Changelog

Source: NEWS.md

pharmhand 0.4.3.9000

• Safety hierarchy tables now retain treatment groups with zero events using names_expand = TRUE.
• Improved robustness in AE comparisons by short-circuiting when denominators are zero.
• Efficacy baseline calculations now use accessor helpers (get_filtered_data, get_trt_var) instead of direct slot access.
• Added input validation for treatment variables in laboratory tables to prevent unsafe glue syntax.
• Bayesian meta-analysis warmup parameter now requires a positive integer (>= 1) for consistency with brms.
• Standardized on rlang::sym across all table creator functions.
• Added footnotes property to ClinicalTable class and updated reporting engine to populate it.
• Fixed ph_abort() to properly concatenate multiple arguments before passing to rlang::abort().
• Documentation and test suite improvements, including explicit skip guards for mmrm and cmdstanr dependencies.

pharmhand 0.4.2.9000

• Validation helpers now reject NA consistently across assertion functions.
• Reporting engine refactor: title/footnotes helper, ADaMData coercion, empty gt styling.
• Export improvements: preserve ClinicalTable styling, HTML title handling.
• Safety/efficacy robustness: NNH bounds ordering, warnings on empty filters, Treatment column guard, case-insensitive DISCONTINUED matching.
• Added .format_n_over_n formatting helper for n/N display.
• Test refactors for clarity and maintainability.

pharmhand 0.4.1.9000

Code Quality Improvements

Validation Utilities

• Added comprehensive assertion functions for input validation:
  • assert_all_positive(), assert_positive(), assert_non_negative()
  • assert_positive_integer(), assert_non_negative_integer()
  • assert_character_scalar(), assert_character_vector()
  • assert_numeric_scalar(), assert_numeric_vector()
  • assert_integer_scalar(), assert_logical_scalar()
  • assert_data_frame(), assert_column_exists()
  • assert_lengths_match(), assert_no_na(), assert_in_range()

Messaging and Helpers

• Added ph_abort() for consistent error handling with abort messaging
• Added ph_inform() for informational messages
• Added ph_warn() for warning messages
• Added get_subject_n() for extracting subject counts from ADaM datasets
• Added get_summary_label() for generating standardized summary labels
• Added || operator (grapes-or-or-grapes) for flexible default value handling

Documentation Improvements

• Added man pages for internal helper functions:
  • .build_evidence_summary_row(), .build_league_matrix(), .build_rob_summary_row()
  • .build_study_characteristic_row(), .calculate_ae_comparisons()
  • .empty_ae_exposure_df(), .ensure_adam_data(), .summarize_ae_exposure(), .summarize_ae_hierarchy()
• Added man pages for utility functions: ph_abort(), ph_inform(), ph_warn()
• Added man pages for new helpers: get_subject_n(), get_summary_label(), ||

Cleanup and Maintenance

• Removed .tldr/ directory and cache files (TLDR code analysis cache)
• Deprecated safe_pct() function (removed man page)
• Removed 11 old imputation test problem files from tests/testthat/_problems/
• Added .tldr/ and vignettes/safety-tables_files/ to .gitignore
• Updated README files and pkgdown configuration
• Updated vignettes for baseline, efficacy, and safety tables

pharmhand 0.4.0.9000

Phase 3: Quality Assessment

Risk of Bias Assessment (RoB 2)

• Added RoB2Result S7 class for storing RoB 2 assessment results.
• Added assess_rob2() for single-study risk of bias assessment using Cochrane RoB 2 tool.
• Added assess_rob2_batch() for batch processing multiple studies from a data frame.
• Added rob2_summary() for creating summary tables of RoB 2 assessments.
• Implements all 5 RoB 2 domains: randomization, deviations, missing data, measurement, selection.
• Automatic overall judgment calculation following RoB 2 algorithm.
• Reference: IQWiG Allgemeine Methoden Section 10.1.4.

Risk of Bias Assessment (ROBINS-I)

• Added ROBINSIResult S7 class for storing ROBINS-I assessment results.
• Added assess_robins_i() for single-study risk of bias assessment using ROBINS-I tool.
• Added assess_robins_i_batch() for batch processing multiple studies from a data frame.
• Added robins_i_summary() for creating summary tables of ROBINS-I assessments.
• Implements all 7 ROBINS-I domains: confounding, participant selection, classification of interventions, deviations, missing data, outcome measurement, selection of reported result.
• Supports 5-level judgments: Low, Moderate, Serious, Critical, No information.
• Reference: IQWiG Allgemeine Methoden Section 10.1.4.

Risk of Bias Visualization

• Added create_rob_traffic_light_plot() for traffic light visualization of domain-level judgments.
• Added create_rob_summary_plot() for stacked bar plot summaries of bias assessments.
• Added save_rob_plot() for exporting plots to PNG, SVG, PDF and other formats.
• Added rob_data_to_tidy() for converting assessment results to tidy data frames.
• Supports both RoB 2 and ROBINS-I assessment results.
• Uses standard risk of bias color scheme (green/yellow/red/gray).

Bias-Adjusted Meta-Analysis

• Added bias_adjusted_meta() for meta-analysis with risk of bias adjustments.
• Added rob_sensitivity_analysis() for sensitivity analysis across RoB scenarios.
• Added calculate_rob_weights() for computing RoB-adjusted study weights.
• Added summarize_bias_adjusted() for comparing original and adjusted estimates.
• Supports three adjustment methods: weight_downgrade, exclude_high, selection_model.
• Integrates with RoB 2 and ROBINS-I assessment results.

IQWiG Evidence Grading

• Added grade_evidence() for assessing evidence certainty using IQWiG methodology.
• Added assess_evidence_domains() for evaluating individual evidence domains.
• Added format_evidence_grade() for German (Beleg/Hinweis/Anhaltspunkt) or English output.
• Implements IQWiG 3.1.4 evidence grading: Beleg (Proof), Hinweis (Indication), Anhaltspunkt (Hint), Kein Beleg (No proof).
• Evaluates: study limitations, inconsistency, imprecision, indirectness, publication bias.
• Integrates with RoB 2 assessments for study quality evaluation.

Evidence Summary Tables

• Added create_evidence_summary_table() for IQWiG-formatted evidence summary tables.
• Added create_study_characteristics_table() for G-BA Module 4 study characteristics.
• Added export_evidence_table() for multi-format export (Word, HTML, Excel).
• Includes: endpoint, N studies, effect estimate, CI, p-value, I², RoB summary, evidence grade.
• Supports German/English bilingual output with IQWiG-compliant formatting.

pharmhand 0.3.4.9000

Fixes and maintenance

• Fixed create_clinical_table() to accept AnalysisResults objects (uses @stats).
• Improved Bayesian meta-analysis test performance and stability.
• Cleaned up documentation/signature mismatches and R CMD check issues.

Major Architecture Improvements

Package-Wide Defaults System

• New defaults infrastructure: Added centralized default parameter system via ph_default() function
  • Default values for common parameters (e.g., trt_var = "TRT01P", autofit = TRUE, conf_level = 0.95)
  • Users can override via options: options(pharmhand.trt_var = "ARM")
  • Reduces boilerplate across 40+ functions
  • Improves consistency and maintainability

Code Reusability Improvements

• New create_clinical_table() factory function: Reduces boilerplate for table creation
  • Consolidates repeated pattern of creating flextable + wrapping in ClinicalTable
  • Used across 25+ table creation functions
  • ~200 lines of duplicate code eliminated
• Extracted estimate_tau2() helper: Eliminates duplication in meta-analysis functions
  • Single implementation of DerSimonian-Laird, Paule-Mandel, and REML methods
  • Used by both meta_analysis() and calculate_heterogeneity()
  • ~100 lines of duplicate code eliminated

Enhanced User Experience

• Automatic data.frame coercion: User-facing functions now accept data.frames directly
  • create_demographics_table() automatically wraps data.frames in ADaMData
  • Eliminates need for users to manually create S7 objects
  • Backward compatible with existing ADaMData inputs
• Improved error messages: Context-rich error reporting throughout
  • Error messages now show what was received vs. what was expected
  • Available values suggested when appropriate (e.g., treatment arms)
  • Missing column errors show available columns for easier debugging
• New workflow helpers: High-level convenience functions for common tasks
  • quick_demographics_report(): Generate demographics report in one call
  • quick_safety_report(): Generate multi-table safety report in one call
  • Ideal for quick analyses and prototyping

API Consistency

• Standardized parameter names: Using ph_default() across the package
  • trt_var (not treatment_var) for treatment variable
  • conf_level (not ci_level or confidence_level) for confidence level
  • autofit consistently for table formatting
  • Reduces cognitive load for users
• Consistent return types: All create_*_table() functions return ClinicalTable objects
  • Fixed inconsistencies where some returned bare flextables
  • Enables consistent downstream processing

Breaking Changes

Source File Reorganization

• Efficacy module split: R/efficacy_tables.R has been split into focused modules:
  • R/efficacy_primary.R - Primary endpoint tables (create_primary_endpoint_table())
  • R/efficacy_cfb.R - Change from baseline tables
  • R/efficacy_lab.R - Laboratory tables
  • R/efficacy_tte.R - Time-to-event tables
  • R/efficacy_responder.R - Responder analysis tables
  • R/efficacy_subgroup.R - Subgroup analysis tables
• Safety module split: R/safety_tables.R has been split into focused modules:
  • R/safety_summary.R - AE summary tables (create_ae_summary_table())
  • R/safety_comparison.R - AE comparison tables
  • R/safety_hierarchy.R - SOC/PT hierarchy tables
  • R/safety_tte.R - Safety time-to-event analysis
  • R/safety_exposure.R - Exposure-adjusted analysis
• Migration guidance: Users who previously sourced these files directly should:
  • Use the package namespace (e.g., pharmhand::create_primary_endpoint_table())
  • Or source the new module files (e.g., source('R/efficacy_primary.R'))
  • Update any direct imports to reference the new module names

Bug Fixes

Robustness Improvements

• Fixed create_primary_endpoint_table() to handle empty data or all-NA values gracefully, returning “-” instead of NaN/Inf
• Fixed create_responder_analysis_table() to compute response rate in separate mutate step, avoiding fragile self-reference in summarise
• Fixed create_tte_summary_table() to correctly extract landmark survival estimates for each treatment stratum instead of using first stratum for all
• Fixed create_tte_summary_table() to use dynamic CI column names based on conf_level instead of hardcoded “lower .95”/“upper .95”
• Fixed create_league_table() to use NMA result’s stored confidence level when available
• Fixed leave_one_out() to read ci_level from correct slot (meta_result@ci_level instead of @metadata$conf_level)

Input Validation

• Require positive, non-missing standard errors in eggers_test(), trim_and_fill(), meta_analysis(), and calculate_heterogeneity()
• Added length(yi) == length(sei) validation in eggers_test() and trim_and_fill()
• Added anyNA(yi) check in eggers_test() returning NA-filled result with clear interpretation
• Added anyNA(yi) check in trim_and_fill() with error listing invalid indices

Code Quality

• Changed stop() to ph_abort() in example script for consistent error handling
• Fixed vignette code fence indentation mismatch in efficacy-tables.Rmd
• Added := import from rlang to fix undefined global function warning
• Fixed create_mean_plot() n calculation to count non-missing values
• Added guards for CI computation when n <= 1 to avoid Inf values

Plotting Fixes

• Fixed create_efficacy_waterfall_plot() to compute category counts in separate mutate step, avoiding fragile self-reference in summarise
• Fixed create_forest_plot() Cox model CI column extraction to use positional indexing instead of fragile name construction
• Fixed create_km_plot() risk table grid background by explicitly setting panel.grid.major/minor to element_blank()
• Fixed safe_pct() division by zero in safety_summary.R by adding helper function applied to 9 locations

Bayesian Meta-Analysis Fixes

• Fixed bayesian_meta_analysis() to use brms::neff_ratio() instead of non-exported brms::ess_bulk()/brms::ess_tail()
• Fixed bayesian_meta_analysis() to compute BFMI from nuts_params energy values
• Fixed bayesian_meta_analysis() to use brms::nuts_params() for divergent transitions instead of accessing internal fit$fit@sim$divergent__
• Added tryCatch wrapper around brms::brm() calls for better error handling
• Changed all brms tests to use cores = 1 to avoid parallel rstan issues in CI
• Added bayesplot to Suggests, removed direct rstan dependency
• Fixed roxygen documentation to avoid \describe{} block Rd parsing issues
• Added brms test caching helper to speed up test suite

Documentation

• Added @title tags and completed @return documentation in meta_bias.R
• Added within-study variation warning documentation in assess_transitivity()
• Split long @return lines in safety_tte.R
• Added description field to meta-analysis vignette
• Updated brms section in meta-analysis vignette with installation guidance

Test Improvements

• Refactored test-plotting_forest.R, test-efficacy_tte.R, and test-efficacy_subgroup.R to use shared test fixtures
• Included documentation examples to exported functions in efficacy and safety modules
• Updated test-meta_bayesian.R with 8-study test data and adapt_delta = 0.99 to prevent divergent transitions
• Seeded RNG in 11 tests in test-pro-analysis.R
• Applied set.seed() to test-efficacy_tte.R before sample() call
• Removed duplicate tests in test-efficacy_responder.R and test-efficacy_subgroup.R

pharmhand 0.3.0.9000

Major Changes

Test Suite Reorganization

• One-test-file-per-source-file convention: Reorganized test files to follow R testing best practices.
  • Split test-efficacy_tables.R into test-efficacy_cfb.R, test-efficacy_lab.R, test-efficacy_primary.R, test-efficacy_responder.R, test-efficacy_subgroup.R, and test-efficacy_tte.R
  • Split test-meta-analysis.R into test-meta_core.R, test-meta_bayesian.R, test-meta_bias.R, test-meta_indirect.R, test-meta_network.R, and test-meta_plots.R
  • Split test-plotting.R into test-plotting_survival.R, test-plotting_efficacy.R, and test-plotting_forest.R
  • Renamed test-safety_tables.R to test-safety_summary.R and extracted test-safety_comparison.R and test-safety_hierarchy.R

New Test Coverage

• Added comprehensive tests for Configuration API (test-config_api.R)
• Added comprehensive tests for Configuration Classes (test-config_classes.R)

Bug Fixes

Test Data Improvements

• Fixed test data in safety TTE tests to use larger sample sizes (20+ subjects per arm) to avoid Cox model convergence issues
• Fixed test data to avoid duplicate landmark timepoints in TTE summary tests
• Fixed test data for subgroup analysis to have adequate subjects per subgroup (25+)
• Fixed S7 class checks from expect_s3_class() to S7::S7_inherits() for proper S7 class validation
• Fixed error message regex patterns to match actual function outputs

Source Code Fixes

• Fixed km_summary matrix conversion in efficacy_tte.R to preserve column names for single treatment arms
• Fixed pct column warning in safety_summary.R by using proper dplyr::mutate() instead of for-loop assignment
• Fixed survfit n.censor matrix handling in plotting_survival.R by converting to vector
• Fixed min() returning -Inf in pro_analysis.R when no events present
• Fixed S7 class checking in config_api.R to use S7::S7_inherits() instead of inherits()

Test Cleanup

• Removed tests for unimplemented maic() and stc() functions
• Fixed missing registry variable in config API tests
• Changed REML tests to use DerSimonian-Laird method to avoid convergence warnings
• Replaced any(is.na()) with anyNA() per jarl recommendations

Code Quality

• All tests pass (1874 tests)
• No warnings or skipped tests
• lintr clean (line length compliance)
• jarl check clean
• devtools::check() passes with 0 errors, 0 warnings, 0 notes

pharmhand 0.2.3.9000

Documentation

• Added runnable examples for all meta-analysis and network meta-analysis functions including meta_analysis(), indirect_comparison(), network_meta(), create_network_plot(), assess_transitivity(), node_splitting(), calculate_sucra(), and create_league_table().

pharmhand 0.2.2.9000

Phase 2: Evidence Synthesis

Meta-Analysis Functions

• Added meta_analysis() for fixed-effect and random-effects meta-analysis with support for multiple tau² estimators (DerSimonian-Laird, REML, Paule-Mandel, ML) and optional Knapp-Hartung adjustment.

• Added calculate_heterogeneity() for comprehensive heterogeneity assessment including Q, I², τ², and H² statistics with interpretation.

• Added leave_one_out() for sensitivity analysis identifying influential studies.

• Added create_meta_forest_plot() for meta-analysis forest plots with study weights, heterogeneity statistics, and prediction intervals.

• Added create_funnel_plot() for publication bias visualization.

• Added eggers_test() for statistical assessment of funnel plot asymmetry.

• Added trim_and_fill() for Duval & Tweedie publication bias adjustment.

Indirect Comparison

• Added indirect_comparison() using the Bucher method for anchored indirect treatment comparisons.

• Added compare_direct_indirect() for consistency testing between direct and indirect evidence.

Network Meta-Analysis

• Added network_meta() for network meta-analysis comparing multiple treatments.

• Added create_network_plot() for network geometry visualization.

• Added assess_transitivity() for evaluating the transitivity assumption.

• Added node_splitting() for testing inconsistency between direct and indirect evidence.

• Added calculate_sucra() for treatment ranking with SUCRA/P-scores.

• Added create_league_table() for pairwise comparison tables.

Bayesian Analysis

• Added bayesian_meta_analysis() interface for Bayesian meta-analysis using brms when available, with automatic fallback to frequentist methods.

Phase 1 Completion

PRO Analysis

• Added calculate_mcid_anchor() for anchor-based MCID calculation.

• Added calculate_mcid_distribution() for distribution-based MCID (0.5 SD, 1 SEM, etc.).

• Added calculate_mcid() wrapper combining both MCID approaches.

• Added create_ttd_analysis() for time-to-deterioration analysis with Kaplan-Meier estimation.

Visualization

• Added create_mean_plot() for longitudinal mean plots with confidence intervals.

• Added create_spider_plot() for individual patient trajectory visualization.

Safety Analysis

• Added create_ae_hierarchy_table() for full MedDRA hierarchy analysis (SOC → HLGT → HLT → PT).

Subgroup Analysis

• Added assess_iceman() for ICEMAN criteria assessment of subgroup credibility.

S7 Classes

• Added comprehensive tests for MultiArmStudy S7 class.

Documentation

• Added extensive test coverage for meta-analysis functions.
• Added tests for PRO analysis functions.

pharmhand 0.2.1.9000

Improvements

• Added conf_level parameter to create_ae_cumulative_incidence_plot() (was hard-coded at 0.95).
• Exposed subgroup_counts and min_subgroup_size in create_subgroup_analysis_table() metadata for programmatic access.
• Wrapped test_ph_assumption() call in create_tte_summary_table() with tryCatch for robustness with sparse Cox models.

Plotting

• Created .pharmhand_theme() internal helper for consistent white backgrounds across all plots.
• Refactored all plotting functions to use .pharmhand_theme() with base_size parameter.
• Added conf_level parameter to create_loglog_plot().
• Added base_size parameter to create_ae_cumulative_incidence_plot() and create_forest_plot().
• Updated build_schoenfeld_plot() to use .pharmhand_theme() for consistent styling.

Documentation

• Clarified that method parameter in test_non_inferiority() applies only to binary endpoints.
• Added @examples to detect_floor_ceiling(), test_non_inferiority(), ancova_adjust_continuous(), create_loglog_plot(), and create_ae_cumulative_incidence_plot().
• Added Time-to-Event Safety Analysis section to safety-tables vignette.
• Added NNH (Number Needed to Harm) documentation to safety-tables vignette.
• Added Non-Inferiority Testing section to efficacy-tables vignette.
• Added ANCOVA Adjustment section to efficacy-tables vignette.
• Added Floor/Ceiling Detection section to efficacy-tables and baseline-tables vignettes.
• Improved documentation for test_non_inferiority() with clarified conf_level, decision logic, and higher_better parameter.
• Improved documentation for ancova_adjust_continuous() with edge case behavior and runnable simulated example.

Code Quality

• Added Newcombe-Wilson reference citation for binary difference in proportions CI calculation.
• Removed unused comparison_results stub from AE comparison function.

Tests

• Added test for NNH calculation when risk difference is negative (treatment beneficial).

pharmhand 0.2.0.9000

Code Quality

• Refactored duplicated palette resolution logic into .resolve_palette() helper.
• Added warning when log-log plot drops data points with non-finite values.
• Expanded PH assumption disable comment with rationale for time-to-first AE.
• Added clarifying comments in floor/ceiling detection and event conversion.
• Added documentation note about CI construction methods for non-inferiority binary endpoints.
• Removed redundant @keywords internal from exported ancova_adjust_continuous().

Tests

• Made color palette test assertions order-agnostic with expect_setequal().
• Added edge case tests for calculate_exposure_adjusted_rate() (zero events, small exposure, large counts).

pharmhand 0.1.21.9000

Bug Fixes

• Fixed ComparisonResult defaults from integer(0) to numeric(0) for consistency with S7 class.
• Fixed line length lint errors in test_non_inferiority() documentation.

Documentation

• Updated pkgdown Utilities section description to include “statistical diagnostics”.
• Added missing topics to pkgdown reference index: calculate_exposure_adjusted_rate(), create_ae_exposure_table(), create_time_to_first_ae(), and test_non_inferiority().
• Cleaned up duplicate NEWS.md heading.
• Added explanatory comment for PH assumption check disable in time-to-first AE function.

Plotting

• Added show_censor support to create_loglog_plot() and passed through from create_km_plot(type = "loglog").
• Updated documentation to clarify that censor marks are supported on log-log plots while median lines, CI bands, risk tables, and landmarks are intentionally omitted.
• Added test for log-log plot censor marks.

pharmhand 0.1.20.9000

Improvements

• Added AE cumulative incidence plots via create_ae_cumulative_incidence_plot().

pharmhand 0.1.19.9000

Improvements

• Added ANCOVA adjustment for continuous endpoints via ancova_adjust_continuous().

pharmhand 0.1.18.9000

Improvements

• Added non-inferiority testing for continuous and binary endpoints via test_non_inferiority().

pharmhand 0.1.17.9000

Improvements

• Added time-to-first adverse event analysis with KM tables, plots, and Cox HRs via create_time_to_first_ae().

pharmhand 0.1.16.9000

Improvements

• Added exposure-adjusted incidence rate tables with Poisson confidence intervals via calculate_exposure_adjusted_rate() and create_ae_exposure_table().

pharmhand 0.1.15.9000

Improvements

• Added configurable warnings for small subgroups in create_subgroup_table() and create_subgroup_analysis_table() via min_subgroup_size (default: 20).

pharmhand 0.1.14.9000

Documentation

• Documented the complete proportional hazards assumption testing suite for Cox models: Schoenfeld residual tests via test_ph_assumption(), log-log survival plots via create_loglog_plot(), and automatic warnings in create_tte_summary_table().

pharmhand 0.1.13.9000

Improvements

• Added automatic PH violation warnings in create_tte_summary_table() with guidance on stratified or alternative survival models.

pharmhand 0.1.12.9000

Improvements

• Added log-log survival plots for PH diagnostics via create_loglog_plot() and create_km_plot(type = "loglog").

pharmhand 0.1.11.9000

Improvements

• Added test_ph_assumption() to run Schoenfeld residual tests with optional diagnostic plots.

pharmhand 0.1.10.9000

Improvements

• Added detect_floor_ceiling() to flag floor/ceiling effects in PRO scores by visit and treatment arm.

pharmhand 0.1.9.9000

Improvements

• Added NNH (Number Needed to Harm) to AE comparison tables with estimability handling.

pharmhand 0.1.8.9000

Breaking Changes

• build_time_to_event() in the safety report example no longer assumes a 100-day default duration. Supply default_duration when TRTDURD cannot be derived from TRTSDT/TRTEDT.

pharmhand 0.1.7.9000

Improvements

• Standardized NA display via pharmhand.na_string across reporting themes and clinical table styling.

pharmhand 0.1.6.9000

Documentation

• Documented fixed-column handling for the Module 4 table template.

pharmhand 0.1.5.9000

Documentation

• Clarified IQWiG flextable theme formatting and decimal separator behavior.

pharmhand 0.1.4.9000

Documentation

• Finalized IQWiG-compliant CI formatter documentation.

pharmhand 0.1.3.9000

Improvements

• Added validation for default_duration in the safety report example.

Documentation

• Clarified palette resolution for create_km_plot() and CI formatting guidance.
• Added a note on comparison-table entry points in the safety tables vignette.
• Fixed the pkgdown badge URL in the README.

Maintenance

• Updated compliance tests to use flextable’s public header removal API.

pharmhand 0.1.1.9000

Bug Fixes

• Fixed redundant null check in check_gba_compliance() title validation.
• Fixed to_gba_template() list handling that would overwrite same file path.
• Fixed list handling in check_gba_compliance() for unnamed elements.

Improvements

• theme_iqwig() and theme_gba() now apply the decimal_separator parameter to numeric columns (was declared but unused).
• SingleArmStudy now has a treatment_var property for consistency with TwoArmStudy and MultiArmStudy.

Documentation

• Updated G-BA citation to 2024 reference.
• Added missing adae and adrs datasets to README examples.

Maintenance

• Added notes/ to .gitignore.
• Fixed lintr line-length violations.

pharmhand 0.1.0.9000

Breaking Changes

• Renamed create_ae_table() to create_ae_summary_table() for clarity.

Improvements

• format_pvalue() and format_ci() now accept trim and na_string parameters for more flexible output formatting.

Documentation

• Added README.Rmd with rendered demo table and KM plot figures.
• Improved pkgdown site with sandstone theme and dark mode support.
• Simplified Articles dropdown to flat list.
• Added description to baseline-tables vignette.
• Fixed dev status badges visibility with development: mode: unreleased.

Maintenance

• Cleaned up _pkgdown.yml by removing redundant defaults.
• Simplified extra.scss to only flextable dark mode fixes.
• Updated example scripts and vignettes for function rename.

pharmhand 0.0.18.9000

New Features

• Add to_gba_template() to export ClinicalTable/ClinicalReport objects in G-BA Module 4 format.

Tests

• Add coverage for G-BA template export.

pharmhand 0.0.17.9000

New Features

• Add create_hta_module4_table() for standardized G-BA Module 4 tables.

Tests

• Add coverage for Module 4 table creation.

pharmhand 0.0.16.9000

New Features

• Add check_gba_compliance() for pre-export validation of G-BA Module 4 tables.
• Mark IQWiG and G-BA flextable themes with a pharmhand_theme attribute.

Tests

• Add coverage for G-BA compliance checks.

pharmhand 0.0.15.9000

Breaking Changes

• Remove fmt_pvalue() format preset. Use format_pvalue() for IQWiG-compliant p-value formatting.

New Features

• IQWiG-compliant formatting functions: format_pvalue() and format_ci().
• Locale-aware helpers: format_number() and format_percentage().
• New flextable themes: theme_iqwig() and theme_gba().

Tests

• Add coverage for IQWiG formatting and theme functions.

Maintenance

• Remove duplicate formatting helpers to avoid shadowing.

pharmhand 0.0.14.9000

Breaking Changes

• Endpoint classes unified: PrimaryEndpoint, SecondaryEndpoint, and SafetyEndpoint have been replaced with a single Endpoint class. Use the category property (“primary”, “secondary”, “safety”, “exploratory”) instead.

• Study class renamed: OneArmStudy is now SingleArmStudy for clarity.

• New property name: TwoArmStudy@group_var is now TwoArmStudy@treatment_var.

New Features

• Study base class: New abstract Study class provides common properties for all study types.

• MultiArmStudy: New class for studies with 3+ treatment arms.

• StudySet: New class for collections of studies (meta-analysis, NMA).

• Statistical result classes: New StatResult hierarchy with ComparisonResult, MetaResult, and EvidenceGrade for type-safe results.

Documentation

• Added notes/ARCHITECTURE_DECISIONS.md documenting class hierarchy and naming conventions.

pharmhand 0.0.13.9000

Fixes

• Make chef integration resilient to missing stratification columns
• Add data.table guardrails and endpoints validation for chef pipeline usage

Maintenance

• Add internal ph_inform helper for consistent messaging
• Scope custom translation keys to requested locale in list_translation_keys()

pharmhand 0.0.12.9000

Maintenance

• Wrap add_translation assignment to satisfy lintr line-length limits

pharmhand 0.0.11.9000

Fixes

• Avoid Rplot.pdf during tests by building plots instead of printing

Documentation

• Simplify LayeredTable usage default to data.frame()

Maintenance

• Relax R version requirement to >= 4.1.0
• Consolidate development artifacts under notes/ and exclude the directory in .Rbuildignore
• Use paste0() for missing-translation warnings

pharmhand 0.0.10.9000

Fixes

• Enforce unique ADSL USUBJID when joining subgroup variables
• Standardize data frame validation in lab summary, shift, and subgroup tables

Documentation

• Clarify Love plot threshold override for BalanceAssessment inputs
• Clarify multi-level categorical SMD return semantics

Tests

• Add coverage for duplicate ADSL USUBJID checks in subgroup analysis

Maintenance

• Use modifyList for translation merges
• Drop explicit tidyselect import (dplyr re-exports used)

pharmhand 0.0.9.9000

Fixes

• NNT CI crossing zero now returns non-estimable bounds
• NA handling in assert_character_scalar
• Translations merge precedence
• ClinicalContent validation now uses assert_true for clearer errors

Documentation

• Logit/raw SMD method documentation
• Hommel method in adjust_method

Tests

• calculate_nnt coverage

pharmhand 0.0.8.9000

Breaking changes

• Removed save_as_pdf() and PDF output for ClinicalTable format_content.

Fixes

• AE comparison multi-arm columns
• Zebra striping order
• Subgroup vars from adsl
• Required column validation for primary/CFB tables

Improvements

• continuous_threshold parameter + multi-arm warning
• Love plot annotation placement
• Rounded KM y labels/pt constant
• Clarified errors/ph_abort consistency

pharmhand 0.0.7.9000

Maintenance

• Removed cli and scales from Imports and replaced with base R helpers
• Moved data.table and patchwork to Suggests (optional for chef integration and risk tables)
• Removed unused Suggests: yaml, checkmate, webshot2
• Updated example report scripts to use base R messaging

pharmhand 0.0.6.9000

New features

• Added multiplicity adjustment for p-values:
  • adjust_pvalues(): Apply Bonferroni, Holm, BH, or Hochberg corrections
  • create_subgroup_table() now supports adjust_method parameter
• Added calculate_nnt() for Number Needed to Treat calculation with confidence intervals

pharmhand 0.0.5.9000

New features

• Added comprehensive localization support for German HTA (G-BA/AMNOG) dossiers:
  • set_locale() / get_locale(): Set and retrieve current locale (“en” or “de”)
  • tr(): Translate keys to current locale
  • tr_col(): Translate data frame column names
  • get_translations(): Get all translations for a locale
  • add_translation(): Add custom translations
  • reset_custom_translations(): Clear custom translations
  • list_translation_keys(): List all available translation keys
  • Built-in German translations for common clinical trial terms (demographics, AEs, efficacy endpoints, populations)

pharmhand 0.0.4.9000

New features

• Unified treatment variable API across safety and efficacy tables:
  • All table functions now consistently use trt_var = "TRT01P" as default
  • create_primary_endpoint_table(), create_cfb_summary_table(), create_vs_by_visit_table() now accept trt_var parameter
  • Safety tables (create_ae_table()) default changed from “TRT01A” to “TRT01P” for consistency
• Added create_ae_comparison_table() for adverse event comparisons with risk metrics:
  • Accessible via create_ae_table(type = "comparison")
  • Calculates risk difference (RD) and risk ratio (RR) with confidence intervals
  • Supports grouping by SOC, PT, or overall
  • Configurable reference group and incidence threshold filtering
  • Chi-square or Fisher’s exact test for p-values

pharmhand 0.0.3.9000

New features

• Added standardized mean difference (SMD) functions for GBA baseline balance assessment:
  • calculate_smd(): Cohen’s d and Hedges’ g for continuous variables
  • calculate_smd_binary(): Arcsine/logit methods for binary variables
  • calculate_smd_from_data(): Auto-detect variable type from data
  • add_smd_to_table(): Add SMD column to demographics tables
  • assess_baseline_balance(): Comprehensive multi-variable assessment
  • create_love_plot(): Covariate balance visualization (Love plot)
  • BalanceAssessment S7 class for storing assessment results
• Updated baseline-tables vignette with SMD examples

pharmhand 0.0.2.9000

Documentation

• Vignettes now display all generated tables using flextable rendering

• Large tables (SOC/PT, lab shift, medical history, conmeds) are collapsible with <details> tags

• Added dark mode support for pkgdown site tables via CSS overrides for flextable compatibility

• Fixed format_spec() roxygen documentation

Bug fixes

• Fixed vignette data loading to use correct pharmaverseadam dataset names (adtte_onco, adrs_onco)

• Fixed treatment variable usage (ARM vs TRT01P) for oncology datasets

pharmhand 0.0.1.9000

New features

• S7 classes: ADaMData, ClinicalTable, ClinicalPlot, ClinicalReport, StudyResult

• ADaMData wraps ADaM datasets with automatic population filtering

• create_ae_table() generates AE tables with type argument

• create_tte_summary_table() for time-to-event analysis

• create_responder_table() for response rates with CI and comparisons

• create_subgroup_table() and create_forest_plot() for subgroups

• create_km_plot() for Kaplan-Meier curves

• create_cfb_summary_table() for change-from-baseline

• create_lab_shift_table() for lab shifts

• create_demographics_table(), create_disposition_table(), create_population_summary_table()

Reporting

• ClinicalReport collects tables and plots for Word export

• generate_word() and write_docx() export to .docx

Formatting

• format_pvalue(), format_ci(), format_percentage(), format_number()

• FormatSpec and CompositeFormat classes

chef integration

• run_chef_pipeline() executes chef pipelines

• chef_to_analysis_results() converts chef output

pharmhand 0.0.0.9000

New features

• Initial package release
• S7 class system for clinical content (ADaMData, ClinicalTable, ClinicalPlot, ClinicalReport, StudyResult)
• ADaM dataset handling with population filters
• Safety tables: AE overview, SOC/PT summaries, severity, relationship, SAEs, deaths
• Efficacy tables: primary endpoints, subgroup analyses
• Word document generation via officer/flextable
• Integration with chef package for HTA/AMNOG analyses

Documentation

• Added package documentation and examples