% specificity, used on 1,000 people. In a clinic where 50% have the disease, 450 of the 500 diseased test positive and 50 of the 500 healthy test positive, so 450 of 500 positives are true: PPV 90%. In a screening camp where 1% have the disease, 9 of the 10 diseased test positive but 99 of the 990 healthy also test positive, so only 9 of 108 positives are true: PPV 8.3%.
The test did not change. The healthy group became much larger, and 10% of a large group is a large number of false positives. This is why screening programmes confirm every positive with a second, more specific test.
To go further, the Bayes' rule chapter shows how to turn any pre-test probability into a post-test probability with likelihood ratios.
Writing the answer
How to write this in a 10-mark DNB answer
- Draw the two-by-two table first, with disease status in the columns and test result in the rows.
- Define each measure in words, then give its formula from the table.
- Work one example with round numbers, as above.
- State that sensitivity and specificity do not depend on prevalence, while PPV and NPV do, and show it with a second prevalence.
- Add SnNout and SpPin, and a line on choosing a cut-off (a sensitive cut-off to screen, a specific one to confirm).
- If the question mentions an ROC curve or likelihood ratios, add a line on each.
The one thing to remember
Sensitivity and specificity belong to the test. PPV and NPV belong to the test used in a particular population.
Past papers
DNB questions this game helps you answer
Questions from NBE theory papers for DNB, DrNB and Diploma courses, most often asked first. The tag after each question is the specialty and the year it was set.
- ×2Sensitivity and Specificity of a test.
- ×2Positive predictive value.
- ×1Define sensitivity and specificity. [2]
- ×1Sensitivity and specificity. [5]
- ×1Performance characteristics of a screening test.
- ×1Potential sources of bias in diagnostic accuracy studies.
- ×1A 35-year-old lady comes for a consultation. She is worried since her both parents are on treatment for diabetes. She seeks advice from you in taking care of her health. b) What is screening, what are the attributes for a good screening test.
- ×1A 35-year-old lady comes for a consultation. She is worried since her both parents are on treatment for diabetes. She seeks advice from you in taking care of her health. e) What are the benefits and disadvantages of screening test.
- ×1Define 'Yield'. Enlist and describe the factors which affects the 'Yield' with proper examples. Also add a note on evaluation of a screening test.
- ×1Describe the following for a given test: a) Sensitivity. [2.5] b) Specificity. [2.5] c) Positive predictive value. [2.5] d) Negative predictive value. [2.5]
- ×1How will you evaluate diagnostic accuracy of a test?
- ×1Positive and negative predictive value of a test.
- ×1What is sensitivity, specificity and predictive value?
- ×1Discuss the performance characteristics of a screening test.
- ×1Write notes on Sensitivity, Specificity, Positive Predictive Value and Hazard ratio. Give Examples.
- ×1Explain negative predictive value.
- ×1Explain what is meant by the sensitivity, specificity, and predictive value of a laboratory test.
- ×1Sensitivity, Specificity, Positive Predictive Value and Negative Predictive Value.
- ×1Also differentiate between specificity and sensitivity of a laboratory test.
- ×1Define positive predictive value.
- ×1Define positive predictive value. Suggest method to improve positive predictive value of a screening test.
- ×1Optimality of sepsis screen as an ideal screening test.
- ×1Screening Tests - Lead Time, Sensitivity, Specificity, Positive and Negative Predictive Values.
- ×1Explain the assessment of the ‘Validity (accuracy) of a Screening Test’ using a suitable example comparing a screening test with a gold standard.
- ×1Negative and positive predictive value.
- ×1Sensitivity and Specificity
- ×1Explain positive and negative predictive values.
- ×1How does sensitivity and specificity vary when multiple test is used sequentially and simultaneously? Explain with example.
- ×1Negative Predictive Value (NPV).
- ×1Positive Predictive Value (PPV).
- ×1Positive Predictive Value and Negative Predictive Value.
- ×1Problems in assessing the sensitivity and specificity of screening tests and cost benefit analysis of screening.
- ×1What is Sensitivity, Specificity, PPV and NPV?
- ×1What is sensitivity and specificity of diagnostic test?
- ×1What is sensitivity, specificity, positive predictive value, negative predictive value and accuracy?
- ×1Define the statistical terms “accuracy”, “positive predictive value” and “negative predictive value”, citing their clinical usefulness.
- ×1Positive predictive value
- ×1Predictive value.
- ×1Enumerate the criteria a screening test should fulfill.
- ×1What do the terms specificity and sensitivity mean?
- ×1With regards to a diagnostic test, define sensitivity, specificity, positive predictive value and negative predictive value.
- ×1In a laboratory test, explain with examples, the importance of: e) Positive predictive value
- ×1Uses and abuses of screening tests?
- ×1What is sensitivity, specificity, negative predictive value and positive predictive value of a diagnostic investigation?
- ×1Define sensitivity, specificity & positive predictive value.
- ×1List and briefly explain four measures of validity of a screening test.
- ×1List and briefly explain the terms "validity" and "reliability" in the context of screening test.
- ×1What is screening? What is the difference between a screening test and a diagnostic test? Describe briefly various types of screening. What are the uses of screening?
- ×1What is Bayes' theorem? Discuss sensitivity and specificity of exercise electrocardiography in coronary artery disease.
Read next