Trials Flashcards
What are the differences between observational studies and randomised controlled trials ?
Observational Studies:
- draws conclusions about the possible association of an exposure on a health outcome
- researchers observe subjects without intervening
- includes cohort, case control, and cross-sectional studies
- prone to confounding and bias
- can establish correlation but not causation
RCTs:
- each participant is randomly assigned to a treated group or a control group before the intervention is given which allows for a direct comparison of outcomes
- used to determine causality between interventions and outcomes
- minimises bias and confounding errors due to randomisation
- considered gold standard for evaluating interventions
Describe the key features of RCTs:
Randomisation - participants are randomly assigned to groups minimising selection bias and confounding
Control group - used a comparison to assess the effect of the intervention
Blinding – participants, researchers, or analysts may be unaware of group assignments to prevent bias
Pre specified outcome measures - defined before the trial to prevent selective reporting
Follow-up & Adherence Monitoring – Ensures compliance and reduces loss to follow-up
Ethical Considerations – Requires informed consent, ethical approval, and stopping rules for safety
What are the steps
involved in planning
and conducting a
randomised trial?
Formulation of a clear question to be answered
Definition of the inclusion criteria for participants
Performance of a sample size calculation
Choice of the method of randomisation to ensure adequate sequence generation and and allocation concealment
Blinding of participants and and observers where appropriate
Choice and measurement of appropriate outcomes
Monitoring of compliance
Conduct of a statistical analysis
Appropriate reporting of results
Informed by a systematic review
Why use randomisation ?
Reduces bias - ensures comparable groups at baseline
Balances confounding errors - both known and unknown confounders are distributed equally (confounding error is when an uncontrolled third variable (confounder) affects both the independent variable (IV) and the dependent variable (DV), leading to a false association between them)
Facilitates causal inference - observed effect can be attributed to the intervention rather than external factors
Provides a valid basis for statistical test of significance
Controls for time trends by having a concurrent comparison group
What are the different types of RCTs ?
Parallel group RCT:
- participants randomly assigned to a group
- Each group receives a different intervention simultaneously and remains in that group throughout
Crossover RCT:
- participants receive both interventions at different time points (sequential order)
- subjects act as their own control → reduces variability
- requires a washout period to prevent carryover
Cluster RCT:
- Randomises groups (clusters) rather than individuals
- Reduces contamination bias but requires larger sample sizes
What are the different types of randomisation ?
Simple Randomisation:
- Each participant has an equal chance of assignment
- risk as can lead to an imbalance in group sizes for small trials
Block Randomisation:
- ensures equal numbers in treatment and control groups by randomising in fixed “blocks”
- Used in smaller studies to maintain balance
Stratified Randomisation:
- Ensures equal distribution of confounding variables
- Participants are stratified into subgroups before randomisation
Minimisation:
- Assigns participants based on minimising imbalances in key factors
Describe the term “intention to treat analysis” :
A statistical principle where participants are analysed in the groups to which they were originally assigned, regardless of adherence or dropout.
✔ Maintains randomisation benefits by preserving baseline comparability.
✔ Provides real-world effectiveness rather than ideal conditions.
✔ Reduces bias from loss to follow-up and non-compliance.
✔ Contrasts with per-protocol analysis, which only includes participants who fully followed the intervention.
What are the strengths of RCTs ?
Causal inference - most efficient design for investigating causality, because we can ensure that the ‘cause’ precedes the ‘effect’
Control of bias - possible confounding don’t confuse the results
Interventions are compared efficiently
Uncertainty is resolved in a fair way
Randomisation facilitates statistical analysis
What are the limitations of RCTs ?
Cost and time - can be long and expensive
Complex interventions - public health interventions may not be easily randomised
Not suitable for rare disease or diseases with a long latency
Withholding potentially effective interventions from the control group may raise ethical issues
Applicability - vulnerable groups such as the very young, very old and pregnant women may have be screened out
Volunteer bias
External validity - the controlled environment of the RCT may not reflect the real world environment, limits generalisability
Describe some ethical considerations needed when doing RCTs:
Informed Consent - ensuring participants are fully aware of risks and benefits
Special care must be taken when dealing with vulnerable groups (e.g., children, low-income communities) to ensure they understand the risks and benefits
Approval by Ethics Committee - mandatory in human research
Stopping Rules - criteria for ending the trial early (e.g., if the intervention proves harmful or highly effective)
Equipoise - randomised trials require equipoise, meaning there must be genuine uncertainty about whether the intervention will benefit participants
