Research Methodology, Evidence-Based Practice, and Ethics
Research Methodology, Evidence-Based Practice, and Ethics
This section tests appraisal skills and professional conduct - reading a paper's design, numbers and integrity correctly.
Core rules
- Hierarchy: systematic reviews/meta-analyses of RCTs > RCTs > cohort > case-control > cross-sectional > case reports. Design alone does not guarantee quality; appraise risk of bias (randomization, allocation concealment, blinding, completeness).
- Confounding differs from bias: a confounder is associated with both exposure and outcome.
- Diagnostic performance: sensitivity, specificity, predictive values. PPV depends on prevalence - a highly specific test in a low-prevalence population still yields many false positives. Likelihood ratios are prevalence-independent.
- Statistics: p-values indicate compatibility with the null, not effect size; report confidence intervals and clinically meaningful thresholds; distinguish statistical from clinical significance.
- Ethics: informed consent (capacity, disclosure, voluntariness), confidentiality, conflicts of interest, ICMJE authorship criteria, IRB approval and the Declaration of Helsinki for research, no fabrication or plagiarism, and referral within one's scope of competence.
Worked example
A paper reports a salivary test for periodontitis with sensitivity 0.85 and specificity 0.90. In a screening population with 5% prevalence, PPV = (0.85 x 0.05) / ((0.85 x 0.05) + (0.10 x 0.95)) ≈ 0.31 - most positives are false. The test is unsuitable as a stand-alone screen despite impressive sensitivity and specificity.
Traps
Equating sensitivity with accuracy; reading a p-value as the probability the hypothesis is true; assuming one small trial is practice-changing; ignoring industry funding and publication bias; treating consent as a signature rather than a process.