Added a second version of the ICD10-AM score including U codes to "increases the capture of some chronic conditions up to 15-fold", according to the Australian Institute of Health and Welfare (#72, thanks to @DesiQuintans for contributing this);
Updated Swedish version of the Charlson comorbidity score according to the corrigendum published online at: https://doi.org/10.2147/CLEP.S425901;
Removed startup message highlighting changes in the package API introduced with {comorbidity} 1.0.0;
Updated hex sticker;
Minor improvements to {pkgdown} website accessibility;
Fixed invalid language field in DESCRIPTION file (#75).
Adjusted name of weights (leftover from #53, thanks @sciatro for reporting this). This bug did not seem to affect any calculation.
Updated CITATION file to use the new interface.
Some internal housekeeping (removed old documentation, updated GitHub actions to new version, etc.).
Updated maintainer e-mail.
Updated hex sticker.
Improved documentation for arguments of the comorbidity() and score() functions (#61, thanks @dsjoho).
Fixed broken references to the old vignettes' names (#61, thanks @dsjoho).
Overall, package documentation has been tidied up a bunch.
The startup message pointing out changes in the API now appears less often (10% probability).
The copd and ami comorbidities for the Charlson index have been renamed to cpd and mi, respectively (#53, thanks @DrYan1102). Please be aware that this might break some old code if you were selecting comorbidities by name.
New dataset: ICD10-CM, 2022 version, named icd10cm_2022.
The startup message pointing out changes in the API now appears less often (40% probability).
Fixed a bug that was causing all comorbidities to be assigned a value of one when there was missing data in the code column (#50), thanks @Chris-M-P for reporting this.
comorbidity version 1.0.0 is a release that substantially modifies and updates the package API.
There are a lot of improvements in terms of performance, supported algorithms, and user-facing functions; see e.g. this thread on GitHub.
Specific user-facing changes, including examples of the new API, are discussed in more detail here: https://ellessenne.github.io/comorbidity/articles/C-changes.html
If required, you can revert to the previous release by installing from GitHub: remotes::install_github("ellessenne/[email protected]").
A startup message pointing out the changes in the API is now displayed with a 50% probability when attaching the package and will be removed in future releases.
New contributors: Sing Yi Chia and Edmund Teo.
comorbidity now accepts data.table objects as input (#23);
comorbidity can deal with non-syntactically valid names (#25);
Reduced number of random tests, as R CMD check was taking too long on some platforms getting killed on CRAN (as reported by Kurt Hornik).
data.table that only occurred when the input dataset had extra columns other than id and code.The melt generic in data.table has been deprecated; the dependency on reshape2 has now been removed;
Fixed bugs introduced by switching to data.table in comorbidity 0.5.0: column names id and code were expecting these specific names, now fixed and behaves as expected.
comorbidity is now 10+ times faster, thanks to Jonathan Williman contributing code based on the data.table package;
Fixed typo in vignette regarding weighting algorithm for the AHRQ Elixhauser comorbidity score (#14, thanks to @cornflakegrl);
Added pkgdown website: https://ellessenne.github.io/comorbidity.
Modified the behaviour of the assign0 argument after further discussion with Anders Alexandersson in Issue #9: now there is no default, forcing the user to decide whether to apply a hierarchy of comorbidity codes or not. This will make the algorithm more transparent to the end user, allowing an informed choice. See ?comorbidity::comorbidity and vignette("comorbidityscores", package = "comorbidity") for further details on the hierarchy being applied.
comorbidity now returns two Elixhauser scores, one computed using the algorithm of van Walraven et al. (2009) and a second one computed using the AHRQ algorithm (Moore et al., 2017). Thanks to Yumiko Abe-Jones for feedback and the discussion regarding weighted Elixhauser scores.
More information can be found on the package vignette: vignette("comorbidityscores", package = "comorbidity").
assign0 argument of comorbidity now defaults to FALSE;comorbidity function:
assign0 now explains in details what hierarchy of comorbidities is applied;comorbidity is faster, with a conservative estimated speed-up of >60%;comorbidity.The score argument from comorbidity has been split into score and icd. For instance, the command comorbidity(x = x, id = "id", code = "code", score = "charlson_icd10") has to be modified as r comorbidity(x = x, id = "id", code = "code", score = "charlson", icd = "icd10"). The default value of icd is icd10, for ICD-10 codes, and possible values are icd10 and icd9.
nhds2010 and australia10 datasets, imported from Stata version 15.Bug fix:
comorbidity() (@corinne-riddell, #5);citation("comorbidity") now returns a properly formatted entry.comorbiditysample_diag_icd10() function renamed back to sample_diag(), as now can simulate ICD-9-CM codes tooicd10_2009 and icd10_2011 with ICD-10 codes, 2009 and 2011 versions (respectively)sample_diag is now sample_diag_icd10 and simulates proper ICD-10 codescomorbidity as it now can compute more than just the Charlson scorecharlson is marginally fastercharlson using only base R functionsNEWS.md file to track changes to the packagetestthat