Research Methodology, Biostatistics, and Evidence-Based Practice
Research Methodology, Biostatistics, and Evidence-Based Practice
This domain tests appraisal of study design, bias, diagnostic statistics, and clinical guidelines. The exam expects you to interpret results rather than memorize formulas alone.
Core rules
- Randomized trials reduce selection bias; intention-to-treat analysis preserves randomization.
- Sensitivity detects disease; specificity detects absence of disease; predictive values depend on prevalence.
- A p-value indicates compatibility with the null hypothesis, not the size or importance of an effect.
- Confidence intervals show precision and range; wide intervals mean uncertainty.
- Evidence-based practice combines best evidence, clinical expertise, and patient values.
Worked example
A test has 90% sensitivity and 80% specificity; prevalence is 10%. Reasoning: in 1,000 people, 100 have disease and 90 test positive; 900 are healthy and 180 falsely test positive. Positive predictive value = 90 ÷ (90 + 180) = 33%. A positive result does not mean a 90% chance of disease when prevalence is low.
Traps
Equating statistical significance with clinical importance; ignoring prevalence; confusing sensitivity with predictive value; selective reporting; applying trial results to excluded patients.