Epidemiology and Biostatistics
This domain tests whether you can read a paper and a population table critically: measures of frequency and association, study designs, bias, validity of screening tests and basic inference. Questions are usually short calculations with one conceptual trap.
Core rules
- Frequency: incidence measures new cases per population at risk per time; prevalence measures all existing cases at a point or period. Prevalence ≈ incidence × duration for stable diseases.
- Association: relative risk from cohort studies; odds ratio from case-control studies (approximates RR when the disease is rare); attributable risk and population attributable fraction for impact.
- Effect size in trials: absolute risk reduction = control rate minus intervention rate; number needed to treat = 1/ARR.
- Validity of a test: sensitivity = true positives/(true positives + false negatives); specificity = true negatives/(true negatives + false positives); positive predictive value depends on prevalence, unlike sensitivity and specificity.
- Bias: selection, information/recall, confounding (a third factor linked to both exposure and outcome, handled by randomisation, restriction, matching or adjustment), lead-time bias and length bias in screening.
- Inference: 95% confidence interval expresses precision; if a CI for a difference includes 1 for ratios or 0 for differences, the result is not statistically significant at 5%. A p-value is not the probability that the hypothesis is true.
Worked example
A fasting glucose test is applied to 1,000 people. Of 100 with diabetes, 90 test positive; of 900 without, 90 test positive. Sensitivity = 90/100 = 90%. Specificity = 810/900 = 90%. PPV = 90/180 = 50%. Now suppose the same test is used where prevalence is 1%: of 1,000 people, 10 have disease (9 detected) and 990 do not (99 false positives), so PPV = 9/108 ≈ 8%. Reasoning: sensitivity and specificity are unchanged, but PPV collapses at low prevalence, which is why screening programmes target high-risk groups.
Traps that cost marks
- Reporting PPV as if it were a fixed property of the test.
- Using an odds ratio from a case-control study to claim incidence.
- Confusing statistical significance with clinical importance.