Research Methodology, Biostatistics, and Evidence-Based Practice
Core concepts
Research methodology asks whether a study is valid, and biostatistics asks how certain we are. Study designs: randomized controlled trials minimize confounding; cohort studies measure incidence; case–control studies suit rare outcomes; cross-sectional studies measure prevalence. Bias includes selection, information, and confounding. Randomization and allocation concealment protect against selection bias; blinding protects against measurement bias; intention-to-treat preserves prognosis. Sensitivity is the proportion of diseased people who test positive; specificity is the proportion of non-diseased who test negative. Positive predictive value depends on prevalence. Confidence intervals describe precision; a p-value is the probability of data as extreme or more extreme if the null hypothesis is true, not the probability that the null is true. Number needed to treat is the reciprocal of absolute risk reduction. Evidence-based practice integrates best evidence, clinical expertise, and patient values.
Traps
Do not interpret a small p-value as a large effect. Do not compare PPV across populations with different prevalence. Do not confuse statistical significance with clinical importance. Beware surrogate outcomes and composite endpoints.
Worked example
A test has sensitivity 90% and specificity 95%. In a population with 1% prevalence, PPV = (0.90 × 0.01) / [(0.90 × 0.01) + (0.05 × 0.99)] ≈ 15%. Most positives are false positives despite good sensitivity and specificity.
Rule of thumb: sensitivity and specificity are properties of the test; PPV and NPV are properties of the test in a population.
Ethics, Professionalism, Patient Safety, and Health Care Quality
Core concepts
Ethics and professionalism govern clinical decisions and system safety. The four principles are autonomy, beneficence, non-maleficence, and justice. Informed consent requires capacity, disclosure, understanding, and voluntariness. Confidentiality may be breached only when required by law or to prevent serious harm. In emergencies, treat under the doctrine of necessity when consent cannot be obtained. Research ethics require institutional review, informed consent, and protection of vulnerable populations. Patient safety focuses on system failures rather than individual blame: handoffs, medication reconciliation, surgical safety checklists, and root cause analysis. Quality improvement uses measurement, feedback, and iterative change. Never events include wrong-site surgery and retained instruments.
Traps
Do not confuse consent with a signature; the discussion matters. Do not assume a pregnant woman’s refusal of treatment can be overridden for fetal benefit; maternal autonomy generally prevails when she has capacity. Disclose harmful errors promptly and honestly.
Worked example
A patient receives the wrong dose of a medication and develops hypoglycemia. The response includes immediate treatment, disclosure to the patient, reporting through the incident system, and a root cause analysis focused on prescribing and administration systems-not public blame.
Rule of thumb: when an error occurs, ask what the system allowed, then fix the system and support the patient.