Package: gsDesignNB 0.3.2

Purpose | Example task | Time-scale setup | Wald versus score sizing | Calendar-time group sequential design | Simulate, cut, and test a small data set | Production workflow reminder | What this skill is and is not

Overview | Dataset 1: Near-Zero Blinded Information | Event Counts Summary | Event Distribution | Event Rates by Patient | Violin Plot of Event Rates | What Happened with the Information Calculations | Root Cause Analysis | Method of Moments Comparison | Dataset 2: Extreme High Blinded Information | Why the Mutze Test Falls Back to Poisson | Summary of Issues | Recommendations

Trial design parameters | Sample size calculation | Group sequential design | Simulation study | Load simulation results | Simulation results summary | Summary of verification results | Overall operating characteristics | Power comparison by analysis | Visualization of Z-statistics | Notes

Notation | Methodology | Negative binomial distribution | Gamma-Poisson mixture motivation | Graphical illustration | Sample size formula | Relationship between Zhu-Lakkis, Friede-Schmidli, and Mutze et al. | Non-inferiority and super-superiority | Average exposure | Setup | Calendar exposure without dropout | Calendar exposure with dropout | Maximum follow-up | Weighted average across segments | Group-specific parameters | Variance inflation for variable follow-up ($Q$ factor) | Event gaps | Simulation verification of average exposure | Statistical information | Per-subject Fisher information | Total information | Information in practice: blinded and unblinded estimation | Connection to sample size | Examples | Basic calculation [@zhu2014sample] | Piecewise constant accrual | Accrual with dropout and max follow-up | Group-specific dropout rates | Calculating power for fixed design | Unequal allocation | Example with event gap | References

Introduction | Load pre-computed results | Scenario grid | Type I error comparison | Power comparison | Z-score density comparison (null simulations) | Fallback method frequency | Summary

Simulation setup | Defining input parameters | Running the simulation | Analyzing the data | Inspect first ten records | Plotting events | Cutting data by analysis date | Missing data assumptions and imputation | Wald test [@mutze2019group] | Finding analysis date for target events | Generating and verifying negative binomial data | Simulation with dispersion | Verifying the dispersion parameter | Visualizing the distribution | References

Introduction | Planned trial design | Fixed design | Group sequential design | Group sequential sample size under each nuisance scenario | Expected information fraction at planned time of each interim | Scenario grid | Simulation engine | Running simulations | Results | Starting sample-size sensitivity | Power by rate ratio and SSR strategy | Power by nuisance scenario | Adapted sample size distribution | Expected sample size at study stop | Calendar and information at each analysis | Discussion | Key findings | Information-based interim timing | Futility at low information | Blinded information fallback | Computational considerations | References

Introduction | Simulation design | Parameters | Theoretical calculation | Simulation results | Power comparison | Summary of verification results (corrected design) | Distribution of the test statistic | Wald vs score Z-score comparison | Distribution statistics for log(RR) | Type I error: Wald vs score test | Scenario sweep: impact of the Jensen correction | Discussion: why we apply the correction

Introduction | Statistical background | Negative binomial count model | Imputation draw | Bootstrap–MI combination | Worked example | Simulating longitudinal count data | Introducing missing data by mechanism | Running impute_nb() | Inspecting imputed values | Comparing imputed means by strategy | Pooled analysis with Rubin's rules (standard MI) | Bootstrap–MI for variance estimation | Using the building-block functions directly | Step 1 — Fit the GLMM | Step 2 — MAR imputation | Step 3 — MNAR reference-based imputation | Step 4 — Composite strategy (no model required) | Key considerations | Session info

Introduction | Trial setup and initial design | Initial sample size calculation | End-to-end helper workflow | Simulation | Interim analysis | Information computation | Sample size re-estimation | Final analysis | Updating bounds | References

Non-inferiority design | Super-superiority design

Simulation setup | Simulation loop | Results summary | Visualization | Group sequential design evaluation

Simulation setup | Define parameters | Run simulation | Analysis at a cut date | Seasonal and treatment effect estimation | Variance and information