TimelyScholarFree Tools › Trial Sequential Analysis Calculator
Check whether a cumulative meta-analysis has enough participants to be conclusive. The tool calculates the required information size and plots the cumulative Z-curve against O'Brien-Fleming-type monitoring boundaries.
FreeNo account neededYour data stays in your browser
| Required information size | 12,894 participantsα = 0.05, power 90%, control risk 0.03, RRR 30% |
|---|---|
| Diversity-adjusted (DARIS) | 251,168 participantsD² = 94.9% |
| Accrued information | 357,347 participants (142.3% of required) |
| Cumulative estimate | RR 0.49 (95% CI 0.34 to 0.70), Z = -4.00 |
The cumulative Z-curve crosses a monitoring boundary (or the conventional boundary after the required information size is reached): the evidence may be firm for the anticipated effect.
Boundaries use the Lan-DeMets O'Brien-Fleming-type alpha-spending function at each trial's information fraction (checked against the ldbounds R package). Futility boundaries are not shown. For a registered review, confirm results with the TSA software from the Copenhagen Trial Unit.
Study name and event counts in each arm.
Alpha, power, control event proportion and the relative risk reduction you want to detect.
Whether the Z-curve crosses a monitoring boundary or the required information size.
A meta-analysis updated as new trials appear performs repeated significance tests, which increases the risk of a false-positive result. Trial sequential analysis adjusts for this by comparing the cumulative Z-score with boundaries that are strict early on and relax as information accrues.
The required information size is the number of participants a single adequately powered trial would need, inflated for heterogeneity using the diversity (D²) of the random-effects meta-analysis.
The evidence may be firm enough to conclude on the assumed effect, even before the required information size is reached.
Use it to plan and check. For the final analysis, confirm results with the TSA software from the Copenhagen Trial Unit, which also provides futility boundaries.