OnCo
ideasIdea

A pre-registered standard for emulating trials with real-world data

When researchers use hospital records to ask 'would drug A have beaten drug B in a trial', they should follow a published recipe and register their plan first, so the answer can be trusted.

Target trial emulation (specifying the hypothetical randomised trial, then mimicking it in observational data) is the accepted framework for causal inference from real-world data, but oncology emulations vary widely in rigour. The proposal is a formal standard: a protocol template (eligibility, treatment strategies, assignment, time zero, outcomes, causal contrast, analysis plan), mandatory pre-registration, and a requirement to report a positive control emulation of a known RCT in the same dataset, adopted by regulators and journals as a condition of considering RWE.

Hypothesis
Emulations that follow the standard and pass a positive-control calibration will agree with subsequent randomised trials on the direction of effect in more than 85 percent of cases, versus roughly 60 percent for unstandardised observational oncology studies.
Rationale
The RCT-DUPLICATE project showed that carefully designed emulations can reproduce RCT results when design elements are matched; the failures came from avoidable design flaws such as immortal time bias. A standard forces the fixes.
What would test it
Apply the standard prospectively to ten oncology questions with ongoing RCTs; lock the emulation results before the trials read out; compare.
Maturity
early clinical
Who has to act
regulator
Cost to try
Small (under $1M)
Years to first evidence
3
Bottlenecks it attacks

Connected

6top

Pages like this

not linked directly; found by shared links