Sensitivity, specificity and predictive values

9 ASCP MLS practice questions on Sensitivity, specificity and predictive values, part of Laboratory Operations. The first 9 are below; the first 5 with answers and explanations (tap Show answer after you have chosen), the rest in Learn with a free account.

1. Question 1 · Hard

   A public health laboratory screens 20,000 people for an infection with a prevalence of 0.5%. The screening immunoassay has a sensitivity of 99% and a specificity of 99%. Every reactive screen is then tested with a supplemental assay that has a sensitivity of 98% and a specificity of 99.5%; assume that the two tests make their errors independently. A clinician asks how reliable a reactive result is. The positive predictive values of a reactive screen alone, and of a result reactive on both tests, are:

   1. AAbout 99% and about 99%
   2. BAbout 67% and about 99%
   3. CAbout 33% and about 99%
   4. DAbout 33% and about 33%

   Show answerHide answer

   Answer: C · About 33% and about 99%

   Of 20,000 people, 100 are infected and 19,900 are not. The screen detects 99 true positives and gives 1% of 19,900, or 199, false positives, so its PPV is 99 ÷ 298 = about 33%, even though its sensitivity and specificity are both 99%. The supplemental test confirms 98% of the true positives (about 97) but only 0.5% of the false positives (about 1), so the PPV of a double-reactive result is about 97 ÷ 98 = 99%. Testing in series raises specificity and PPV at a small cost in sensitivity, which is why low-prevalence screening programs confirm reactive screens before reporting them.

   • A: 99% is the screen's sensitivity and specificity, not its predictive value at a prevalence of 0.5%.
   • B: About 67% is the share of reactive screens that are false positives, not the PPV.
   • D: The supplemental test removes nearly all false positives, so the PPV of a double-reactive result rises.

2. Question 2 · Medium

   A hospital wants a troponin I cutoff for a rule-out pathway that misses as few myocardial infarctions as possible. Based on the ROC curve shown, which cutoff should it choose?

   

   EnlargeClose

   

   1. A20 ng/L
   2. B80 ng/L
   3. C40 ng/L
   4. D10 ng/L

   Show answerHide answer

   Answer: D · 10 ng/L

   Each point plots the sensitivity of one cutoff against its false-positive rate (1 - specificity). Missing the fewest infarctions means the fewest false negatives, which is the highest sensitivity: 10 ng/L, at about 99%, with a specificity of about 60%. The 20 ng/L cutoff lies closest to the top-left corner and gives the best overall balance, but it misses about 8% of infarctions. Lowering a cutoff raises sensitivity at the cost of more false positives, which then need further testing.

   • A: 20 ng/L gives the best balance but a sensitivity of only about 92%.
   • B: 80 ng/L has the highest specificity but a sensitivity of only about 55%.
   • C: 40 ng/L has a sensitivity of about 78% and misses about one infarction in five.

3. Question 3 · Hard

   Of 500 patients with chest pain, 110 had a myocardial infarction. A new troponin assay was positive in 120 patients, 100 of whom had an infarction. What is the assay's specificity?

   1. A83.3%
   2. B94.9%
   3. C97.4%
   4. D90.9%

   Show answerHide answer

   Answer: B · 94.9%

   First build the 2 x 2 table. True positives are 100 and false positives 120 - 100 = 20; 500 - 110 = 390 patients had no infarction, so true negatives are 390 - 20 = 370. Specificity = 370 ÷ 390 = 94.9%. The same table gives sensitivity 100 ÷ 110 = 90.9%, PPV 100 ÷ 120 = 83.3% and NPV 370 ÷ 380 = 97.4%.

   • A: 83.3% is the PPV: 100 of 120 positive results.
   • C: 97.4% is the NPV: 370 of 380 negative results.
   • D: 90.9% is the sensitivity: 100 of 110 infarctions detected.

4. Question 4 · Hard

   A screening test with 95% sensitivity and 95% specificity is used on 1,000 people, of whom 2% have the disease. About what percentage of the positive results will be true positives?

   1. AAbout 28%
   2. BAbout 95%
   3. CAbout 99.9%
   4. DAbout 72%

   Show answerHide answer

   Answer: A · About 28%

   Of the 1,000 people, 20 have the disease, and 95% of them (19) test positive. Of the 980 without the disease, 5% (49) also test positive. PPV = 19 ÷ (19 + 49) = 28%, so most positive results are false. In a low-prevalence population, positive screens must be confirmed with a more specific test.

   • B: 95% is the sensitivity, not the PPV.
   • C: 99.9% is the NPV (931 ÷ 932).
   • D: 72% is the share of positive results that are false.

5. Question 5 · Hard

   A 34-year-old woman in the emergency department has calf pain, and her clinical probability of deep vein thrombosis (DVT) is low. A negative D-dimer is used to avoid ultrasound. This approach relies mainly on the D-dimer test's:

   1. AHigh PPV, so a positive confirms DVT
   2. BHigh specificity, so its PPV is high
   3. CLow imprecision, so repeats agree
   4. DHigh sensitivity, so its NPV is high

   Show answerHide answer

   Answer: D · High sensitivity, so its NPV is high

   A rule-out test must rarely miss disease, so it needs high sensitivity; with few false negatives, a negative result has a high NPV. D-dimer is very sensitive for venous thrombosis but not specific, because it also rises with surgery, infection, pregnancy, cancer and age. Its NPV is highest when the pretest probability is low, which is why it is used only in low-probability patients. A positive D-dimer does not diagnose DVT and is followed by imaging.

   • A: A positive D-dimer is nonspecific and needs imaging to confirm DVT.
   • B: D-dimer has low specificity, so its PPV is poor.
   • C: Precision concerns repeatability, not the ability to rule out disease.

6. Question 6 · Medium

   A test is moved from a specialist clinic, where the disease is common, to population screening, where it is rare. Its sensitivity and specificity do not change. What happens to its predictive values?

   1. APPV falls and NPV rises
   2. BNeither changes, as the test is the same
   3. CBoth fall with lower prevalence
   4. DPPV rises and NPV falls

   Answer and explanation in LearnFree account

7. Question 7 · Medium

   A new assay is compared with a reference method in 400 patients: 80 true positives, 30 false positives, 20 false negatives and 270 true negatives. Its positive predictive value is:

   1. A80.0%
   2. B72.7%
   3. C90.0%
   4. D93.1%

   Answer and explanation in LearnFree account

8. Question 8 · Easy

   The specificity of a test is the percentage of:

   1. ANegative results that are true negatives
   2. BPeople with the disease who test positive
   3. CPositive results that are true positives
   4. DPeople without the disease who test negative

   Answer and explanation in LearnFree account

9. Question 9 · Medium

   A new test is positive in 90 of 100 patients known to have the disease. What is its sensitivity?

   1. A10%
   2. BIt cannot be calculated without the prevalence
   3. C100%
   4. D90%

   Answer and explanation in LearnFree account