Study Design
Understanding study design is fundamental to evidence-based medicine, with each design having characteristic strengths and limitations that determine its position in the hierarchy of evidence and its applicability to clinical questions.
Key Facts
Hierarchy of evidence (highest to lowest): systematic reviews/meta-analyses → RCTs → cohort studies → case-control studies → cross-sectional → case reports → expert opinion RCTs are the gold standard for evaluating interventions; randomisation minimises confounding Cohort studies follow groups over time; prospective or retrospective; measure incidence and relative risk Case-control studies compare exposure history in cases vs controls; efficient for rare diseases; measure odds ratio Cross-sectional studies measure exposure and outcome simultaneously; measure prevalence; cannot determine causality Ecological studies use population-level data; subject to the ecological fallacy (cannot infer individual-level associations) CONSORT (RCTs), STROBE (observational), PRISMA (systematic reviews) are key reporting standards Intention-to-treat (ITT) analysis preserves the benefit of randomisation by analysing participants in their allocated groups
Overview
Key Facts
Study design selection depends on the research question, available resources, ethical considerations, and disease frequency. Each design occupies a position in the evidence hierarchy and has characteristic strengths, limitations, and measures of association.
Experimental Studies
- Randomised controlled trial (RCT): participants randomly allocated to intervention or control; gold standard for treatment evaluation; minimises selection bias and confounding
- Cluster RCT: randomisation at group level (e.g. GP practices, schools)
- Crossover trial: participants act as their own controls; receive both treatments sequentially
- Non-randomised controlled trial: intervention assigned without randomisation; more prone to confounding
Observational Analytical Studies
- Cohort study: follows exposed and unexposed groups over time; can be prospective or retrospective; measures incidence and relative risk
- Case-control study: identifies cases (with disease) and controls (without), then compares past exposure; retrospective; measures odds ratio
- Cross-sectional study: snapshot of exposure and outcome at one time point; measures prevalence; useful for needs assessment
Descriptive Studies
- Case report: single patient; useful for rare conditions or novel presentations
- Case series: group of similar cases; no comparison group
- Ecological study: population-level correlations; subject to ecological fallacy
Qualitative Research
- Explores experiences, meanings, and processes
- Methods: interviews, focus groups, ethnography
- Analysis: thematic analysis, grounded theory, phenomenology
Clinical Presentation
Choosing the Right Study Design
- Treatment efficacy → RCT
- Aetiology/risk factors (common exposure) → Cohort study
- Aetiology/risk factors (rare disease) → Case-control study
- Disease burden/prevalence → Cross-sectional study
- Screening test accuracy → Cross-sectional (diagnostic accuracy study)
- Prognosis → Cohort study
- Harm/side effects → Cohort or case-control
- Systematic overview → Systematic review ± meta-analysis
Exam Tips
- Always identify the study design before interpreting results
- Know which measure of association belongs to which design (RR for cohort, OR for case-control)
- Understand the direction of enquiry (prospective vs retrospective)
- Be able to identify potential sources of bias for each design
Differential Diagnosis
| Study Design | Direction | Measure | Strengths | Weaknesses |
|---|---|---|---|---|
| RCT | Prospective | RR, ARR, NNT | Minimises confounding | Expensive, ethical constraints |
| Cohort (prospective) | Prospective | RR, incidence | Temporal sequence clear | Time-consuming, attrition |
| Cohort (retrospective) | Retrospective | RR | Faster, cheaper than prospective | Relies on existing data quality |
| Case-control | Retrospective | OR | Efficient for rare diseases | Recall bias, cannot measure incidence |
| Cross-sectional | Snapshot | Prevalence, OR | Quick, cheap | No temporal sequence |
| Ecological | Population level | Correlation | Generates hypotheses | Ecological fallacy |
Diagnosis / Investigation
Critical Appraisal Tools
- CASP checklists: structured tools for appraising different study designs
- Cochrane Risk of Bias tool: for assessing RCT quality
- Newcastle-Ottawa Scale: for cohort and case-control study quality
- GRADE framework: rates overall quality of evidence across studies
Key Questions for Appraisal
- Was the study design appropriate for the question?
- Was randomisation/allocation concealment adequate (RCTs)?
- Were participants/assessors blinded?
- Were groups comparable at baseline?
- Was follow-up adequate (>80%)?
- Were outcomes measured objectively?
- Were confounders identified and adjusted for?
- Are results clinically significant as well as statistically significant?
Management
Application to Clinical Practice
- Use evidence hierarchy to guide clinical decisions
- Prefer systematic reviews and RCTs for treatment decisions
- Recognise that observational studies may be the best available evidence for some questions (rare diseases, long-term outcomes, harms)
- Apply GRADE framework: considers study design, risk of bias, inconsistency, indirectness, imprecision, publication bias
- Translate findings: calculate NNT/NNH for patient discussions
Reporting Standards
- CONSORT: Consolidated Standards of Reporting Trials (RCTs); includes flow diagram and checklist
- STROBE: Strengthening the Reporting of Observational Studies in Epidemiology
- PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses
- STARD: Standards for Reporting of Diagnostic Accuracy Studies
- SPIRIT: Standard Protocol Items: Recommendations for Interventional Trials
Prognosis
- Study design directly determines the strength of evidence for clinical decisions
- RCTs provide the strongest evidence for treatment effects but may have limited external validity
- Real-world evidence from observational studies increasingly complements RCT data
- Meta-analyses pool results but are limited by heterogeneity and publication bias
- Poorly designed studies can lead to incorrect clinical conclusions and patient harm
Other Relevant Information
Summary Table of Study Designs
| Design | Question Answered | Key Measure | Time Direction |
|---|---|---|---|
| Systematic review | Synthesis of evidence | Pooled effect estimate | N/A |
| RCT | Does treatment work? | RR, ARR, NNT | Prospective |
| Cohort | Does exposure cause disease? | RR, incidence rate | Prospective/retrospective |
| Case-control | Was disease caused by exposure? | OR | Retrospective |
| Cross-sectional | How common is disease/exposure? | Prevalence, OR | Snapshot |
| Case series | What are features of condition? | Descriptive | Retrospective |