Package Updates

Recent releases relevant to gsDesignSkills

gsDesign 3.10.0 (June 2026)

Main branch at github.com/keaven/gsDesign; not yet on CRAN.

New features

• gsSurvPower(): Power computation for group sequential survival designs with fixed enrollment, dropout, treatment effect, and analysis timing. Supports informationRates to cap spending at planned vs. realized information fractions and fullSpendingAtFinal to force spending fraction to 1 at final analysis.
• testUpper / testLower / testHarm: Selective bound testing at interim analyses for gsDesign(), gsSurv(), and gsSurvCalendar(). Inactive bounds set to extreme values and displayed as NA.
• test.type 7/8: Three-boundary designs (efficacy + futility + harm) with sfharm/sfharmparam for independent harm-bound spending.
• hr > hr0 support: Survival design functions handle reversed HR conventions for time-to-response or safety endpoints.
• Exact binomial functions: repeatedPValueBinomialExact(), sequentialPValueBinomialExact(), simBinomialSeasonalExact() for rare-event monitoring.
• toInteger() improvements: Preserves selective-bound flags and harm-bound spending; uses fixed-calendar enrollment inflation.
• sfExtremeValue2: Two-parameter flipped extreme value spending function for futility bound calibration.

Key references

• gsSurvPower vignette
• Selective bound testing vignette
• Harm bound vignette

---

gsDesign2 1.1.9 (2026)

Main branch at github.com/Merck/gsDesign2.

New features

• sequential_pval(): Sequential p-value for AHR group sequential designs. Use with gs_design_ahr() output. For gsDesign objects (from gsSurv(), gsSurvCalendar(), gsSurvPower()), continue to use gsDesign::sequentialPValue().
• Minimal risk weighting added to gs_design_rd() and gs_power_rd().
• gs_design_ahr() spending time output (v1.1.8): Can now output spending time directly.

---

rpact 4.4.0 (2026)

CRAN and github.com/rpact-com/rpact.

New features (4.3.0–4.4.0)

• getFutilityBounds(): Convert futility bounds between scales (z-value, p-value, conditional power, predictive power, reverse conditional power, effect estimate).
• futilityBoundsScale argument in getDesignInverseNormal() and getDesignGroupSequential() for specifying futility bounds on alternative scales.
• alpha0Scale argument in getDesignFisher() for alternative-scale alpha0 bounds.
• efficacyStops / futilityStops (v4.2.1): Control which stages have stopping.
• Dose-response (v4.2.0): Multi-arm simulations support doseLevels for linear or sigmoid Emax models.
• Unequal variances (v4.2.0): getSampleSizeMeans(), getPowerMeans(), getSimulationMeans() support different variances per group.

Key references

• rpact documentation
• Wassmer G, Brannath W. Group Sequential and Confirmatory Adaptive Designs in Clinical Trials. Springer, 2025.

---

graphicalMCP 0.2.9 (March 2026)

CRAN and github.com/openpharma/graphicalMCP.

Changes

• Fixed precision issue in graph_test_closure() parametric tests (#90).
• Added Hochberg-based procedures (v0.2.7).
• Internal validation via power simulations (v0.2.7).

---

How sequential p-values work with multiplicity

Design package	Sequential p-value function	Use with
gsDesign (gsSurv, gsSurvCalendar, gsSurvPower)	gsDesign::sequentialPValue()	gsDesign objects
gsDesign2 (gs_design_ahr, gs_power_ahr)	gsDesign2::sequential_pval()	gsDesign2 objects

Both produce a single p-value per hypothesis per analysis that can be passed to graphicalMCP::graph_test_shortcut() for FWER-controlled multiplicity testing.