Research Methods, Statistics, and Evidence-Based Practice
Core concepts
Understand study design: randomized controlled trials, cohort studies, case-control studies, and cross-sectional studies, and their strengths and limitations. Know measures of association: relative risk, odds ratio, absolute risk reduction, number needed to treat, and their interpretation. Understand sensitivity, specificity, positive and negative predictive value, and how prevalence affects predictive values. Know confidence intervals and p-values and their limitations. Understand bias (selection, information, confounding) and confounding control (randomization, restriction, matching, stratification, multivariable analysis).
Understand systematic reviews and meta-analysis, including heterogeneity and publication bias. Know the principles of clinical practice guidelines and grading of evidence. Be familiar with intention-to-treat analysis and per-protocol analysis.
Traps
- Confusing relative risk with absolute risk; a large relative risk may be a small absolute effect.
- Assuming a p-value indicates the size or importance of an effect.
- Forgetting that predictive values depend on prevalence.
- Misinterpreting odds ratios as relative risks in high-prevalence settings.
- Overlooking that intention-to-treat analysis preserves randomization benefits.
Rule of thumb
When reading a study, ask: what is the design, what is the effect size, what is the precision, and what are the threats to validity? Then decide whether the result applies to your patient.
Worked example: A trial reports a relative risk reduction of 50% for a rare outcome. Calculate the absolute risk reduction and number needed to treat before concluding the intervention is clinically meaningful.