Ethics, Research Methodology, and Evidence-Based Medicine
What the exam tests
This domain tests consent, confidentiality, research ethics, levels of evidence, and interpretation of diagnostic and treatment studies.
Core rules
- Informed consent requires capacity, disclosure, and voluntariness; a patient may refuse even life-saving treatment.
- Research ethics: independent approval, informed consent, risk minimisation, and equitable selection. Children and pregnant women need extra safeguards.
- Evidence hierarchy: systematic reviews of randomised trials are highest for treatment; cohort studies for prognosis; case series lowest.
- Sensitivity detects disease; specificity excludes it. PPV depends on prevalence.
- NNT = 1/absolute risk reduction. Intention-to-treat analysis preserves randomisation.
- A randomised trial allocates by chance; blinding reduces performance and detection bias. A confidence interval crossing 1 for a risk ratio is not statistically significant at the 5% level. Audit measures practice against standards; research generates generalisable knowledge. Conflicts of interest must be declared.
Worked example
A screening test has sensitivity 90%, specificity 95%, and prevalence 1%. In 10,000 women: 90 true positives, 10 false negatives, 495 false positives, and 9,405 true negatives. PPV = 90/(90+495) = 15.4%. The reasoning: even a good test has a low PPV when prevalence is low, so confirmatory testing is essential. If prevalence were 10%, PPV would rise to about 67%.
Traps
Confusing sensitivity with PPV; reading a p-value as the size of effect; ignoring loss to follow-up; and using surrogate outcomes as if they were patient-important.