NEWS

---

healthiar 0.2.7 (2026-10-12)

• 12 October 2026

Bug Fixes

• summarize_uncertainty() simulated all rows with the mean of the whole input vector. Now each row (e.g. each age group or geographic unit) is simulated with its own central, lower and upper estimates.
• attribute_lifetable() split one life table into pieces if the exposure or the info differed by age (e.g. an exposure distribution by age group), so that the cohorts did not get older from one piece to the next. Now the life tables are only identified by geographic unit, sex, uncertainty and info that differs within the same age.
• attribute_lifetable() showed an error if erf_eq was entered as a function.
• attribute_lifetable() did not detect an age missing in only one life table (e.g. only for males). Now the error names the life table concerned.
• prepare_lifetable() produced negative deaths at high hazard rates, because the probability of dying in one single year was not consistent with the hazard rate entered by the user.
• In attribute_health() with absolute risk and an exposure distribution, the population was summed across exposure categories and therefore impact_per_100k_inhab was underestimated. The same happened in multiexpose(), where the population was summed across exposures.
• multiexpose() merged the exposures across sub-groups defined in info (e.g. urban and rural). Now the exposures are merged within each sub-group.
• compare() silently used the exposure-response function of scenario 2 in both scenarios when erf_eq was entered as a function. Now an error is shown if the functions differ.
• socialize() summed the central, lower and upper estimates if the output of attribute_health() had uncertainty. Now only the central estimates are used.

Documentation

• attribute_lifetable() documents that the life table approach is only applicable to all-cause mortality.
• prepare_lifetable() documents how to enter an open-ended last age group.
• The argument info documents how its values define sub-groups.
• Edits in the vignette chapters on cut-off vs. threshold and on shifting the exposure-response function.

Testing

• 11 additional internal tests. Now a total of 464 tests.

healthiar 0.2.6 (2026-09-11)

• 11 September 2026

New features

• attribute_health() and attribute_lifetable() have the new argument main_results_by to quantify multiple exposure-outcome pairs at once. The dimensions named here (e.g. the info columns) are kept as separate rows in the main results instead of being summed.
• attribute_health() and attribute_lifetable() have the new argument threshold, now clearly distinguished from cutoff. The threshold is the anchor of the exposure-response function (subtracted from the exposure), while the cutoff truncates it.
• monetize(), cba() and get_inflation_factor() now accept year-specific inflation rates (a vector) and not only a constant rate.
• get_discount_factor() now returns one discount factor per value of n_years. Previously, only one value was returned for a vector of years.
• The info columns are now named after the names entered by the user (info_yourname) instead of info_column_1, info_column_2, etc.
• summarize_uncertainty() now shares the random draws across scenarios if seed is NULL, uses truncated (instead of reflected) draws for positive quantities, aggregates macro results from the totals of each simulation and supports the comparison of population impact fractions.

Bug Fixes

• The wrong results of attribute_lifetable() when time_horizon was different to the number of age groups have been corrected. Moreover, time_horizon = 1 produced a reversed sequence of years and max_age was ignored.
• The age groups are now sorted internally and validated (consecutive and of equal length), so that the order of the rows entered by the user does not change the results.
• The wrong behaviour of approach_newborns, silently ignored for deaths and single-year exposure, has been fixed with a warning.
• The wrong use of fraction_lived in prepare_lifetable() has been corrected.
• The wrong behaviour of multiexpose(), multiplying risks across all rows instead of within a row, has been fixed.
• Values of prop_pop_exp that do not sum up to 1 were silently renormalised. The equation in get_pop_fraction() has been corrected.
• In socialize(), the geographic units with a missing social indicator were assigned to the least deprived quantile. They are now dropped with a warning.
• Several defects in standardize() have been fixed (missing bhd_per_100k_inhab_std, wrong standardized exposure, missing grouping by uncertainty columns and wrong use of ref_prop_pop).
• The error "object exp_scen_1 not found" in compare() has been fixed, as well as the duplicated erf_eq in the labels.
• In daly(), the mis-named rate column that was then summed as an impact has been renamed to impact_per_100k_inhab.
• impact_per_100k_inhab was Inf if the population was 0. Now it is NA.
• The binning in prepare_exposure() dropped the extreme cells of the grid.
• prepare_mdi() aborted with a cryptic error if there was any missing value. Now missing values are handled and a warning is shown. Moreover, the results are now assigned and not lost.
• discount() showed an error when using its own documented default of discount_shape.
• monetize() did not keep real_growth_rate among the relevant columns.
• Entering two names in main_results_by was wrongly rejected.
• attribute_mod() did not tag the arguments with new values as entered by the user.
• health_outcome is now validated (error if NULL or not one of the options).
• More comprehensive validation of the input data: no exposure-response function data, increment equal to 0, missing erf_shape or rr_increment, non-consecutive age groups and inconsistent life table approaches in multiexpose().

Other improvements

• get_output() and the projection in attribute_lifetable() are now faster.
• The code has been streamlined overall: new internal function validate_args(), shorter and more consistent variable names, more code comments and removal of dead code.
• The data sets have been recompressed and the duplicated test data removed.
• Changes to prevent errors, warnings and notes in the CRAN checks, e.g. requireNamespace() for suggested packages and tests skipped if a suggested package is not installed.

Documentation

• New vignette chapters on cut-off vs. threshold and on multiple exposure-outcome pairs.
• More extensive explanation of the differences between healthiar and AirQ+ in the life table approach.
• The pkgdown website is now built from the gh-pages branch, so the folder docs is not part of the master branch anymore.
• The badges of the website have been reorganized, the JOSS badge added and the Zenodo DOI moved to the new section "Stay updated".
• A new acknowledgements section has been added.
• The presentation of healthiar at the ISEE conference is now available.
• The minimum R version in the readme file (4.2.0) is now consistent with DESCRIPTION.

Testing

• 26 fake examples in the tests have been replaced with real published studies.
• 71 additional internal tests. Now a total of 453 tests.

healthiar 0.2.5 (2026-08-21)

• 11 August 2026

New features

• attribute_lifetable() and prepare_lifetable() now have the new argument fraction_lived to enter the fraction of year lived specific for each age group. Previously, 0.5 was internally fixed for all age groups.
• attribute_health() has been expanded to now accept the argument cutoff_ together with erf_eq_ in the absolute risk approach (consistent with relative risk risk). A warning is shown in this case because, for absolute risk, it is assumed that erf_eq_ already contains a cutoff.
• Fractional positive deaths and bhd values (with values between 0 and 1) are now allowed for lifetable calculations.
• The attributable deaths can now be quantifed with constant exposure in attribute_lifetable(). Previously, only possible with single year exposure.

Bug Fixes

• Error when installing the package without package dependences has been prevented.
• The wrong behaviour of erf_eq_lower and erf_eq_upper as string in attribute_health(), providing the same result for both, has been corrected.
• The wrong behaviour of socialize() providing deciles for other requested quantiles has been fixed.
• The wrong behaviour of get_paf() for categorical exposure distributions has been fixed.

Documentation

• Bugs in references of function descriptions and vignette have been fixed.
• More clear instructions on how to install the package has been provided.
• Contributing and a Code of Conduct files (sections in pkgdown web pages) have been created to provide information on how to interact with the community.
• The final version of paper published in the Journal of Open Source Software (JOSS) is now available.
• An updated citation of the package is now available refering to the JOSS paper.
• CITATION.cff has been added.
• The function descriptions and vignette have been updated to include instructions of new features.

Testing

• Additional internal tests were added. Now a total of 382 tests.

healthiar 0.2.4 (2026-03-12)

• 12 March 2026

Bug Fixes

• Previously, monetize() used the argument inflation_factor for both adjusting discount_rate and increasing value overtime. This has been resolved by introducing a new argument called real_growth_rate.
  Now, inflation_rate is used exclusively for adjusting discount_rate, while real_growth_rate handles value growth overtime.

Other improvements

• get_discount_factor() previously had inflation_rate as argument. Now, not anymore to keep different concepts separated.
• get_inflation_factor() previously had discount_rate as argument. Now, not anymore to keep different concepts separated.

healthiar 0.2.3 (2026-02-19)

• 19 February 2026

New features

• Using compare() after standardizing results from attribute_health() now works
• For the sub-group analysis, exposure value can be specific for each geo_id_micro AND info column
• The argument social_indicator and geo_id_micro in socialize() can now use tidy data
• monetize() works after compare()
• prepare_exposure() works for gridded population data

Bug Fixes

• Using socialize() and standardize() without ref_prop_pop now works properly
• Now attribute_lifetable() allows all lengths for age group (before, only 100 age categories were allowed)
• The data validation now identifies correctly whether the argument n_years was entered by the user or not

Testing

• 31 additional internal tests

Other improvements

• The argument n_years has no default value anymore in monetize() and discount()
• Error message if monetize() is used after compare() with different year of analysis or baseline health data in the two scenarios (not attributable to policy intervention)
• More complete warning message if absolute risk and cutoff is provided

healthiar 0.2.2 (2026-01-09)

• 08 January 2026

Bug Fixes

• results_raw now stratifies by info arguments

Documentation

• New structure of vignette (by topic instead of by function)
• Amendments in readme file in terms of contents and structure including URL to new healthiar website
• Updated and fixed citation

Testing

• Around 70 additional internal tests for attribute_health()

healthiar 0.2.1 (2025-11-11)

• 06 November 2025

Bug Fixes

• time_horizon did not work in attribute_lifetable()
• population was not summed correct in attribute_lifetable()
• health_detailed in attribute_health() provided a duplicated rows for results by geo_id_micro

Other improvements

• Better sampling in summarize_uncertainty() using RNG and package parallel
• Other changes to comply with with manual review of CRAN

healthiar 0.1.1

• 19 September 2025

Other improvements

• Small changes to comply with with automatic review of CRAN

healthiar 0.1.0

• 19 September 2025 (1st submission to CRAN, automatic review)

New Features

• New function get_inflation_rate()
• New argument in cba(): inflation rate
• geo ids can now have different number of exposure categories

Bug Fixes

• Argument time_horizon was not working

Renamings

• discount_years was renamed to n_years
• inflatoin was renamed to inflation_rate
• attribute_by_sim was renamed to impact_by_sim
• positive_impact was renamed to impact_benefit

Other improvements

• attribute_by_sim_disaggregated is not anymore available as output
• attribute_by_geo_id_micro is not anymore available as output
• Faster performance of summarize_uncertainty()

healthiar 0.0.4

• 01 September 2025

New Features

• New function prepare_lifetable()
• Argument info can have different values
• Argument info also available in monetize()
• Data validation in socialize()

Renamings

• geo_id_disaggregated is now geo_id_micro
• geo_id_aggregated is now geo_id_macro
• output_attribute_1 in attribute_mod() is now output_attribute
• Arguments with suffix _1 and _2 in compare() are now _scen_1 and _scen_2
• Arguments with exposure_ are now exp_
• Arguments with suffix _1 and _2 in multiexpose() are now _exp_1 and _exp_2
• And many other renames in output columns and variables

Other improvements

• More consistent sum of results
• Higher speed because of shorter and/or optimized code in attribute_...() functions

healthiar 0.0.3

• 14 July 2025

New Features

• No lists (vectors) as input for multiple geo units
• Arguments age_group & sex in attribute_health() and attribute_lifetable()
• Detailed results of simulations by geo unit in summarize_uncertainty()
• New structure of input_args

Bug Fixes

• Fixed bug in socialize()
• Fixed bug attribute_lifetable()

Renamings

• Rename: impact_raw is now results_raw
• Rename: listed_output_attribute is now output_attribute

Other improvements

• Higher performance of attribute_()
• Data validation in monetize()
• Data validation in summarize_uncertainty()
• More validation in attribute_...()

healthiar 0.0.2

• 02 June 2025

New Features

• New function standardize()
• Improved and corrected function socialize()
• Expanded usability of summarize_uncertainty()
• Enabled single exposure in absolute risk

Bug Fixes

• Fixed compare() to avoid errors when using erf_eq
• Fixed warning in socialize()
• Fixed bug in prepare_exposure()

Renamings

• Arguments results or output_healthiar become output_attribute
• Internal variable rr_conc (visible in results of attribute functions) becomes rr_at_exp

Other improvements

• Input data validation in compare()
• Warning if cutoff is NULL and 0 as default

healthiar 0.0.1

• 05 May 2025

New Features

• Version number was added
• Columns with health impacts were moved to the front in results

Bug Fixes

• Results for impact per 100k inhab. have been corrected
• Exposure lower than cut-off must result in zero health impact

Renamings

• get_pop_fraction() becomes intern function
• get_mdi() was renamed to prepare_mdi()
• get_multiexposure() was renamed to multiexpose()
• get_daly was renamed to daly()

Other improvements

• Custom warning and error messages are now available in attribute_health()