Clinical Trials and Evidence-Based Medicine
Clinical trials are structured research studies evaluating interventions in humans. Evidence-based medicine integrates best research evidence with clinical expertise and patient values.
Key Facts
Phase I trials assess safety and dosing in healthy volunteers (20-100 participants); Phase II assess efficacy; Phase III are large RCTs comparing to standard treatment Randomisation minimises confounding and selection bias; allocation concealment prevents foreknowledge of group assignment Double-blinding reduces observer and performance bias — neither participant nor investigator knows allocation Intention-to-treat (ITT) analysis includes all randomised participants regardless of compliance — preserves randomisation Per-protocol analysis includes only those who completed the study as planned — may overestimate treatment effect CONSORT statement provides reporting guidelines for RCTs — includes flow diagram of participant progress Equipoise is the ethical requirement that genuine uncertainty exists about which treatment is superior before conducting an RCT GCP (Good Clinical Practice) guidelines ensure trials are ethically and scientifically sound — required by MHRA
Overview
Key Facts
Clinical trials are the cornerstone of evidence-based medicine, providing the most reliable evidence for treatment efficacy and safety. The UK has a strong clinical trials infrastructure with the NIHR coordinating much of the research activity.
Epidemiology
Approximately 600,000 people participate in clinical trials in the UK each year. The NIHR supports over 3,500 active studies. UK clinical trials legislation is governed by the Medicines for Human Use (Clinical Trials) Regulations 2004, overseen by the MHRA.
Aetiology
Trial phases:
- Phase I: First-in-human, safety, pharmacokinetics, dose-finding (20-100 healthy volunteers)
- Phase II: Efficacy and side effects in target population (100-300 patients)
- Phase III: Large-scale RCT comparing to standard care or placebo (300-3,000+ patients)
- Phase IV: Post-marketing surveillance, rare side effects, long-term outcomes
Pathophysiology
Key methodological concepts:
- Randomisation methods: Simple, block, stratified, minimisation
- Blinding levels: Open-label, single-blind, double-blind, triple-blind
- Control types: Placebo, active comparator, standard care
- Endpoints: Primary (main outcome), secondary, surrogate (biomarker proxy), composite
- Non-inferiority trials: Show new treatment is not unacceptably worse than standard (pre-specified margin)
- Adaptive designs: Allow modifications based on interim data (e.g., RECOVERY trial for COVID-19 treatments)
Clinical Presentation
Applying Evidence to Clinical Practice
- Ask: Formulate a clinical question (PICO: Population, Intervention, Comparator, Outcome)
- Acquire: Search for best available evidence (PubMed, Cochrane, NICE)
- Appraise: Critically evaluate validity, results, and applicability
- Apply: Integrate evidence with clinical expertise and patient preferences
- Assess: Audit outcomes of clinical decisions
Ethical Considerations
- Informed consent — voluntary, informed, competent
- Research Ethics Committee (REC) approval required
- Declaration of Helsinki principles
- Data Safety Monitoring Board (DSMB) reviews interim data for safety
Red Flags in Trial Interpretation
- Unblinded trials with subjective outcomes — high risk of performance bias
- Large loss to follow-up (>20%) — threatens validity
- Surrogate endpoints without validated link to clinical outcomes
- Selective outcome reporting — check trial registry (ClinicalTrials.gov) for pre-specified outcomes
Differential Diagnosis
| Study Design | Strengths | Limitations |
|---|---|---|
| RCT | Minimises confounding, highest internal validity | Expensive, may lack external validity, ethical constraints |
| Cohort study | Can study rare exposures, temporal relationship | Confounding, loss to follow-up, expensive if prospective |
| Case-control | Efficient for rare diseases, quick and cheap | Recall bias, cannot calculate incidence/RR directly |
| Cross-sectional | Quick, cheap, estimates prevalence | Cannot determine causation, temporal ambiguity |
| Systematic review/meta-analysis | Pools evidence, increases power | Publication bias, heterogeneity, garbage in-garbage out |
| Qualitative research | Explores experiences and perspectives | Not generalisable, subjective interpretation |
Diagnosis / Investigation
Critical Appraisal Framework (CASP)
- Validity: Was randomisation adequate? Was blinding maintained? Were groups similar at baseline?
- Results: What is the effect size (RR, OR, NNT)? How precise (confidence intervals)?
- Applicability: Are patients similar to mine? Are all important outcomes considered? Are benefits worth harms and costs?
Key Statistical Measures
- Absolute risk reduction (ARR): Difference in event rates between groups
- Relative risk reduction (RRR): ARR ÷ control event rate
- Number needed to treat (NNT): 1/ARR
- Hazard ratio: Instantaneous risk ratio (from Cox regression, survival analysis)
- Forest plot: Visual display of individual study and pooled results in meta-analysis
- Funnel plot: Detects publication bias in meta-analyses (asymmetry suggests bias)
Regulatory Framework
- MHRA approval for investigational medicinal products
- Health Research Authority (HRA) approval
- Research Ethics Committee (REC) review
- Sponsor responsibilities (university, NHS trust, or pharmaceutical company)
Management
Non-pharmacological
- Shared decision-making using evidence summaries (NICE patient decision aids)
- Clinical guidelines (NICE, SIGN, RCGP) synthesise evidence for practice
- Clinical audit measures practice against evidence-based standards
Pharmacological
Landmark UK Trials (examples):
- RECOVERY trial: Dexamethasone reduces COVID-19 mortality in ventilated patients by 36%
- UKPDS: Intensive glucose control reduces microvascular complications in T2DM
- ISIS-2: Aspirin + streptokinase reduces MI mortality
- 4S trial: Simvastatin reduces CV events in hypercholesterolaemia
- NICE guidance is based on systematic review of clinical and cost-effectiveness evidence
Referral Criteria
- Consider referring suitable patients to clinical trials — all NHS trusts should facilitate research participation
- Research nurse support available through NIHR Clinical Research Network
- Patient and public involvement (PPI) is now mandatory in NIHR-funded research
Prognosis
- Evidence-based medicine has improved outcomes across all medical specialties
- Implementation of trial evidence into guidelines typically takes 17 years on average — a key challenge
- The RECOVERY trial demonstrated the value of large, simple adaptive platform trials during pandemics
- Approximately 50% of treatments used in clinical practice have been tested in RCTs
- UK clinical trials contribute significantly to global medical knowledge through the NIHR infrastructure
Other Relevant Information
Hierarchy of Evidence
| Level | Study Type | Strength |
|---|---|---|
| 1a | Systematic review of RCTs | Highest |
| 1b | Individual RCT with narrow CI | High |
| 2a | Systematic review of cohort studies | Moderate-high |
| 2b | Individual cohort study | Moderate |
| 3 | Case-control study | Moderate-low |
| 4 | Case series | Low |
| 5 | Expert opinion | Lowest |
Common Biases in Clinical Research
| Bias | Definition | Prevention |
|---|---|---|
| Selection | Non-random group allocation | Randomisation, allocation concealment |
| Performance | Differential treatment of groups | Blinding |
| Detection | Differential outcome assessment | Blinding of assessors |
| Attrition | Differential loss to follow-up | ITT analysis, minimise dropouts |
| Reporting | Selective outcome reporting | Pre-registration, CONSORT |
| Publication | Positive results preferentially published | Trial registries, funnel plots |