Skip to content

Protocol Generator · ST

Sequential Trial Emulation protocol

Draft a target-trial-emulation protocol as a sequence of nested trials — one emulated trial at each eligible time origin. Where the control arm is those not yet treated, patients who stay eligible enter several trials, so estimates are pooled across trials and the uncertainty has to account for that re-use — by a bootstrap over patients, or a patient-clustered robust variance. Handles time-varying eligibility, staggered entry, and the prevalent-user problem. Everything runs in your browser.

← All designs
1 · Identity

📚 Databases / sites (fills “Data source(s)”)

Taiwan

South Korea

Japan

Hong Kong

Australia

Finland

United Kingdom

Netherlands

United States

2 · Causal question
Population
Treatment (index)
Comparator
Outcome
3 · Sequence of trials
In one sentence: instead of one trial starting on one day, you run the same trial again on every day someone could have been randomised.

A single-origin new-user study has to pick one moment and discard everyone who becomes eligible later. Here each origin gets its own trial: eligibility, treatment assignment and the start of follow-up are all re-assessed at that origin, so each nested trial is itself a complete new-user trial and time zero stays aligned — which is what keeps immortal time out.

Two versions of this design exist and they are not interchangeable, so section 3 asks which you mean. If the control arm is those not yet treated at each origin (Hernán 2008, Danaei 2013, Gran 2010 — every paper cited on this page), a patient enters a trial at every origin until they initiate or have the outcome. If instead both arms initiate at the origin — an active comparator — a patient is assigned at the single origin at which they initiate.

The price of the first is that the same person appears in several trials, often with overlapping follow-up. Those trials are not independent samples, and two things follow: the variance must account for the re-use (Section 10 of the protocol), and per-trial estimates cannot be pooled as if they came from separate studies. Under the second, each patient sits in one trial and none of that applies.

4 · Assignment & enrolment
5 · Confounding, estimand & analysis
Subgroup analyses (tick to perform)
Sensitivity analyses (tick to perform)
Auto-saved to this browser.

Live protocol preview

 
 
 
 
Consistency checks · target-trial 防呆
 
 
 
 
Study design diagram
 
 
Within one trial — anchored at its origin
 
 
 
Across trials — the same trial repeated at every origin