Sequential / Group Sequential Trial Design
Also known as: group sequential design, adaptive stopping design, Ardışık Deneme Tasarımı (Sequential / Group Sequential)
Sequential and group sequential trial designs allow a study to be stopped early — or continued — based on interim analyses conducted as data accumulate. The core framework was formalised by O'Brien and Fleming in 1979 and extended by Lan and DeMets's alpha-spending approach, and it controls the overall Type I error rate across all planned looks by pre-specifying both efficacy and futility boundaries before enrolment begins.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use sequential design when continuous accrual of outcomes makes early stopping ethically or economically important — for example, in clinical trials where a large treatment benefit or serious harm should not be withheld or continued unnecessarily. The design suits continuous, binary, and ordinal outcomes in longitudinal studies. Four assumptions must be respected: the number and timing of interim analyses must be fixed in advance; the alpha-spending function must be tied to the information fraction (Lan-DeMets framework); operational bias must be controlled by keeping interim results blinded to the study team; and early stopping for benefit must be weighed carefully because it can overestimate treatment effects in the final analysis.
Strengths & limitations
- Controls the overall Type I error rate despite multiple interim looks, using rigorous alpha-spending functions.
- Enables ethical early stopping when evidence of efficacy or harm is overwhelming, reducing patient exposure to inferior treatment.
- Can reduce the expected sample size relative to a fixed design when a large effect is present.
- Applicable to continuous, binary, and ordinal outcomes within the same framework.
- Design complexity is substantially higher than a fixed-sample trial; incorrect implementation inflates the false-positive rate.
- Early stopping for benefit tends to overestimate the true treatment effect, which complicates post-trial meta-analyses.
- Requires prospective commitment to the number, timing, and statistical boundaries of interim looks — retrospective additions are not valid.
- Minimum sample of around 30 per group is needed before meaningful interim analyses are feasible.
Frequently asked
How many interim looks can I plan?
There is no hard limit, but each additional look reduces the per-look boundary and increases design complexity. In practice, two to five looks are most common. More frequent monitoring — approaching continuous sequential testing — can be handled with alpha-spending functions that allow information-fraction-based timing rather than equally spaced looks.
O'Brien-Fleming or Pocock — which boundaries should I choose?
O'Brien-Fleming boundaries are very conservative early in the trial (requiring very large Z values to stop early) and relax toward the end; they are preferred in most clinical trials because they preserve most of the alpha for the final analysis. Pocock boundaries are constant across looks and easier to cross early, but they require a stricter final boundary and may leave the trial with insufficient power at the end.
Does stopping early for efficacy bias the treatment effect estimate?
Yes. When a trial stops early because of a favourable result, the observed effect is drawn from the upper tail of the sampling distribution and is likely an overestimate of the true effect. Adjusted estimators (e.g. the conditional mean estimator or shrinkage methods) should be used in the final report, especially if the result will feed into a meta-analysis.
Can I add an interim look that was not in the original plan?
Not without invalidating the Type I error control. Any unplanned look consumes alpha that was not accounted for in the original spending function. If a truly unplanned look becomes necessary for safety reasons, the resulting p-values must be interpreted with extreme caution and regulatory guidance should be sought.
Sources
- O'Brien, P.C. & Fleming, T.R. (1979). A Multiple Testing Procedure for Clinical Trials. Biometrics, 35(3), 549–556. DOI: 10.2307/2530245 ↗
- Jennison, C. & Turnbull, B.W. (2000). Group Sequential Methods with Applications to Clinical Trials. CRC Press. ISBN: 978-0849303166
How to cite this page
ScholarGate. (2026, June 1). Sequential / Group Sequential Trial Design. ScholarGate. https://scholargate.app/en/experimental-design/sequential-design
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Adaptive Clinical Trial DesignExperimental design↔ compare
- Power analysisStatistics↔ compare
- Randomized Controlled TrialExperimental design↔ compare