Research, Evidence-Based Practice, and Professional Ethics
What this domain tests
This domain tests appraisal of study design, bias, statistics, and ethical conduct in dietetics research and practice.
Core rules
- Evidence hierarchy: systematic reviews and randomized controlled trials rank above cohort and case-control studies; expert opinion is weakest.
- Randomization reduces selection bias; allocation concealment protects the sequence; blinding reduces performance and detection bias. Intention-to-treat analysis preserves randomization.
- Confounding is a third variable associated with exposure and outcome. Adjust in design or analysis; do not infer causation from association.
- P-values indicate compatibility with the null hypothesis, not effect size. Report confidence intervals and absolute effects. NNT = 1 ÷ absolute risk reduction.
- Sensitivity is the proportion of true positives detected; specificity is the proportion of true negatives detected. Predictive values depend on prevalence.
- Ethics: autonomy, beneficence, nonmaleficence, and justice. Obtain informed consent, protect confidentiality, and declare conflicts of interest.
Worked example
A trial reports 20% relative risk reduction, from 20% to 16% absolute risk. Absolute risk reduction = 4%, or 0.04. NNT = 1 ÷ 0.04 = 25. The relative figure sounds larger than the absolute benefit; report both. If the 95% confidence interval for ARR includes zero, the result is compatible with no effect.
Traps
Treating statistical significance as clinical importance; ignoring attrition; using surrogate outcomes such as cholesterol as if they were hard outcomes; confusing sensitivity with positive predictive value; failing to declare industry funding; quoting relative risk without baseline risk.