Epidemiology and Biostatistics
Epidemiology and biostatistics tests whether you can interpret rates, study designs, diagnostic tests and causal claims. The exam often presents a short scenario or 2×2 table and asks for the measure or the main bias.
Core rules
- Incidence is new cases per population at risk per time; prevalence is all existing cases at a point or period.
- Sensitivity = TP/(TP+FN); specificity = TN/(TN+FP); PPV = TP/(TP+FP). PPV rises with prevalence.
- Cohort studies estimate incidence and relative risk; case-control studies estimate odds ratios; cross-sectional studies estimate prevalence but not temporality.
- Confounding is a third factor associated with exposure and outcome; randomisation or adjustment addresses it, matching does not guarantee control.
- Confidence intervals show precision; a p-value is not the probability that the null is true.
Worked example
In 1,000 people, 100 have disease. A test has 90% sensitivity and 95% specificity. True positives = 90. False positives = 5% of 900 = 45. PPV = 90/(90+45) = 66.7%. The reasoning: most positive results are true, but one-third are false, so a positive screening test needs confirmation. If prevalence fell, PPV would fall even with the same sensitivity and specificity.
Traps
Calling a small p-value “clinically important”; confusing relative risk reduction with absolute risk reduction; using a case-control study to calculate incidence; and ignoring selection or recall bias.