Research Methodology, Biostatistics, and Evidence-Based Practice
This section tests whether you can frame a clinical question, appraise study design, interpret statistics, and apply evidence without overclaiming.
Core rules and concepts
- PICO: population, intervention, comparison, outcome.
- Study designs: randomised controlled trial for effectiveness; cohort for prognosis; case–control for rare exposures; cross-sectional for prevalence; diagnostic accuracy study for tests.
- Bias includes selection, measurement, recall, and confounding. Intention-to-treat analysis preserves randomisation.
- Sensitivity detects disease; specificity detects absence. Positive predictive value depends on prevalence.
- A p-value is the probability of the observed result, or a more extreme one, if the null hypothesis is true; it is not the probability that the null is true.
Worked example
A test has sensitivity 90% and specificity 80% in a population with 10% prevalence. Of 1,000 people, 100 have disease; 90 are true positives and 10 are false negatives. Of 900 without disease, 720 are true negatives and 180 are false positives. PPV = 90 ÷ (90 + 180) = 33%. Reasoning: even a good test has low positive predictive value when prevalence is low, so a positive result needs confirmation.
Traps that cost marks
- Treating a p-value as the probability that the hypothesis is true.
- Confusing sensitivity with positive predictive value.
- Assuming correlation proves causation.
- Ignoring confidence intervals and relying only on statistical significance.