Resources

Public health bioethicists evaluate genetic screening policies for a rare adult-onset neurodegenerative condition with a population prevalence of P(D)=0.00200P(D) = 0.00200 (0.200%0.200\%, or 11 in 500500). A commercial high-throughput genomic assay advertises high analytical performance: sensitivity (true positive rate) of P(+∣D)=0.990P(+\mid D) = 0.990 (99.0%99.0\%) and specificity (true negative rate) of P(−∣Dc)=0.980P(-\mid D^c) = 0.980 (98.0%98.0\%). Despite high accuracy metrics, student intuition frequently presumes that an individual testing positive has a greater than 90%90\% chance of having the disease. Applying Bayes' Theorem to a cohort of N=100,000N = 100{,}000 asymptomatic individuals: (a) calculate the positive predictive value PPV=P(D∣+)=P(+∣D)P(D)P(+∣D)P(D)+P(+∣Dc)P(Dc)PPV = P(D\mid +) = \frac{P(+\mid D)P(D)}{P(+\mid D)P(D) + P(+\mid D^c)P(D^c)}, and (b) determine the false discovery rate FDR=1−PPVFDR = 1 - PPV, expressing both as percentages rounded to three significant figures.

Answer and steps

Final answer

(a) PPV=(0.990)(0.00200)(0.990)(0.00200)+(1−0.980)(1−0.00200)=0.001980.00198+0.01996=0.001980.02194≈0.0902=9.02%PPV = \frac{(0.990)(0.00200)}{(0.990)(0.00200) + (1 - 0.980)(1 - 0.00200)} = \frac{0.00198}{0.00198 + 0.01996} = \frac{0.00198}{0.02194} \approx 0.0902 = 9.02\%; (b) FDR=100%−9.02%=90.98%≈91.0%FDR = 100\% - 9.02\% = 90.98\% \approx 91.0\%

Steps

01

Calculate the Positive Predictive Value (PPVPPV)

Understanding the Given Parameters

From the population and test metrics:

  • Prevalence of disease (P(D)P(D)): 0.002000.00200 (or 0.200%0.200\%)
  • Complement (proportion healthy, P(Dc)P(D^c)): P(Dc)=1−P(D)=1−0.00200=0.99800P(D^c) = 1 - P(D) = 1 - 0.00200 = 0.99800
  • Sensitivity (P(+∣D)P(+\mid D)): 0.9900.990 (true positive rate)
  • Specificity (P(−∣Dc)P(-\mid D^c)): 0.9800.980 (true negative rate)
  • False positive rate (P(+∣Dc)P(+\mid D^c)): P(+∣Dc)=1−P(−∣Dc)=1−0.980=0.0200P(+\mid D^c) = 1 - P(-\mid D^c) = 1 - 0.980 = 0.0200

The Positive Predictive Value (PPV=P(D∣+)PPV = P(D\mid +)) represents the probability that a person who receives a positive test result genuinely has the disease. Using Bayes' Theorem:

PPV=P(+∣D)P(D)P(+∣D)P(D)+P(+∣Dc)P(Dc)PPV = \frac{P(+\mid D)P(D)}{P(+\mid D)P(D) + P(+\mid D^c)P(D^c)}

  1. Calculate the joint probability of a true positive (TPTP): P(+∣D)P(D)=0.990⋅0.00200=0.00198P(+\mid D)P(D) = 0.990 \cdot 0.00200 = 0.00198
  2. Calculate the joint probability of a false positive (FPFP): P(+∣Dc)P(Dc)=(1−0.980)(1−0.00200)=0.0200⋅0.99800=0.01996P(+\mid D^c)P(D^c) = (1 - 0.980)(1 - 0.00200) = 0.0200 \cdot 0.99800 = 0.01996
  3. Calculate the total probability of testing positive (P(+)P(+)): P(+)=0.00198+0.01996=0.02194P(+) = 0.00198 + 0.01996 = 0.02194
  4. Compute PPVPPV: PPV=0.001980.02194≈0.090246…PPV = \frac{0.00198}{0.02194} \approx 0.090246 \dots

Expressed as a percentage to three significant figures: PPV≈9.02%PPV \approx 9.02\%


02

Determine the False Discovery Rate (FDRFDR)

The False Discovery Rate represents the proportion of positive results that are incorrect (false alarms):

FDR=1−PPV=100%−9.02%=90.98%FDR = 1 - PPV = 100\% - 9.02\% = 90.98\%

Rounding to three significant figures gives: FDR≈91.0%FDR \approx 91.0\%


Natural Frequency Interpretation (Cohort of 100,000100{,}000)

To visualize why this occurs despite high test sensitivity (99%99\%) and specificity (98%98\%):

  • In a cohort of 100,000100{,}000 individuals, only 200200 actually have the condition (100,000⋅0.00200100{,}000 \cdot 0.00200). The test correctly flags 198198 of them (200⋅0.990200 \cdot 0.990).
  • Meanwhile, 99,80099{,}800 individuals do not have the condition (100,000⋅0.99800100{,}000 \cdot 0.99800). The test incorrectly flags 2%2\% of them as positive: 99,800⋅0.0200=1,996 false positives99{,}800 \cdot 0.0200 = 1{,}996\text{ false positives}
  • Out of all 198+1,996=2,194198 + 1{,}996 = 2{,}194 positive results, only 198198 are true cases: 1982,194≈9.02%\frac{198}{2{,}194} \approx 9.02\%

Key Pitfall: The "base-rate fallacy" occurs when one equates high analytical sensitivity (99%99\%) with the probability of having the disease given a positive result. When a condition is very rare, the vast pool of healthy individuals generates far more absolute false positives than the tiny pool of affected individuals generates true positives.

Typical errors

OverviewACEpath Teacher's Diagnosis

01 The student confuses specificity with the false positive rate by plugging 0.980 directly into the denominator instead of 1 - 0.980 = 0.020.

Sample Student Work

Reviewed by ACEpath Teacher
Step-by-Step Review
Walk Through It with Your Teacher
Incorrect
  1. Step 1

    P(D)=0.00200,  P(Dc)=0.998  ⟹  P(+∣D)=0.990,  P(+∣Dc)=0.980P(D) = 0.00200,\; P(D^c) = 0.998 \implies P(+\mid D) = 0.990,\; P(+\mid D^c) = 0.980

    Correct

    P(D)=0.00200,  P(Dc)=0.998  ⟹  P(+∣D)=0.990,  P(+∣Dc)=1−0.980=0.0200P(D) = 0.00200,\; P(D^c) = 0.998 \implies P(+\mid D) = 0.990,\; P(+\mid D^c) = 1 - 0.980 = 0.0200

    You correctly identified disease prevalence P(D)P(D), healthy proportion P(Dc)P(D^c), and sensitivity P(+∣D)P(+\mid D). However, 0.9800.980 is the test specificity P(−∣Dc)P(-\mid D^c) (the probability of testing negative given no disease). The false positive rate required by the formula is P(+∣Dc)=1−P(−∣Dc)=1−0.980=0.0200P(+\mid D^c) = 1 - P(-\mid D^c) = 1 - 0.980 = 0.0200.

  2. Step 2

    PPV=0.990×0.00200(0.990×0.00200)+(0.980×0.998)=0.001980.00198+0.97804=0.001980.98002≈0.00202  ⟹  PPV=0.202%PPV = \frac{0.990 \times 0.00200}{(0.990 \times 0.00200) + (0.980 \times 0.998)} = \frac{0.00198}{0.00198 + 0.97804} = \frac{0.00198}{0.98002} \approx 0.00202 \implies PPV = 0.202\%

    Correct

    PPV=0.990×0.00200(0.990×0.00200)+(0.0200×0.998)=0.001980.00198+0.01996=0.001980.02194≈9.02%PPV = \frac{0.990 \times 0.00200}{(0.990 \times 0.00200) + (0.0200 \times 0.998)} = \frac{0.00198}{0.00198 + 0.01996} = \frac{0.00198}{0.02194} \approx 9.02\%

    Your arithmetic and algebraic structure were consistent with your chosen values, but your result was distorted by using 0.9800.980 instead of 0.02000.0200 in the denominator.

  3. Step 3

    FDR=1−0.00202=0.99798  ⟹  FDR=99.8%FDR = 1 - 0.00202 = 0.99798 \implies FDR = 99.8\%

    Correct

    FDR=1−0.0902=0.9098≈91.0%FDR = 1 - 0.0902 = 0.9098 \approx 91.0\%

    You correctly used the definition FDR=1−PPVFDR = 1 - PPV, but your answer inherited the numerical error from the previous step.

WHAT YOU DID WELL
Compliments from Your Teacher

You correctly identified the base prevalence P(D)P(D) and calculated its complement P(Dc)=0.998P(D^c) = 0.998.

You correctly applied the relation FDR=1−PPVFDR = 1 - PPV.

THE ERROR
Where Your ACEpath Teacher Spotted the Mistake
Specificity vs. False Positive Rate Confusion

no

WHY IT HAPPENED
ACEpath Teacher's Diagnosis

You treated specificity P(−∣Dc)P(-\mid D^c) as if it were the false positive rate P(+∣Dc)P(+\mid D^c), rather than recognizing that testing positive when healthy is the complement of testing negative when healthy (P(+∣Dc)=1−specificityP(+\mid D^c) = 1 - \text{specificity}).

REASONING PATH
Your ACEpath teacher’s suggested approach
Define Test Metrics by Outcome Sign
  • Specificity is the true negative rate, P(−∣Dc)P(-\mid D^c). The false positive rate is its complement among healthy individuals: P(+∣Dc)=1−P(−∣Dc)P(+\mid D^c) = 1 - P(-\mid D^c).

  • Always write down the full conditional notation P(−∣Dc)P(-\mid D^c) for specificity first, then compute P(+∣Dc)=1−specificityP(+\mid D^c) = 1 - \text{specificity} before substituting into Bayes' Theorem.

CORRECTED PATH
ACEpath Teacher's Demonstration
  1. Determine the false positive rate: P(+∣Dc)=1−P(−∣Dc)=1−0.980=0.0200P(+\mid D^c) = 1 - P(-\mid D^c) = 1 - 0.980 = 0.0200.

  2. Substitute into Bayes' Theorem: PPV=0.990×0.00200(0.990×0.00200)+(0.0200×0.99800)=0.001980.00198+0.01996=0.001980.02194≈0.090246PPV = \frac{0.990 \times 0.00200}{(0.990 \times 0.00200) + (0.0200 \times 0.99800)} = \frac{0.00198}{0.00198 + 0.01996} = \frac{0.00198}{0.02194} \approx 0.090246.

  3. Convert to percentage and round to three significant figures: PPV≈9.02%PPV \approx 9.02\%.

  4. Calculate the False Discovery Rate: FDR=1−PPV=1−0.090246≈0.9098  ⟹  FDR≈91.0%FDR = 1 - PPV = 1 - 0.090246 \approx 0.9098 \implies FDR \approx 91.0\%.

Final answer

(a) PPV≈9.02%PPV \approx 9.02\%; (b) FDR≈91.0%FDR \approx 91.0\%

NEXT STEPS
Final Takeaways from Your Teacher
  • Review the distinction between specificity P(−∣Dc)P(-\mid D^c) and false positive rate P(+∣Dc)P(+\mid D^c).

  • Use natural frequencies (e.g., imagining a cohort of 100,000100{,}000 people) as an intuitive sanity check for Bayesian screening problems.

OverviewACEpath Teacher's Diagnosis

02 The student commits the base rate fallacy by omitting disease prevalence and equating PPV to sensitivity divided by sensitivity plus false positive rate.

Sample Student Work

Reviewed by ACEpath Teacher
Step-by-Step Review
Walk Through It with Your Teacher
Incorrect
  1. Step 1

    P(+∣Dc)=1−0.980=0.020P(+\mid D^c) = 1 - 0.980 = 0.020

  2. Step 2

    PPV=0.990(0.990+0.020)=0.9901.010≈0.980198  ⟹  PPV=98.0%PPV = \frac{0.990}{(0.990 + 0.020)} = \frac{0.990}{1.010} \approx 0.980198 \implies PPV = 98.0\%

    Correct

    PPV=P(+∣D)P(D)P(+∣D)P(D)+P(+∣Dc)P(Dc)=(0.990)(0.00200)(0.990)(0.00200)+(0.020)(0.99800)=0.001980.02194≈9.02%PPV = \frac{P(+\mid D)P(D)}{P(+\mid D)P(D) + P(+\mid D^c)P(D^c)} = \frac{(0.990)(0.00200)}{(0.990)(0.00200) + (0.020)(0.99800)} = \frac{0.00198}{0.02194} \approx 9.02\%

    You omitted the prior probabilities (prevalence P(D)=0.00200P(D) = 0.00200 and healthy proportion P(Dc)=0.99800P(D^c) = 0.99800) from both the numerator and denominator of Bayes' Theorem.

  3. Step 3

    FDR=1−0.9802=0.0198  ⟹  FDR=1.98%FDR = 1 - 0.9802 = 0.0198 \implies FDR = 1.98\%

    Correct

    FDR=1−PPV=1−0.0902=0.9098≈91.0%FDR = 1 - PPV = 1 - 0.0902 = 0.9098 \approx 91.0\%

    You correctly applied the relationship FDR=1−PPVFDR = 1 - PPV, but your result was affected by the incorrect value of PPVPPV from the previous step.

WHAT YOU DID WELL
Compliments from Your Teacher

You correctly calculated the false positive rate as the complement of specificity.

You correctly set up the formula for the false discovery rate as the complement of the positive predictive value.

THE ERROR
Where Your ACEpath Teacher Spotted the Mistake
Base-rate neglect in Bayes' Theorem

none

WHY IT HAPPENED
ACEpath Teacher's Diagnosis

You treated the conditional probabilities (sensitivity and false positive rate) as absolute counts or probabilities, ignoring the extremely low population prevalence of the disease (the base-rate fallacy).

REASONING PATH
Your ACEpath teacher’s suggested approach
Weighting by Prior Prevalence
  • PPV=Sensitivity×P(D)[Sensitivity×P(D)]+[False Positive Rate×P(Dc)]PPV = \frac{\text{Sensitivity} \times P(D)}{[\text{Sensitivity} \times P(D)] + [\text{False Positive Rate} \times P(D^c)]}

  • Always multiply conditional test rates by their respective subpopulation sizes before calculating predictive values.

CORRECTED PATH
ACEpath Teacher's Demonstration
  1. Calculate true positive joint probability: P(+∩D)=0.990×0.00200=0.00198P(+\cap D) = 0.990 \times 0.00200 = 0.00198.

  2. Calculate false positive joint probability: P(+∩Dc)=0.020×(1−0.00200)=0.020×0.99800=0.01996P(+\cap D^c) = 0.020 \times (1 - 0.00200) = 0.020 \times 0.99800 = 0.01996.

  3. Sum both to find total positive probability: P(+)=0.00198+0.01996=0.02194P(+) = 0.00198 + 0.01996 = 0.02194.

  4. Compute positive predictive value: PPV=0.001980.02194≈0.090246≈9.02%PPV = \frac{0.00198}{0.02194} \approx 0.090246 \approx 9.02\%.

  5. Compute false discovery rate: FDR=1−0.090246≈0.90975≈91.0%FDR = 1 - 0.090246 \approx 0.90975 \approx 91.0\%.

Final answer

(a) PPV≈9.02%PPV \approx 9.02\%; (b) FDR≈91.0%FDR \approx 91.0\%

NEXT STEPS
Final Takeaways from Your Teacher
  • Practice converting Bayes' Theorem problems into natural frequency tables (e.g., using a hypothetical cohort of 100,000 individuals) to make false positive counts intuitive.

  • Remember that when a disease is rare, false positives from the large healthy population often vastly outnumber true positives from the small diseased group.

OverviewACEpath Teacher's Diagnosis

03 The student ignores healthy false positives in the cohort, assuming everyone without the disease tests negative so that all positive tests are true positives.

Sample Student Work

Reviewed by ACEpath Teacher
Step-by-Step Review
Walk Through It with Your Teacher
Incorrect
  1. Step 1

    True positives=100000×0.00200×0.990=198\text{True positives} = 100000 \times 0.00200 \times 0.990 = 198

  2. Step 2

    Total positive tests=198\text{Total positive tests} = 198

    Correct

    Total positive tests=198+100,000×(1−0.00200)×(1−0.980)=198+1,996=2,194\text{Total positive tests} = 198 + 100{,}000 \times (1 - 0.00200) \times (1 - 0.980) = 198 + 1{,}996 = 2{,}194

    You assumed that only diseased individuals test positive. Because the specificity is 98.0%98.0\%, the false positive rate is 1−0.980=2.0%1 - 0.980 = 2.0\%. Across the 99,80099{,}800 healthy individuals, this produces 1,9961{,}996 false positives, making the total positive tests 198+1,996=2,194198 + 1{,}996 = 2{,}194.

  3. Step 3

    PPV = 198 / 198 = 1.00   ⟹  \implies PPV = 100%

    Correct

    PPV=1982,194≈0.0902=9.02%PPV = \frac{198}{2{,}194} \approx 0.0902 = 9.02\%

    You evaluated PPV=TPTotal PositivesPPV = \frac{TP}{\text{Total Positives}} using your denominator from Step 2, which led to an incorrect value of 100%100\%.

  4. Step 4

    FDR = 1 - 1.00 = 0   ⟹  \implies FDR = 0.00%

    Correct

    FDR=1−0.0902=0.9098≈91.0%FDR = 1 - 0.0902 = 0.9098 \approx 91.0\%

    You applied the complement rule FDR=1−PPVFDR = 1 - PPV correctly to your previous value, but the result was skewed by the earlier omission of false positives.

WHAT YOU DID WELL
Compliments from Your Teacher

You correctly identified and applied the disease prevalence and test sensitivity to find the true positive count.

You used the correct algebraic relationship FDR=1−PPVFDR = 1 - PPV to relate the false discovery rate to the positive predictive value.

THE ERROR
Where Your ACEpath Teacher Spotted the Mistake
Base-Rate Neglect / Omission of False Positives

no

WHY IT HAPPENED
ACEpath Teacher's Diagnosis

You treated all positive test results as originating strictly from affected individuals, overlooking the fact that a test with 98.0%98.0\% specificity still produces a 2.0%2.0\% false positive rate among healthy people.

REASONING PATH
Your ACEpath teacher’s suggested approach
Account for Both Sources of Positive Tests
  • Total positive tests always consist of True Positives (TPTP) plus False Positives (FPFP).

  • Always compute false positives by multiplying the healthy population N×(1−P(D))N \times (1 - P(D)) by the false positive rate (1−specificity)(1 - \text{specificity}) before forming the denominator.

Evaluate Base-Rate Influence on Predictive Value
  • When a disease is rare (P(D)≪1P(D) \ll 1), healthy individuals heavily outnumber diseased individuals, meaning even a small false positive rate can yield more false alarms than true positives.

  • Compare TPTP and FPFP directly in natural frequencies to verify that your PPVPPV denominator reflects the entire testing cohort.

CORRECTED PATH
ACEpath Teacher's Demonstration
  1. Compute healthy individuals: N×P(Dc)=100,000×(1−0.00200)=99,800N \times P(D^c) = 100{,}000 \times (1 - 0.00200) = 99{,}800.

  2. Calculate false positives: FP=99,800×(1−0.980)=99,800×0.020=1,996FP = 99{,}800 \times (1 - 0.980) = 99{,}800 \times 0.020 = 1{,}996.

  3. Find total positive tests: Total Positives=TP+FP=198+1,996=2,194\text{Total Positives} = TP + FP = 198 + 1{,}996 = 2{,}194.

  4. Calculate positive predictive value: PPV=1982,194≈0.0902=9.02%PPV = \frac{198}{2{,}194} \approx 0.0902 = 9.02\%.

  5. Calculate false discovery rate: FDR=100%−9.02%=90.98%≈91.0%FDR = 100\% - 9.02\% = 90.98\% \approx 91.0\%.

Final answer

(a) PPV≈9.02%PPV \approx 9.02\%; (b) FDR≈91.0%FDR \approx 91.0\%

NEXT STEPS
Final Takeaways from Your Teacher
  • Practice building a 2×22 \times 2 contingency table with natural frequencies for rare disease screening scenarios.

  • Double-check whether any positive results can come from the disease-free group whenever test specificity is less than 100%100\%.

  • Review how low population prevalence dramatically lowers positive predictive value despite high analytical sensitivity.

Lock it in

Similar question

01

A rare metabolic disorder has prevalence P(D)=0.010P(D) = 0.010. A screening test has sensitivity 0.9500.950 and specificity 0.9000.900. Calculate the Positive Predictive Value (PPVPPV) as a percentage rounded to three significant figures.

Answer

PPV=0.950×0.010(0.950×0.010)+(1−0.900)×0.990=0.00950.0095+0.0990=0.00950.1085≈8.76%PPV = \frac{0.950 \times 0.010}{(0.950 \times 0.010) + (1 - 0.900) \times 0.990} = \frac{0.0095}{0.0095 + 0.0990} = \frac{0.0095}{0.1085} \approx 8.76\%.

02

An oncogenomic panel screens for a mutation present in 0.50%0.50\% of the population. The assay has sensitivity 99.0%99.0\% and specificity 99.5%99.5\%. In a screening cohort of 200,000200{,}000 asymptomatic individuals, compute: (a) total true positives, (b) total false positives, and (c) the positive predictive value (PPVPPV).

Answer

(a) True positives: 200,000×0.0050×0.990=990200{,}000 \times 0.0050 \times 0.990 = 990. (b) False positives: 200,000×0.9950×(1−0.995)=199,000×0.005=995200{,}000 \times 0.9950 \times (1 - 0.995) = 199{,}000 \times 0.005 = 995. (c) PPV=990990+995=9901,985≈49.9%PPV = \frac{990}{990 + 995} = \frac{990}{1{,}985} \approx 49.9\%.

03

A clinical geneticist wants a screening test for a condition with prevalence P(D)=0.00100P(D) = 0.00100 to achieve a positive predictive value (PPVPPV) of at least 50.0%50.0\%. Assuming the test has perfect sensitivity (100%100\%), what is the minimum specificity required? Express as a percentage to three significant figures.

Answer

Set PPV=1.0×0.00100(1.0×0.00100)+(1−Spec)×0.999≥0.500PPV = \frac{1.0 \times 0.00100}{(1.0 \times 0.00100) + (1 - \text{Spec}) \times 0.999} \ge 0.500. This requires (1−Spec)×0.999≤0.00100  ⟹  1−Spec≤0.001000.999≈0.001001(1 - \text{Spec}) \times 0.999 \le 0.00100 \implies 1 - \text{Spec} \le \frac{0.00100}{0.999} \approx 0.001001. Thus, Spec≥1−0.001001=0.998999≈99.9%\text{Spec} \ge 1 - 0.001001 = 0.998999 \approx 99.9\%.

Loading workspace…