Skip to contents

Changelog

Source: NEWS.md

ssutil 1.2.0

CRAN release: 2026-09-17

  • Fix bug on power_best_binomial()
  • Improve performance of sim_power_best_binomial()
  • Add test to confirm power_best_binomial() produces a valid result using simulations.
  • constrain parameter in wcs_power_best_binomial() to avoid search probabilities lower than the difference.
  • Modify sim_power_nbinom() to explicitly indicate the direction of the alternative hypothesis as it could be wrongly inferred from the rr alone.
  • Modify multz() and multq() to follow R convention in distributions. A new parameter lower.tail is added
  • Fix a bug an ss_best_normal() returns now an integer value
  • ss_ni_ve() now computes the total number of events using Freedman’s (1982) log-rank sample size formula instead of gsDesign::nBinomial1Sample(), which better reproduces Table 1 of Fleming et al. (2021) and removes the gsDesign package dependency
  • Added ve_exp/ve_ac parameters to ss_ni_ve(), allowing the experimental vaccine’s assumed efficacy to differ from the active comparator’s, which reproduces Table 2 of Fleming et al. (2021)
  • Added a true_diff parameter to sim_power_ni_normal() so simulations are no longer restricted to a true difference of 0 between groups, allowing sensitivity evaluation when the experimental group is truly slightly worse or better than control
  • Fixed prophr() to accept vector inputs for p0/hr without an “condition has length > 1” error
  • Corrected documentation: sim_power_best_bin_rank() selects the group with the highest total rank as best (roxygen said “lowest”); power_best_normal() returns a probability, not an integer sample size; sim_power_ni_normal()’s and sim_power_equivalence_normal()’s t_level examples/docs clarified
  • Removed dead/superseded code and a missing-comma typo in sim_power_nbinom()’s example

ssutil 1.1.0

CRAN release: 2026-09-06

  • Added the power_events_rate() function to compute the exact binomial probability of observing at least a given number of events, across combinations of sample sizes and risks.
  • Updated the “Power for rare events” vignette (formerly “Detectable Event Rate for Safety Signal Detection”) to cover power_events_rate() alongside power_single_rate().
  • Corrected several grammar, spelling, and terminology errors across the package vignettes.

ssutil 1.0.0

CRAN release: 2025-06-12

  • CRAN submission

ssutil 0.15.0

  • Remove lf_config functions as they are not useful. Replace by the worst case scenario function.
  • Modify sim_best_binomial() function to select the best using ranks rather than the custom approach of which == max().
  • Make uniform the parameters names
  • Provide standardized output for the simulations

ssutil 0.14.0

  • Re-factoring of the entire package.
  • Functions that calculate power based on simulation now start with the prefix sim_.
  • ni_fleming has been renamed to ni_ve.
  • The multz function was modified to correctly handle the case when there is only one group.
  • Improve documentation

ssutil 0.13.0

  • Added the probhr function to estimate the probability of an event in the experimental group based on the control group probability and the hazard ratio, assuming proportional hazards.

ssutil 0.12.0

  • Added the option to preserve the effect.
  • Modified the formula used in the Fleming method.

ssutil 0.11.0

  • Added a 30% criterion to the ni_fleming function.

ssutil 0.10.1

  • Updated the messages displayed by ni_fleming.

ssutil 0.10.0

  • power_best_binomial and power_best_normal were updated to match their rank-based equivalents.
  • Now, dif represents the difference between the most promising group and the rest.
  • For binomial tests, prob is now the probability in the most promising group (previously it referred to the other groups), and dif defines how much lower the other groups are.

ssutil 0.9.0

  • Added support for weights in power_best_norm_ranks and power_best_bin_rank.
  • Note: the API is not backward compatible for these two functions.

ssutil 0.8.0

  • Added power_best_norm_rank to empirically estimate power for normal distributions based on ranks.

ssutil 0.7.0

  • Added a function to calculate non-inferiority in vaccine trials following Fleming et al.

ssutil 0.6.0

  • Fixed an error in power_test_binomial where, in the absence of ties, the sample function randomly selected a number instead of sampling from a range.
  • Fixed the same issue in power_best_normal.
  • power_best_binomial now returns a data frame with power and 95% confidence interval.
  • power_best_normal now returns a data frame with power and 95% confidence interval.
  • power_ni_normal now includes the number of simulations in the output data frame.
  • Added power_best_bin_rank, which selects the best group based on ranks, instead of assuming the highest value as in power_best_binomial.

ssutil 0.5.0

  • Added the power_ni_normal function.

ssutil 0.4.1

  • Fixed an error in the ggplot_prob_lowest_power graph.
  • Fixed some global variables and cleaned up imports.

ssutil 0.4.0

  • Added the power_equivalence_normal function.

ssutil 0.3.0

  • Added the power_best_binomial function and related helpers.

ssutil 0.2.0

  • Added the power_best_normal function and related helpers.

ssutil 0.1.1

  • Added a NEWS.md file to track package changes.
  • Added the power_single_rate function.