Research, Evidence-Based Practice, and Quality Improvement
What this domain tests
Critical appraisal of plastic surgery literature, correct interpretation of diagnostic and statistical measures, and the difference between research and quality improvement.
Evidence and study design
- Levels of evidence: systematic review of randomised trials at level 1, individual randomised trial level 2, cohort level 3, case-control and case series level 4, expert opinion level 5.
- Appraise randomised trials for randomisation, allocation concealment, blinding, intention-to-treat analysis, sample size and power, and effect size with confidence intervals.
- A p value is the probability of the observed data (or more extreme) given the null hypothesis - not the probability that the null is true. A confidence interval for a ratio crossing 1, or for a difference crossing 0, means no significant difference.
- Absolute risk reduction drives number needed to treat = 1/ARR. Odds ratios overstate risk when events are common; report risk ratios when possible.
- Meta-analysis: heterogeneity measured by I-squared, fixed versus random effects, funnel plots for publication bias. Reporting standards: CONSORT for trials, STROBE for observational studies, PRISMA for reviews, SQUIRE for improvement work.
- Patient-reported outcomes relevant here include BREAST-Q, FACE-Q and DASH, alongside scar scales such as POSAS and the Vancouver Scar Scale.
Quality improvement
QI tests a change in local practice with PDSA cycles and run charts; research generates generalisable new knowledge and needs ethics approval and consent. Distinguish structure, process and outcome measures, and recognise special-cause variation on a control chart.
Worked example
A new scar assessment tool has 90% sensitivity and 95% specificity; prevalence of problematic scarring in the clinic is 1%. Reasoning with 1000 patients: 10 have the problem, 9 are correctly detected; among 990 without it, 5% (about 50) are false positives. Positive predictive value = 9/59 ≈ 15%. So a positive result in a low-prevalence setting mostly means false positives, and the test cannot replace clinical assessment - a favourite examination twist.
Traps
- Interpreting a non-significant result in an underpowered study as proof of no difference.
- Confusing sensitivity with positive predictive value.
- Treating a surrogate endpoint as a patient-important outcome.
- Describing a local audit as research, or vice versa.
- Quoting relative risk reduction without the absolute figure.