Residency · Residency · Family Medicine
Evidence-Based Medicine at the Bedside
Introduction
Evidence-based medicine (EBM) integrates the best available research evidence with clinical expertise and patient values to guide medical decision-making. For family physicians managing a broad scope of clinical problems, EBM skills are essential to avoid outdated practices, interpret new research critically, and provide high-quality care. This lecture focuses on the practical application of EBM in daily clinical encounters.
The Three Pillars of EBM
The three pillars of EBM are best available evidence (systematically gathered research from the medical literature), clinical expertise (the clinician's accumulated experience, education, and clinical skills), and patient values and preferences (the unique concerns, expectations, and circumstances each patient brings). All three pillars are equally important. Evidence alone does not make decisions; it informs them.
The Five Steps of EBM
Step 1: Ask a Clinical Question (PICO Format)
A well-structured clinical question uses the PICO format: Patient or Population (who is the patient or group?), Intervention (what treatment, test, or exposure is being considered?), Comparison (what is the alternative, such as placebo, standard care, or no treatment?), and Outcome (what is the desired or measured result?). For example: "In adults with mild hypertension (P), does lifestyle modification alone (I) compared with antihypertensive medication (C) reduce cardiovascular events at 5 years (O)?"
Step 2: Acquire the Evidence
Point-of-care resources such as UpToDate, DynaMed, and Essential Evidence Plus provide rapid answers. Filtered resources including Cochrane Reviews, ACP Journal Club, and Clinical Evidence offer pre-appraised evidence. Primary literature is accessed through PubMed and MEDLINE using clinical queries filters. Pre-appraised resources should be prioritized during busy clinical practice, with deep literature searches reserved for complex questions.
Step 3: Appraise the Evidence
Appraisal evaluates validity (internal validity), importance (clinical significance), and applicability (external validity). The hierarchy of evidence ranks study types: systematic reviews are strongest, followed by randomized controlled trials, cohort studies, case-control studies, case series, and expert opinion.
Step 4: Apply the Evidence
Application involves integrating findings with patient preferences and clinical context. Number needed to treat (NNT) and number needed to harm (NNH) are useful for communicating risk and benefit. Shared decision-making is the practical embodiment of applying evidence.
Step 5: Assess Performance
Reflection on whether the application of evidence improved outcomes is important. Clinical practice should be audited against evidence-based guidelines periodically.
Critical Appraisal Essentials
Evaluating Therapy Studies (RCTs)
Key questions when appraising an RCT include whether there was adequate randomization and allocation concealment, whether groups were similar at baseline, whether blinding was used for patients, clinicians, and outcome assessors, whether follow-up was complete (greater than 80%), whether outcomes were analyzed by intention-to-treat, and whether the treatment effect was clinically (not just statistically) significant.
Understanding Key Statistical Concepts
Absolute risk reduction (ARR) is the difference in event rates between groups. Relative risk reduction (RRR) is the proportional reduction and can be misleading when baseline risk is low. Number needed to treat (NNT), calculated as 1 divided by the ARR, tells how many patients must be treated for one to benefit. The confidence interval (CI) represents the range within which the true effect likely falls; if the CI for relative risk crosses 1.0, the result is not statistically significant. The p-value represents the probability the result occurred by chance, with p less than 0.05 being the conventional but arbitrary threshold.
Common Biases
Selection bias results from non-random group assignment. Performance bias arises from differential treatment beyond the intervention. Attrition bias occurs from differential loss to follow-up. Detection or measurement bias results from non-blinded outcome assessment. Publication bias reflects the tendency for positive results to be published more readily.
Applying EBM in Primary Care
Screening Tests
Screening tests are evaluated using sensitivity, specificity, and positive and negative predictive values. PPV and NPV depend on disease prevalence (pre-test probability). The Wilson and Jungner criteria define appropriate screening programs. Not every available test should be ordered, as screening can cause harm through false positives, overdiagnosis, and anxiety.
Clinical Decision Rules
Validated clinical decision rules quantify probability and guide decisions. Examples include the Ottawa Ankle Rules, Wells Score, and CHA2DS2-VASc. These tools are most useful when they help safely reduce testing while maintaining safety. Rules should only be applied to populations similar to their derivation and validation cohorts.
Guidelines: Trustworthy vs. Not
Trustworthy guidelines are transparent, evidence-graded (using systems such as GRADE), and manage conflicts of interest. The GRADE system rates evidence quality as high, moderate, low, or very low and separates evidence quality from recommendation strength. Guidelines funded entirely by industry or based predominantly on expert opinion should be viewed with caution.
Communicating Evidence to Patients
Absolute numbers should be used rather than relative risk: "2 out of 100 instead of 4 out of 100" communicates more clearly than "50% reduction." Visual aids such as icon arrays and frequency diagrams improve patient comprehension. Uncertainty should be acknowledged honestly, as patients respect transparency. Information should be framed in terms of both benefits and harms.
Key Clinical Pearls
Asking answerable questions is fundamental: a well-formed PICO question is the foundation of efficient EBM practice. Relative risk reductions are inherently misleading without context, so the absolute risk reduction and NNT should always be calculated or communicated. Pre-appraised resources such as UpToDate and Cochrane should be used during clinical encounters, with primary literature review reserved for unresolved questions. A statistically significant result is not always clinically significant, and effect size and clinical relevance must be evaluated. EBM is not cookbook medicine: it explicitly incorporates clinical judgment and patient values into every decision.
References
- Straus, S. E., et al. (2019). Evidence-Based Medicine: How to Practice and Teach EBM (5th ed.). Elsevier.
- Guyatt, G. H., et al. (2008). GRADE: An emerging consensus on rating quality of evidence and strength of recommendations. BMJ, 336(7650), 924-926.
- Sackett, D. L., et al. (1996). Evidence-based medicine: What it is and what it isn't. BMJ, 312(7023), 71-72.
- Greenhalgh, T. (2019). How to Read a Paper: The Basics of Evidence-Based Medicine and Healthcare (6th ed.). Wiley-Blackwell.