Evidence-based practice, statistics and quality improvement
What the exam asks
These questions test whether you can appraise a paper, interpret a diagnostic test, and distinguish audit, research and quality improvement.
Core rules
- Study design hierarchy: systematic review and meta-analysis, then RCT, cohort, case-control, cross-sectional, case series. Randomisation and allocation concealment control selection bias; blinding controls ascertainment bias; analyse by intention to treat.
- Sensitivity and specificity are properties of the test; positive predictive value depends on prevalence. NNT = 1/absolute risk reduction.
- A p value is the probability of the data given the null hypothesis, not the probability that the null is true; confidence intervals convey precision and the size of effect.
- Screening needs the Wilson and Jungner criteria: important condition, accepted treatment, adequate test, understood natural history, cost-effective.
- Quality improvement uses the Model for Improvement (aim, measures, changes) with PDSA cycles and run charts; a shift of six or more points suggests special-cause variation.
- Audit compares practice with a standard and re-audits; research generates generalisable new knowledge and needs ethics approval.
Worked example
A test has sensitivity 90% and specificity 95%; prevalence is 1%. In 10,000 people, 100 have disease and 90 are detected; 9,900 are well and 495 are false positives. PPV = 90/(90+495) = 15%. Most positives are false, so confirmatory testing is essential before treatment.
Traps
Quoting relative risk reduction alone; assuming a small p value means a large or important effect; confusing a run chart with a control chart; calling a QI project research.