Target trial emulation demo
A method that uses real-world data to mimic a randomized clinical trial, helping estimate treatment effects as if patients were randomly assigned to treatments.
At a glance
Use when
Assessing long-term treatment effects, rare outcomes, or in populations excluded from RCTs; when RCTs are not feasible or ethical
Avoid when
Critical unmeasured confounders are likely; data quality is poor; treatment definitions or timing are unclear
Inputs
Observational health data (e.g., registries, electronic health records), treatment definitions, eligibility criteria, outcome definitions, follow-up period
Outputs
Estimated treatment effects (e.g., hazard ratios, risk differences), measures of uncertainty (confidence intervals)
How it works
Target trial emulation applies causal inference techniques to observational data by defining a hypothetical randomized trial, specifying eligibility criteria, treatment strategies, follow-up periods, outcomes, and statistical analysis plans, then using methods like inverse probability weighting or g-computation to adjust for confounding.
- Project
- HTx
- Funding
- Horizon 2020
- Project status
- Completed 2024
- HTA domains
- Clinical Effectiveness
- Categories
- RWE
- Technology
- Medicines
- Assumptions
- Key confounders are measured and can be adjusted for; consistency between treatment assignment and observed data; positivity (all treatment options are available to all eligible patients); correct model specification
- Strengths
- Enables causal inference from non-randomized data; flexible design; can answer questions where RCTs are unethical or impractical; uses real-world patient populations
- Limitations
- Results depend on untestable assumptions (e.g., no unmeasured confounding); sensitive to model misspecification; requires high-quality, detailed observational data
Questions this answers
- › What would happen if patients were randomly assigned to different treatments?
- › How do treatments compare in real-world settings?
- › Can we estimate causal effects from observational data?
- › What are the risks and benefits of a treatment over time?
- › How do patient outcomes differ by treatment strategy?
- › Can real-world evidence support regulatory or reimbursement decisions?
Related methods
Similar by meaning
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.