Package ‘medicalrisk’ - The Comprehensive R …

1 Package medicalrisk August 29, 2016 TypePackageTitleMedical Risk and comorbidity Tools for ICD-9-CM risk estimates and comorbidity flags from ICD-9-CMcodes available in administrative medical datasets. The Package supportsthe Charlson comorbidity Index, the Elixhauser Comorbidityclassification, the Revised Cardiac Risk Index, and the Risk StratificationIndex. Methods are table-based, fast, and use the 'plyr' Package , soparallelization is possible for large jobs. Also includes a sample ofreal ICD-9 data for 100 patients from a publicly available (>= )Importsplyr (>= ), reshape2, hashSuggeststestthat, knitr, ggplot2, gridExtraLicenseGPL-3 | file McCormick [aut, cre],Thomas Joseph [aut]MaintainerPatrick 15:45:3112charlson_listRtopics documented:charlson_list.

2 2charlson_weights ..3charlson_weights_orig ..4elixhauser_list ..5generate_charlson_index_df ..5generate_comorbidity_df ..6icd9cm_charlson_deyo ..7icd9cm_charlson_quan ..8icd9cm_charlson_romano ..9icd9cm_elixhauser_ahrq37 .. 11icd9cm_elixhauser_quan .. 12icd9cm_list .. 13icd9cm_rcri .. 14icd9cm_sessler_rsi .. 15melt_icd9list .. 16merge_icd9_dx_and_procs .. 17rsi_beta_1yrpod .. 17rsi_beta_30dlos .. 18rsi_beta_30dpod .. 18rsi_beta_inhosp .. 19rsi_sample_data .. 19rsi_sample_results .. 20sessler_get_single_beta .. 21verify_sessler_rsi.

3 22vt_inp_sample .. 22 Index24charlson_listList of Charlson comorbiditiesDescriptionList of Charlson comorbiditiesUsagecharlson_listFormatA list, with one column for each comorbidity ; value is a textual descriptioncharlson_weights3 Examples# List the strings used to refer to Charlson comorbiditiesnames(charlson_list)# List descriptions of comorbiditiescharlson_listcharlson_weigh tsMap of Charlson comorbidity categories to revised weightsDescriptionList that links the Charlson comorbidity categories to revised weights as calculated by Schneeweissin Table 4 of his list, with Charlson comorbidities as names and weight as valueDetailsRevised Schneeweiss weights.

4 0 = Connective tissue dz, Ulcer1 = MI, PVD, CVD, Diabetes, Hemiplegia2 = CHF, Chronic pulm dz, Mild liver dz, Diabetes with end organ damage, Any tumor, Leukemia,Lymphoma3 = Dementia, Mod or severe renal dz4 = Moderate or severe liver dz, AIDS6 = Metastatic solid tumorReferences1. Schneeweiss S, Wang PS, Avorn J, Glynn RJ: Improved comorbidity adjustment for predictingmortality in Medicare populations. Health services research 2003; 38:1103 Alsocharlson_weights_orig,icd9cm_charlso n_deyo,icd9cm_charlson_romano,icd9cm_cha rlson_quan,melt_icd9listExamplescharlson _weights["dementia"]4charlson_weights_or igcharlson_weights_origMap of Charlson comorbidity categories to weightsDescriptionList that links the Charlson comorbidity categories to the original weights (specified in the originalCharlson paper, Table 3)Usagecharlson_weights_origFormatA list, with Charlson comorbidities as names and weight as valueDetailsOriginal Weights.

5 1 = MI, CHF, PVD, CVD, Dementia, Chronic pulm dz, Connective tissue dz, Ulcer, Mild liver dz,Diabetes2 = Hemiplegia, Mod or severe renal dz, Diabetes with end organ damage, Any tumor, Leukemia,Lymphoma3 = Moderate or severe liver dz6 = Metastatic solid tumor, AIDSR eferences1. Charlson ME, Pompei P, Ales KL, MacKenzie CR: A new method of classifying prognosticcomorbidity in longitudinal studies: development and validation. Journal of chronic diseases 1987;40:373-83 Alsocharlson_weights,icd9cm_charlson_dey o,icd9cm_charlson_romano,icd9cm_charlson _quan,melt_icd9listExamplescharlson_weig hts_orig["aids"]elixhauser_list5elixhaus er_listList of Elixhauser comorbiditiesDescriptionList of Elixhauser comorbiditiesUsageelixhauser_listFormatA list, with one column for each comorbidity .

6 Value is a textual descriptionExamples# List the strings used to refer to Elixhauser comorbiditiesnames(elixhauser_list)# List descriptions of comorbiditieselixhauser_listgenerate_cha rlson_index_dfCalculate the Charlson comorbidity IndexDescriptiongenerate_charlson_index_ dfmerges a data frame of Charlson comorbidities withcharlson_weightsand sums the results per (df, idvar = "id",weights = medicalrisk ::charlson_weights)Argumentsd fa data frame with ID columnidvarand logical columns for each comorbidity ,such as that generated bygenerate_comorbidity_dfidvarstring with name of ID variable withindfweightsdefaults tocharlson_weights6generate_comorbidity_ dfValuea dataframe with two columns,idvarand"index"See Alsogenerate_comorbidity_df,charlson_wei ghts,charlson_weights_origExamples# calculate Charlson comorbidity Index for all patients in the \code{\link{vt_inp_sample}}data(vt_inp_s ample)generate_charlson_index_df(generat e_comorbidity_df(vt_inp_sample))

7 Generate_comorbidity_dfGenerate a comorbidity dataframeDescriptionMerges a given DF of IDs and ICD-9-CM codes to one of the ICD9CM maps, removes redundantcomorbidities, and returns a (df, idvar = "id", icd9var = "icd9cm",icd9mapfn = icd9cm_charlson_quan, .progress = "none", .parallel = FALSE,.paropts = NULL)Argumentsdfa data frame with at least two columns, specified with name of ID variable withindf(defaults to "id")icd9varstring with name of ICD code variable withindf(defaults toicd9cm)icd9mapfnFunction to generate comorbidity data frame from ICD-9 codes (defaults toicd9cm_charlson_quan).

8 Progresspassed toddplyicd9cm_charlson_deyo7 DetailsRedundancy rules: * If "tumor" and "mets", only "mets" will be returned. * If "htn" and "htncx",only "htncx" will be returned. * If "dm" and "dmcx", only "dmcx" will be returned. * If "liver" and"modliver", only "modliver" will be Walraven has a modification adopted here where the following "dmcx" codes are downgradedto "dm" if the specific DM complication is separately coded: * D2(49|50)4x is DM w renal *D2(49|50)6x is DM w neuro * D2(49|50)7x is DM w PVDC ases without any comorbidities will not appear in the returned data dataframe with columnidvarand a logical column for each comorbidityExamplescases <- (id=c(1,1,1,2,2,2,2,2),icd9cm=c("D20206" ,"D24220","D4439","D5064","DE8788","D404 03","D1960","D1958"))generate_comorbidit y_df(cases)# generate categories for patients in the \code{\link{vt_inp_sample}}generate_como rbidity_df(vt_inp_sample)

9 # in this example, D25071 is reduced to "dm" from "dmcx" because D4439 already codes perivasc# also, D20206 "tumor" and D1970 "mets" lead to just "mets"# D25001 and D25040 are just "dmcx"# D45621 and D570 are just "modliver"cases <- (id=c(1,1,1,1,2,2,2,2),icd9cm=c("D1970", "D20206","D25071","D4439","D25001","D250 40","D45621","D570"))generate_comorbidit y_df(cases)icd9cm_charlson_deyoCreate Deyo map of ICD-9-CM to Charlson comorbiditiesDescriptionFunction that generates a data frame linking ICD-9-CM codes to the Charlson comorbidity cate-gories using the Deyo (icd9)Argumentsicd9a unique character vector of ICD-9-CM codes8icd9cm_charlson_quanDetailsNOTE: The input vector of ICD-9-CM codes must be unique, because the output dataframe usesthe ICD-9-CM code as regular expressions created from the paper by Deyo in codes must have periods removed.

10 Diagnostic codes are prefixed with D while proce-dure codes are prefixed with P . So, diagnostic be"D40403".ValueA data frame, with ICD9 codes as row names and one logical column for each comorbidity incharlson_listReferences1. Deyo RA, Cherkin DC, Ciol MA: adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. Journal of clinical epidemiology 1992; 45:613-9 Alsoicd9cm_charlson_quan,icd9cm_charlson _romano,icd9cm_elixhauser_quan,icd9cm_el ixhauser_ahrq37,charlson_weights,Example s# Identify Charlson categories in ICD-9-CM listingcases <- (id=c(1,1,1,2,2,2),icd9cm=c("D20206","D2 4220","D4439","D5064","DE8788","D40403") )cases_with_cm <- merge(cases, icd9cm_charlson_deyo(levels(cases$icd9cm )), "icd9cm", " ", )# generate crude comorbidity summary for each patientlibrary(plyr)ddply(cases_with_cm. )