Practice-Based Learning and Improvement
Core concepts
This domain asks whether you can appraise evidence and improve your own practice. Candidates should understand study design (RCT, cohort, case-control), measures of effect (RR, OR, HR, NNT), and sources of bias (selection, lead-time, publication). In hematology, landmark trial designs-non-inferiority in anticoagulation, superiority in CAR-T-require attention to endpoints and confidence intervals.
Quality improvement frameworks (PDSA cycles, root cause analysis, morbidity and mortality review) and guideline adherence (ASH, NCCN) are commonly tested. Recognize that guidelines are tools, not mandates, and that deviation requires documented rationale.
Traps
- Interpreting a non-inferiority trial as showing equivalence without checking the prespecified margin.
- Confusing statistical significance with clinical importance in large trials.
- Ignoring lead-time bias when screening improves survival without changing mortality.
- Assuming a surrogate endpoint (e.g., response rate) guarantees an overall survival benefit.
Worked example
A trial reports a hazard ratio of 0.92 (95% CI 0.80–1.06) for overall survival: the CI crosses 1, so no statistically significant benefit; do not claim efficacy. Check whether the primary endpoint was met and whether the confidence interval excludes the non-inferiority margin.
Rule of thumb
Before acting on a trial, ask: what was the primary endpoint, what was the comparator, and does the confidence interval exclude the clinically meaningful threshold?