EDAIC Statistics and Evidence-Based Medicine — Questions and Answers
Question 1: A screening test has a sensitivity of 90% and a specificity of 85%. In a population where the disease prevalence is 2%, what is the approximate positive predictive value (PPV)?
- 90%
- 85%
- 11% (Correct answer)
- 50%
Correct answer: 11%
PPV = (sensitivity × prevalence) / [(sensitivity × prevalence) + (1 − specificity) × (1 − prevalence)] = (0.90 × 0.02) / [(0.90 × 0.02) + (0.15 × 0.98)] = 0.018 / (0.018 + 0.147) ≈ 11%. At low disease prevalence, even a highly sensitive and specific test yields a low PPV, meaning most positive results are false positives.
Question 2: A randomised controlled trial reports 30-day mortality of 10% in the treatment group and 20% in the control group. What is the Number Needed to Treat (NNT)?
- 2
- 5
- 10 (Correct answer)
- 20
Correct answer: 10
NNT = 1 / Absolute Risk Reduction (ARR). ARR = 20% − 10% = 10% = 0.10. NNT = 1 / 0.10 = 10. This means 10 patients must be treated to prevent one additional death. NNT is more clinically meaningful than relative risk reduction alone.
Question 3: A study reports a 95% confidence interval for the mean difference in arterial pressure between two groups as −2 to +8 mmHg. What is the most appropriate interpretation?
- The result is statistically significant at p < 0.05
- The result is not statistically significant at p < 0.05 because the interval includes zero (Correct answer)
- The treatment definitely has no effect
- A larger sample would guarantee significance
Correct answer: The result is not statistically significant at p < 0.05 because the interval includes zero
A 95% CI that crosses zero (the null value for a difference) means we cannot exclude no effect, which corresponds to p > 0.05 — the result is not statistically significant. Note this does not prove the treatment has no effect; the study may simply be underpowered. Clinical significance (is a difference of 8 mmHg important?) is a separate question from statistical significance.
Question 4: Which study design provides the highest level of evidence for the efficacy of a therapeutic intervention?
- Prospective cohort study
- Randomised controlled trial (RCT)
- Systematic review with meta-analysis of multiple RCTs (Correct answer)
- Case-control study
Correct answer: Systematic review with meta-analysis of multiple RCTs
The evidence hierarchy places systematic reviews and meta-analyses of high-quality RCTs at the top, because they pool data from multiple randomised trials, maximising statistical power and reducing random error. A single RCT ranks below a well-conducted systematic review; cohort and case-control studies are observational and more prone to confounding.
Question 5: In a diagnostic accuracy study, the area under the ROC curve (AUC) is reported as 0.50. How should this be interpreted?
- The test is perfect
- The test correctly identifies 50% of cases
- The test performs no better than chance (random guessing) (Correct answer)
- The test has 50% sensitivity and 50% specificity
Correct answer: The test performs no better than chance (random guessing)
An AUC of 0.50 corresponds to the diagonal line of the ROC curve, meaning the test discriminates no better than flipping a coin — it has no diagnostic value. An AUC of 1.0 is a perfect test; AUC ≥ 0.70 is generally considered acceptable. AUC represents the probability that the test correctly ranks a randomly chosen diseased individual above a randomly chosen non-diseased individual.
Question 6: A p-value of 0.04 is reported in a clinical trial. Which statement is the correct interpretation?
- There is a 4% probability that the null hypothesis is true
- There is a 96% probability that the alternative hypothesis is true
- There is a 4% probability of obtaining results as extreme as observed (or more so) if the null hypothesis were true (Correct answer)
- The treatment effect is clinically important
Correct answer: There is a 4% probability of obtaining results as extreme as observed (or more so) if the null hypothesis were true
A p-value is the probability of observing a result as extreme as (or more extreme than) the study data, assuming the null hypothesis is true. It does NOT give the probability that the null hypothesis is true or false. A small p-value means the data are unlikely under the null hypothesis, but does not tell us about clinical significance or effect size — a large study can produce tiny p-values for clinically trivial differences.
EDAIC Statistics and Evidence-Based Medicine
A screening test has a sensitivity of 90% and a specificity of 85%.
In a population where the disease prevalence is 2%, what is the approximate positive predictive value (PPV)?