AtlasSession 3L8: Study designp.49–74, 106–109
Randomised controlled trials
- Study designs primer
- Classification
- Cohort studies
- Randomised controlled trials
Fig. 1
MACI vs microfracture for knee cartilage defects (KOOS at 2 years): how each trial feature moves the seven bias domains
Fig. 1, What makes a trial fair? (First look). A trial of a cell-based cartilage implant vs standard keyhole repair; patients score their knee at 2 years. Allocation: Sealed envelopes: kept secret. Traffic lights: Groups alike at the start? Low risk; Who got into the study? Low risk; Who got what, recorded right? Low risk; Same care in both groups? High risk; Every patient counted? Low risk; Results scored fairly? High risk; All results shown? Can’t tell.
- Low risk
- 4/7
- Can’t tell
- 1
- High risk
- 2
- Next group kept secret?
- yes
Bias Scenario Classifier · First look
Level 1 · First look · Lecture 8 · p.49–74, 106–109
See why tossing a coin to decide treatment, and keeping it secret, makes a fair test.
Explain how random sequence generation, allocation concealment, blinding, ITT analysis, blinded outcome assessment and a published protocol reduce each of the 7 bias domains.
Fair teams (the step you are on)
A coin toss stops a captain stacking one side with the best players.
A coin toss stops a team captain stacking one side with the best players. Here chance picks each patient’s group, and the groups start alike: 4/7 lights show low risk.