o Starts with exposure → follows over time to see outcome.
o Measures relative risk (RR).
o Prospective or retrospective.
o Starts with outcome → looks back for exposure.
o Measures odds ratio (OR).
o Useful for rare diseases.
o Measures exposure and outcome at a single point in time.
o Provides prevalence data.
o Participants randomly allocated to intervention or control.
o Measures efficacy, minimizes confounders.
• Selection bias: differences in baseline characteristics between groups (e.g., healthy volunteer effect).
• Recall bias: cases recall exposure differently than controls (common in case-control studies).
• Measurement (information) bias: systematic errors in data collection (e.g., poorly calibrated instrument).
• Publication bias: positive studies more likely to be published → affects meta-analyses.
• Incidence: new cases over a specified time period / population at risk.
• Prevalence: total cases (new + existing) at a point in time / population.
• Relative risk (RR): risk of outcome in exposed / unexposed.
• Odds ratio (OR): odds of exposure among cases / controls; approximates RR when disease is rare.
Metric Meaning Formula
Sensitivity Proportion of true positives detected TP / (TP + FN)
Specificity Proportion of true negatives correctly identified TN / (TN + FP)
PPV Probability disease present if test positive TP / (TP + FP)
NPV Probability no disease if test negative TN / (TN + FN)
Sensitivity and specificity are intrinsic to the test.
PPV and NPV depend on disease prevalence.
• p-value: probability that observed result is due to chance if null hypothesis true.
o p < 0.05 → typically considered significant.
• Confidence interval (CI):
o Range in which true parameter is likely to lie (usually 95% CI).
o If CI for RR or OR does not include 1 → statistically significant.
• Forest plot:
o Shows individual study effects and pooled effect.
o Squares = individual study effect size; line = CI.
o Diamond = pooled effect estimate.
• Funnel plot:
o Assesses publication bias.
o Symmetry suggests low bias; asymmetry suggests potential publication bias.
Extra Revision Pearls
• Number needed to treat (NNT) = 1 / absolute risk reduction (ARR).
• Lead-time bias: apparent survival benefit due to earlier detection, not actual improvement.
• Length-time bias: screening more likely to detect slowly progressing disease.
• Intention-to-treat analysis: includes all randomised participants; preserves randomisation.
————————————————————————————————————————————————————————————————————————————————————————————————————————-
Author & Educational Disclaimer
Author:
Dr Phillip Cockrell BM FRCP DipClinEd
Dr Phillip Cockrell is a UK Consultant Physician in Internal Medicine, currently working at Queen Alexandra Hospital, Portsmouth University Hospitals NHS Trust. He has previously worked as a registrar across Intensive Care Medicine, Gastroenterology, Cardiology, Stroke Medicine, Acute Medicine, and Respiratory Medicine.
He has held senior leadership roles including Associate Clinical Director of the Acute Medical Unit, Clinical Director of Internal Medicine, and Chief of Medicine. Dr Cockrell has over 15 years’ experience in postgraduate medical education, having lectured extensively across the MRCP syllabus and contributed to MRCP revision teaching and course development.
Dr Cockrell holds a Bachelor of Medicine (BM), Fellowship of the Royal College of Physicians (FRCP), and a Diploma in Clinical Education (DipClinEd). His teaching approach is based on structured consolidation of complex medical topics to support efficient and effective revision for postgraduate examinations.
Purpose of this content:
The material on this page is intended solely for educational purposes to support revision for the MRCP (UK) Part 1 examination. It reflects examination-relevant principles of internal medicine and is designed to aid learning and pattern recognition.
Medical disclaimer:
This content is designed for postgraduate medical examination revision and does not constitute medical advice, diagnosis, or treatment guidance and must not be used as a substitute for professional clinical judgement, local guidelines, or specialist consultation. Clinical decisions should always be made in the context of individual patient circumstances and current national guidance.
