Nothing
R/icd9cm.R
In medicalrisk: Medical Risk and Comorbidity Tools for ICD-9-CM Data
Defines functions
icd9cm_charlson_deyo
icd9cm_charlson_romano
icd9cm_charlson_quan
icd9cm_elixhauser_quan
icd9cm_elixhauser_ahrq37
icd9cm_rcri
melt_icd9list
merge_icd9_dx_and_procs
generate_comorbidity_df
generate_charlson_index_df
Documented in
generate_charlson_index_df
generate_comorbidity_df
icd9cm_charlson_deyo
icd9cm_charlson_quan
icd9cm_charlson_romano
icd9cm_elixhauser_ahrq37
icd9cm_elixhauser_quan
icd9cm_rcri
melt_icd9list
merge_icd9_dx_and_procs
#' Create Deyo map of ICD-9-CM to Charlson comorbidities
#'
#' Function that generates a data frame linking ICD-9-CM codes to the Charlson
#' comorbidity categories using the Deyo mapping.
#'
#' NOTE: The input vector of ICD-9-CM codes must be unique, because the output dataframe
#' uses the ICD-9-CM code as row.name.
#'
#' Uses regular expressions created from the paper by Deyo in 1992.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{charlson_list}}
#' @references 1. 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
#' \url{http://www.ncbi.nlm.nih.gov/pubmed/1607900}
#
#' @seealso \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_charlson_romano}},
#' \code{\link{icd9cm_elixhauser_quan}}, \code{\link{icd9cm_elixhauser_ahrq37}},
#' \code{\link{charlson_weights}},
#' @examples
#' # Identify Charlson categories in ICD-9-CM listing
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_charlson_deyo(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_charlson_deyo <- function(icd9) {
data.frame(
row.names = icd9,
mi = grepl('^D(410|412)', icd9),
chf = grepl('^D428[0-9]', icd9),
perivasc = grepl('^(D(4439|441[0-9]|7854|V434)|P3848)', icd9),
cvd = grepl('^D43[0-8]', icd9),
dementia = grepl('^D290[0-9]', icd9),
chrnlung = grepl('^D(49[0-6]|50[0-5]|5064)', icd9),
rheum = grepl('^D(710[0,1,4]|714[0-2]|71481|725)', icd9),
ulcer = grepl('^D53[1-4]', icd9),
liver = grepl('^D571[2456]', icd9),
dm = grepl('^D250[0-3,7]', icd9),
dmcx = grepl('^D250[4-6]', icd9),
para = grepl('^D(3441|342)', icd9),
renal = grepl('^D(58[2568]|583[0-7])', icd9),
tumor = grepl('^D(1[4-6]|17[0-24-9]|18|19[0-5]|20[0-8])', icd9),
modliver = grepl('^D(456[0-2]|572[2-8])', icd9),
mets = grepl('^D19[6-9]', icd9),
aids = grepl('^D04[2-4]', icd9))
}
#' Create Romano map of ICD-9-CM to Charlson comorbidities
#'
#' Function that creates a dataframe which links ICD-9-CM codes to the Charlson
#' comorbidity categories using the Romano mapping.
#'
#' NOTE: The input vector of ICD-9-CM codes must be unique, because the output dataframe
#' uses the ICD-9-CM code as row.name.
#'
#' Uses regular expressions created from the paper by Romano in 1993.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{charlson_list}}
#' @references 1. Romano PS, Roos LL, Jollis JG: Adapting a clinical comorbidity index for
#' use with ICD-9-CM administrative data: differing perspectives. Journal of
#' clinical epidemiology 1993; 46:1075-9; discussion 1081-90
#' \url{http://www.ncbi.nlm.nih.gov/pubmed/8410092}
#
#' @seealso \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_charlson_deyo}},
#' \code{\link{icd9cm_elixhauser_quan}}, \code{\link{icd9cm_elixhauser_ahrq37}},
#' \code{\link{charlson_weights}},
#' @examples
#' # Identify Charlson categories in ICD-9-CM listing
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_charlson_romano(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_charlson_romano <- function(icd9) {
data.frame(
row.names = icd9,
mi = grepl('^D(410|412)', icd9),
chf = grepl('^D(402[01]1|40291|425|428|4293)', icd9),
perivasc = grepl('^(D(44[012]|443[1-9]|4471|7854)|P(38(1[3468]|3[3468]|4[3468])|392[2-69]))', icd9),
cvd = grepl('^(D(36234|43[0-6]|437[019]|438|7814|7843|9970)|P38[14]2)', icd9),
dementia = grepl('^D(290[0-9]|331[0-2])', icd9),
chrnlung = grepl('^D(4150|416[89]|49[1-46])', icd9),
rheum = F,
ulcer = grepl('^D53[1-4]', icd9),
liver = grepl('^D571[25689]', icd9),
dm = grepl('^D250[0-3]', icd9),
dmcx = grepl('^D250[4-9]', icd9),
para = grepl('^D(34[2,4])', icd9),
renal = grepl('^(D(58[56]|V420|V451|V56)|P(39(27|42|9[3-5])|5498))', icd9),
tumor = grepl('^(D(1[4-6]|17[0-14-9]|18|19[0-5]|20[0-8]|273[03]|V1046)|P(605|6241))', icd9),
modliver = grepl('^(D(456[0-2]|572[2-4])|P(391|4291))', icd9),
mets = grepl('^D19[6-9]', icd9),
aids = grepl('^D04[2-4]', icd9))
}
#' Create Quan map of ICD-9-CM to Charlson comorbidities
#'
#' Function that creates a dataframe that links ICD-9-CM codes to the Charlson comorbidity
#' categories using Quan's method.
#'
#' NOTE: The input vector of ICD-9-CM codes must be unique, because the output dataframe
#' uses the ICD-9-CM code as row.name.
#'
#' Uses regular expressions created from the paper by Quan in 2005.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{charlson_list}}
#' @references 1. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi
#' J-C, Saunders LD, Beck CA, Feasby TE, Ghali WA: Coding algorithms for
#' defining comorbidities in ICD-9-CM and ICD-10 administrative data. Medical
#' care 2005; 43:1130-9
#' \url{http://www.ncbi.nlm.nih.gov/pubmed/16224307}
#
#' @seealso \code{\link{icd9cm_charlson_deyo}}, \code{\link{icd9cm_charlson_romano}},
#' \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_elixhauser_quan}}
#' @examples
#' # Identify Charlson categories in ICD-9-CM listing
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_charlson_quan(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_charlson_quan <- function(icd9) {
data.frame(
row.names = icd9,
mi = grepl('^D(410|412)', icd9),
chf = grepl('^D(39891|402[019]1|404[019][13]|425[4-9]|428)', icd9),
perivasc = grepl('^D(0930|4373|44[01]|443[1-9]|4471|557[19]|V434)', icd9),
cvd = grepl('^D(36234|43[0-8])', icd9),
dementia = grepl('^D(290|2941|3312)', icd9),
chrnlung = grepl('^D(416[89]|49|50[0-5]|5064|508[18])', icd9),
rheum = grepl('^D(4465|710[0-4]|714[0-2]|7148|725)', icd9),
ulcer = grepl('^D53[1-4]', icd9),
liver = grepl('^D(070[23][23]|070[45]4|070[69]|57[01]|573[3489]|V427)', icd9),
dm = grepl('^D250[0-389]', icd9),
dmcx = grepl('^D250[4-7]', icd9),
para = grepl('^D(3341|34[23]|344[0-69])', icd9),
renal = grepl('^D(403[019]1|404[019][23]|582|583[0-7]|58[56]|5880|V420|V451|V56)', icd9),
tumor = grepl('^D(1[4-6]|17[0-24-9]|18|19[0-5]|20[0-8]|2386)', icd9),
modliver = grepl('^D(456[0-2]|572[2-8])', icd9),
mets = grepl('^D19[6-9]', icd9),
aids = grepl('^D04[2-4]', icd9))
}
#' Create Quan map of ICD-9-CM to Elixhauser comorbidities
#'
#' Function to make a dataframe that links ICD-9-CM codes to the Elixhauser comorbidity
#' categories using the Quan mapping.
#'
#' Uses regular expressions created from the Quan paper from 2005.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' Some ICD-9-CM codes will correspond to more than one category. For example,
#' \code{404.03} (Hypertensive heart and chronic kidney disease ... stage V) is
#' in both \code{chf} and \code{renlfail} categories.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{elixhauser_list}}
#' @references 1. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi
#' J-C, Saunders LD, Beck CA, Feasby TE, Ghali WA: Coding algorithms for
#' defining comorbidities in ICD-9-CM and ICD-10 administrative data. Medical
#' care 2005; 43:1130-9
#' \url{http://www.ncbi.nlm.nih.gov/pubmed/16224307}
#
#' @seealso \code{\link{icd9cm_charlson_deyo}}, \code{\link{icd9cm_charlson_romano}},
#' \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_elixhauser_ahrq37}}
#' @examples
#' # Identify Elixhauser categories
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_elixhauser_quan(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_elixhauser_quan <- function(icd9) {
data.frame(
row.names = icd9,
chf = grepl('^D(39891|402[019]1|404[019][13]|425[4-9]|428)', icd9),
arrhythmia = grepl('^D(426([079]|13)|426(10|12)|427[0-46-9]|7850|9960[14]|V450|V533)', icd9),
valve = grepl('^D(0932|39[4567]|424|746[3-6]|V422|V433)', icd9),
pulmcirc = grepl('^D(415[01]|416|417[089])', icd9),
perivasc = grepl('^D(0930|4373|44[01]|443[1-9]|4471|557[19]|V434)', icd9),
htn = grepl('^D401', icd9),
htncx = grepl('^D40[2-5]', icd9),
para = grepl('^D(3341|34[23]|344[0-69])', icd9),
neuro = grepl('^D(3319|332[01]|333[45]|33392|33[45]|3362|34[015]|348[13]|7803|7843)', icd9),
chrnlung = grepl('^D(416[89]|49|50[0-5]|5064|508[18])', icd9),
dm = grepl('^D250[0-3]', icd9),
dmcx = grepl('^D250[4-9]', icd9),
hypothy = grepl('^D(2409|24[34]|246[18])', icd9),
renlfail = grepl('^D(403[019]1|404[019][23]|58[56]|5880|V420|V451|V56)', icd9),
liver = grepl('^D(070[23][23]|070[45]4|070[69]|456[0-2]|57[01]|572[2-8]|573[3489]|V427)', icd9),
ulcer = grepl('^D53[1-4][79]', icd9),
aids = grepl('^D04[2-4]', icd9),
lymph = grepl('^D(20[0-2]|2030|2386)', icd9),
mets = grepl('^D19[6-9]', icd9),
tumor = grepl('^D(1[4-6]|17[0-24-9]|18|19[0-5])', icd9),
rheum = grepl('^D(446|7010|710[0-489]|7112|714|7193|72[05]|7285|72889|72930)', icd9),
coag = grepl('^D(286|287[13-5])', icd9),
obese = grepl('^D2780', icd9),
wghtloss = grepl('^D(26[0-3]|7832|7994)', icd9),
lytes = grepl('^D(2536|276)', icd9),
bldloss = grepl('^D2800', icd9),
anemdef = grepl('^D(280[1-9]|281)', icd9),
alcohol = grepl('^D(2652|291[1-35-9]|303[09]|3050|3575|4255|5353|571[0-3]|980|V113)', icd9),
drug = grepl('^D(292|304|305[2-9]|V6542)', icd9),
psych = grepl('^D(2938|295|296[0145]4|29[78])', icd9),
depress = grepl('^D(296[235]|3004|309|311)', icd9))
}
#' Create AHRQ v3.7 map of ICD-9-CM to Elixhauser comorbidities
#'
#' Function makes a dataframe that links ICD-9-CM codes to the Elixhauser comorbidity
#' categories using the AHRQ v3.7 mapping.
#'
#' Uses regular expressions based on the file "comformat2012-2013.txt" from AHRQ.
#'
#' The Agency for Healthcare Research and Quality (AHRQ) has developed
#' Comorbidity Software as part of the Healthcare Cost and Utilization Project
#' (HCUP). The software was developed to report on the comorbidity measures
#' reported by Elixhauser (1998).
#'
#' The AHRQ software has two parts, one that classifies ICD-9-CM codes by
#' comorbidity, and another that performs heuristics to eliminate duplicate
#' comorbidities and ignore comorbidities which are the primary reason for the
#' hospital visit, as per the DRG.
#'
#' This table is a translation of the first part of the software, the
#' classifier, as implemented in the SAS file \code{Comformat2012-2013.txt}.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{elixhauser_list}}
#' @references 1. \url{http://www.hcup-us.ahrq.gov/toolssoftware/comorbidity/comorbidity.jsp}
#'
#' @seealso \code{\link{icd9cm_charlson_deyo}}, \code{\link{icd9cm_charlson_romano}},
#' \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_elixhauser_quan}}
#' @examples
#' # Identify Elixhauser categories
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_elixhauser_ahrq37(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_elixhauser_ahrq37 <- function(icd9) {
# the below is taken from comformat2012-2013.txt
# htncx: htncx, htnpreg, htnwochf, hrenworf, hhrwohrf
# htncx & chf: htnwchf, hhrwchf
# htncx & renlfail: hrenwrf, hhrwrf
# htncx & chf & renlfail: hhrwhrf
htnpreg <- grepl('^D6422', icd9)
htnwochf <- grepl('^D(402[019]0|405[019]9)', icd9)
htnwchf <- grepl('^D402[019]1', icd9)
hrenworf <- grepl('^D(403[019]0|405[019]1|6421)', icd9)
hrenwrf <- grepl('^D403[019]1', icd9)
hhrwohrf <- grepl('^D404[019]0', icd9)
hhrwchf <- grepl('^D404[019]1', icd9)
hhrwrf <- grepl('^D404[019]2', icd9)
hhrwhrf <- grepl('^D404[019]3', icd9)
ohtnpreg <- grepl('^D642[79]', icd9)
data.frame(
row.names = icd9,
chf = grepl('^D(39891|428)', icd9) | htnwchf | hhrwchf | hhrwhrf,
arrhythmia = F,
valve = grepl('^D(0932|39[456]|397[019]|424|746[3-6]|V422|V433)', icd9),
pulmcirc = grepl('^D(4151|416|4179)', icd9),
perivasc = grepl('^D(44[012]|443[1-9]|4442|4471|449|557[19]|V434)', icd9),
htn = grepl('^D(401[19]|6420)', icd9),
htncx = grepl('^D(4010|4372)', icd9) |
htnpreg | htnwochf | hrenworf | hhrwohrf |
htnwchf | hhrwchf | hrenwrf | hhrwrf | hhrwhrf,
para = grepl('^D(34[234]|438[2-5]|78072)', icd9),
neuro = grepl('^D(33[01]|3320|333[457]|33385|33394|33[45]|3380|34[0157]|6494|7687|7803|78097|7843)', icd9),
chrnlung = grepl('^D(49|50[0-5]|5064)', icd9),
dm = grepl('^D(250[0-3]|6480|249[0-3])', icd9),
dmcx = grepl('^D(250[4-9]|7751|249[4-9])', icd9),
hypothy = grepl('^D(243|244[01289])', icd9),
renlfail = grepl('^D(585[3-9]|586|V420|V451|V56)', icd9) | hrenwrf | hhrwrf,
liver = grepl('^D(070[23][23]|070[45]4|456[0-2]|571[02-9]|572[38]|5735|V427)', icd9),
ulcer = grepl('^D(53[1234][456]1|53[1234]7[01]|53[1234]91)', icd9),
aids = grepl('^D04[2-4]', icd9),
lymph = grepl('^D(20[01]|202[0-35-9]|203|2386|2733)', icd9),
mets = grepl('^D(19[6-8]|199[01]|2097|78951)', icd9),
tumor = grepl('^D(1[4-6]|17[012459]|18|19[0-5]|209[0-3]|2580)', icd9),
rheum = grepl('^D(7010|71[04]|720|725)', icd9),
coag = grepl('^D(286|287[13-5]|6493|28984)', icd9),
obese = grepl('^D(2780|6491|V85[34]|V8554|79391)', icd9),
wghtloss = grepl('^D(26[0-3]|7832)', icd9),
lytes = grepl('^D276', icd9),
bldloss = grepl('^D(2800|6482)', icd9),
anemdef = grepl('^D(280[1-9]|281|285[29])', icd9),
alcohol = grepl('^D(291[0-35-9]|303[09]|3050)', icd9),
drug = grepl('^D(292|304|305[2-9]|6483)', icd9),
psych = grepl('^D(29[5678]|2991)', icd9),
depress = grepl('^D(3004|30112|309[01]|311)', icd9))
}
#' Create Map of ICD-9-CM to Revised Cardiac Risk Index classes
#'
#' Function to generate data frame that links ICD-9-CM codes to the RCRI comorbidity categories.
#'
#' Lee et al in 1999 published a "Revised Cardiac Risk Index" based on the work
#' on Goldman in 1997. The RCRI is used to determine the major cardiac
#' complication risk for a patient about to undergo major noncardiac surgery.
#' The six predictors that make up the RCRI are: 1. high-risk surgery 2. history
#' of ischemic heart disease 3. history of congestive heart failure 4. history
#' of cerebrovascular disease 5. preoperative treatment with insulin 6.
#' preoperative serum creatinine with Cr > 2 mg/dL.
#'
#' In 2005 Boersma et al demonstrated that the Lee indexed can be adapted to use
#' administrative data to predict cardiovascular mortality. They used the
#' following for each point above: 1. retroperitoneal, intrathoracic, or
#' suprainguinal vascular procedure; 2. Ischemia: ICD-9 codes 410.*, 411.*,
#' 412.*, 413.*, 414.*; 3. CVA: ICD-9 428.*; 4. CHF: ICD-9 943.0; 5. DM: ICD-9
#' 425.0; 6. Renal: ICD-9 958.0.
#'
#' This function merges the ICD-9 guidelines
#' used by Boersma with some of the other ICD-9 classifiers in this package.
#' This data set uses the following for each aspect of the RCRI:
#' 1. procedure is left to you
#' 2. 'ischemia' as defined in Boersma
#' 3. 'cvd' as defined by Quan in \code{\link{icd9cm_charlson_quan}}
#' 4. 'chf' as defined by AHRQ in \code{\link{icd9cm_elixhauser_ahrq37}}
#' 5. 'dm' as defined by AHRQ (both 'dm' and 'dmcx')
#' 6. renlfail' as defined by AHRQ.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and logical columns for
#' \code{chf}, \code{cvd}, \code{dm}, \code{ischemia}, and \code{renlfail}.
#' @references 1. Lee TH, Marcantonio ER, Mangione CM, Thomas EJ, Polanczyk CA,
#' Cook EF, Sugarbaker DJ, Donaldson MC, Poss R, Ho KK, Ludwig LE, Pedan A,
#' Goldman L: Derivation and prospective validation of a simple index for
#' prediction of cardiac risk of major noncardiac surgery. Circulation 1999;
#' 100:1043-9 \url{http://www.ncbi.nlm.nih.gov/pubmed/10477528}
#'
#' 2. Boersma E, Kertai MD, Schouten O, Bax JJ, Noordzij P, Steyerberg EW,
#' Schinkel AFL, Santen M van, Simoons ML, Thomson IR, Klein J, Urk H van,
#' Poldermans D: Perioperative cardiovascular mortality in noncardiac surgery:
#' validation of the Lee cardiac risk index. The American journal of medicine
#' 2005; 118:1134-41 \url{http://www.ncbi.nlm.nih.gov/pubmed/16194645}
#'
#' @seealso \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_elixhauser_quan}},
#' \code{\link{icd9cm_elixhauser_ahrq37}}
#' @export
icd9cm_rcri <- function(icd9) {
ahrq <- icd9cm_elixhauser_ahrq37(icd9)
quan <- icd9cm_charlson_quan(icd9)
data.frame(
row.names = icd9,
chf = ahrq$chf,
cvd = quan$cvd,
dm = ahrq$dm | ahrq$dmcx,
ischemia = grepl('^D41[01234]', icd9),
renlfail = ahrq$renlfail)
}
#' Convert ICD-9-CM code list to dataframe
#'
#' \code{melt_icd9list} uses \code{\link{ddply}} to melt a column of comma-separated ICD-9-CM
#' codes into a series of rows, one for each code.
#'
#' @param df a data frame with at least two columns, specified as \code{idvar}
#' and \code{icd9var}.
#' @param idvar string with name of ID variable within \code{df} (defaults to "id")
#' @param icd9var string with name of ICD code variable within \code{df} (defaults to "icd9cm")
#' @param .progress passed to \code{\link{ddply}}
#' @param .parallel passed to \code{\link{ddply}}
#' @param .paropts passed to \code{\link{ddply}}
#' @return a dataframe with two columns, \code{idvar} and \code{"icd9cm"}
#' @importFrom plyr ddply
#' @examples
#' cases <- data.frame(id=c(1,2),
#' icd9list=c('162.4,070.30,155.0,401.9','996.52,E878.8,V45.86'),
#' stringsAsFactors=TRUE)
#' melt_icd9list(cases, "id", "icd9list")
#' @export
melt_icd9list <- function(df, idvar="id", icd9var="icd9cm",
.progress="none", .parallel=FALSE, .paropts=NULL) {
ddply(df, idvar, function(x) {
data.frame(
icd9cm=unlist(strsplit(
paste(gsub('.','',x[,icd9var],fixed=TRUE),collapse=','),',',fixed=TRUE)))
}, .progress=.progress, .parallel=.parallel, .paropts=.paropts)
}
#' Merge ICD-9-CM diagnostic and procedure codes
#'
#' Merges a dataframe containing ICD-9-CM diagostic codes with a dataframe containing ICD-9 procedure codes
#' Diagnostic codes are prefixed with 'D', while procedure codes are prefixed with 'P'
#'
#' @param dx_df a data frame with at least two columns, specified as \code{idvar}
#' and \code{icd9dxvar}, where the values are ICD-9 diagnostic codes
#' @param proc_df a data frame with at least two columns, specified as \code{idvar}
#' and \code{icd9pvar}, where the values are ICD-9 procedure codes
#' @param icd9dxvar name of icd9 diagnostic code column, default "icd9cm"
#' @param icd9pvar name of icd9 procedure code column, default "icd9cm"
#' @return a merged dataframe with common columns and \code{"icd9cm"}
#' @examples
#' cases <- data.frame(id=c(1,2),
#' icd9dxlist=c('162.4,070.30,155.0,401.9','996.52,E878.8,V45.86'),
#' icd9plist=c('38.16','38.42'),
#' stringsAsFactors=TRUE)
#' dx_df <- melt_icd9list(cases, "id", "icd9dxlist")
#' proc_df <- melt_icd9list(cases, "id", "icd9plist")
#' merge_icd9_dx_and_procs(dx_df, proc_df)
#' @export
merge_icd9_dx_and_procs <- function(dx_df, proc_df, icd9dxvar="icd9cm", icd9pvar="icd9cm") {
if (!is.null(dx_df)) {
dx_df$icd9cm <- paste('D',dx_df[,icd9dxvar],sep='')
}
if (!is.null(proc_df)) {
proc_df$icd9cm <- paste('P',proc_df[,icd9pvar],sep='')
}
if (!is.null(proc_df) & !is.null(dx_df)) {
commonCols <- intersect(names(dx_df),names(proc_df))
df <- rbind(dx_df[,commonCols],proc_df[,commonCols])
} else if (is.null(proc_df)) {
df <- dx_df
} else if (is.null(dx_df)) {
df <- proc_df
}
df$icd9cm <- factor(df$icd9cm)
df
}
#' Generate a comorbidity dataframe
#'
#' Merges a given DF of IDs and ICD-9-CM codes to one of the ICD9CM maps,
#' removes redundant comorbidities, and returns a dataframe.
#'
#' Redundancy 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 returned.
#'
#' Van Walraven has a modification adopted here where the following "dmcx" codes
#' are downgraded to "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 PVD
#'
#' Cases without any comorbidities will not appear in the returned data
#' frame.
#'
#' @param df a data frame with at least two columns, specified as \code{idvar}
#' and \code{icd9var}.
#' @param idvar string with name of ID variable within \code{df} (defaults to
#' "id")
#' @param icd9var string with name of ICD code variable within \code{df}
#' (defaults to \code{icd9cm})
#' @param icd9mapfn Function to generate comorbidity data frame from ICD-9 codes
#' (defaults to \code{\link{icd9cm_charlson_quan}})
#' @param .progress passed to \code{\link{ddply}}
#' @param .parallel passed to \code{\link{ddply}}
#' @param .paropts passed to \code{\link{ddply}}
#' @return a dataframe with column \code{idvar} and a logical column for each comorbidity
#' @importFrom plyr ddply
#' @examples
#' cases <- data.frame(id=c(1,1,1,2,2,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403","D1960","D1958"),
#' stringsAsFactors=TRUE)
#' generate_comorbidity_df(cases)
#' # generate categories for patients in the \code{\link{vt_inp_sample}}
#' generate_comorbidity_df(vt_inp_sample)
#' # 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 <- data.frame(id=c(1,1,1,1,2,2,2,2),
#' icd9cm=c("D1970","D20206","D25071","D4439","D25001","D25040","D45621","D570"),
#' stringsAsFactors=TRUE)
#' generate_comorbidity_df(cases)
#' @export
generate_comorbidity_df <- function(df, idvar="id", icd9var="icd9cm",
icd9mapfn=icd9cm_charlson_quan,
.progress="none", .parallel=FALSE, .paropts=NULL) {
# create a unique vector for icd9 codes
uniq_icd9map <-
if (is.factor(df[,icd9var])) levels(df[,icd9var]) else unique(as.character(df[,icd9var]))
icd9map <- icd9mapfn(uniq_icd9map)
cases_merged <- merge(df[,c(idvar,icd9var)], icd9map, by.x=icd9var, by.y="row.names", all.x=T)
ddply(cases_merged, c(idvar),
function(df) {
rows <- c()
# preprocess df with van Walraven step
if ((all(c("renal","dm","dmcx") %in% names(df)) & any(df$renal)) |
(all(c("renlfail","dm","dmcx") %in% names(df)) & any(df$renlfail))) {
rows <- grep('^D2(49|50)4', df[,icd9var])
if (length(rows))
df[rows, c("dm", "dmcx")] <- matrix(rep(c(T,F), length(rows)), byrow = TRUE, ncol = 2)
}
if ((all(c("neuro","dm","dmcx") %in% names(df)) & any(df$neuro)) |
(all(c("cvd","dm","dmcx") %in% names(df)) & any(df$cvd))) {
rows <- grep('^D2(49|50)6', df[,icd9var])
if (length(rows))
df[rows, c("dm", "dmcx")] <- matrix(rep(c(T,F), length(rows)), byrow = TRUE, ncol = 2)
}
if (all(c("perivasc","dm","dmcx") %in% names(df)) & any(df$perivasc)) {
rows <- grep('^D2(49|50)7', df[,icd9var])
if (length(rows))
df[rows, c("dm", "dmcx")] <- matrix(rep(c(T,F), length(rows)), byrow = TRUE, ncol = 2)
}
# now merge rows together with any
out <- data.frame(lapply(df[,names(icd9map)], any))
# postprocess: eliminate redundancy
if (all(c("mets","tumor") %in% names(out))) {
rows <- which(out$mets & out$tumor)
if (length(rows))
out$tumor[rows] <- F
}
if (all(c("htn","htncx") %in% names(out))) {
rows <- which(out$htn & out$htncx)
if (length(rows))
out$htn[rows] <- F
}
if (all(c("dm","dmcx") %in% names(out))) {
rows <- which(out$dm & out$dmcx)
if (length(rows))
out$dm[rows] <- F
}
if (all(c("liver","modliver") %in% names(out))) {
rows <- which(out$liver & out$modliver)
if (length(rows))
out$liver[rows] <- F
}
out
},
.progress=.progress, .parallel=.parallel, .paropts=.paropts)
}
#' Calculate the Charlson Comorbidity Index
#'
#' \code{generate_charlson_index_df} merges a data frame of Charlson
#' comorbidities with \code{\link{charlson_weights}} and sums the results per
#' patient.
#'
#' @param df a data frame with ID column \code{idvar} and logical columns for each
#' comorbidity, such as that generated by \code{\link{generate_comorbidity_df}}
#' @param idvar string with name of ID variable within \code{df}
#' @param weights defaults to \code{\link{charlson_weights}}
#' @return a dataframe with two columns, \code{idvar} and \code{"index"}
#' @seealso \code{\link{generate_comorbidity_df}},
#' \code{\link{charlson_weights}}, \code{\link{charlson_weights_orig}}
#' @importFrom plyr empty
#' @examples
#' # calculate Charlson Comorbidity Index for all patients in the \code{\link{vt_inp_sample}}
#' data(vt_inp_sample)
#' generate_charlson_index_df(generate_comorbidity_df(vt_inp_sample))
#' @export
generate_charlson_index_df <- function(df, idvar="id", weights=medicalrisk::charlson_weights) {
if (empty(df)) {
return(df)
}
if (!exists(idvar,where=df)) {
stop("Data frame must have variable specified in idvar")
}
missing_cols <- setdiff(names(weights), names(df))
if (length(missing_cols)) {
stop(sprintf("Missing Charlson columns: %s", paste(missing_cols, collapse=",")))
}
# to generate index score, multiply weights by logical columns, then take row sum of that
rv <- data.frame(id=df[,idvar],
index=rowSums(df[,names(weights)] * weights))
names(rv)[1] <- idvar
rv
}
Try the medicalrisk package in your browser
Any scripts or data that you put into this service are public.
medicalrisk documentation built on March 26, 2020, 6:31 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions icd9cm_charlson_deyo icd9cm_charlson_romano icd9cm_charlson_quan icd9cm_elixhauser_quan icd9cm_elixhauser_ahrq37 icd9cm_rcri melt_icd9list merge_icd9_dx_and_procs generate_comorbidity_df generate_charlson_index_df
Documented in generate_charlson_index_df generate_comorbidity_df icd9cm_charlson_deyo icd9cm_charlson_quan icd9cm_charlson_romano icd9cm_elixhauser_ahrq37 icd9cm_elixhauser_quan icd9cm_rcri melt_icd9list merge_icd9_dx_and_procs
#' Create Deyo map of ICD-9-CM to Charlson comorbidities
#'
#' Function that generates a data frame linking ICD-9-CM codes to the Charlson
#' comorbidity categories using the Deyo mapping.
#'
#' NOTE: The input vector of ICD-9-CM codes must be unique, because the output dataframe
#' uses the ICD-9-CM code as row.name.
#'
#' Uses regular expressions created from the paper by Deyo in 1992.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{charlson_list}}
#' @references 1. 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
#' \url{http://www.ncbi.nlm.nih.gov/pubmed/1607900}
#
#' @seealso \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_charlson_romano}},
#' \code{\link{icd9cm_elixhauser_quan}}, \code{\link{icd9cm_elixhauser_ahrq37}},
#' \code{\link{charlson_weights}},
#' @examples
#' # Identify Charlson categories in ICD-9-CM listing
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_charlson_deyo(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_charlson_deyo <- function(icd9) {
data.frame(
row.names = icd9,
mi = grepl('^D(410|412)', icd9),
chf = grepl('^D428[0-9]', icd9),
perivasc = grepl('^(D(4439|441[0-9]|7854|V434)|P3848)', icd9),
cvd = grepl('^D43[0-8]', icd9),
dementia = grepl('^D290[0-9]', icd9),
chrnlung = grepl('^D(49[0-6]|50[0-5]|5064)', icd9),
rheum = grepl('^D(710[0,1,4]|714[0-2]|71481|725)', icd9),
ulcer = grepl('^D53[1-4]', icd9),
liver = grepl('^D571[2456]', icd9),
dm = grepl('^D250[0-3,7]', icd9),
dmcx = grepl('^D250[4-6]', icd9),
para = grepl('^D(3441|342)', icd9),
renal = grepl('^D(58[2568]|583[0-7])', icd9),
tumor = grepl('^D(1[4-6]|17[0-24-9]|18|19[0-5]|20[0-8])', icd9),
modliver = grepl('^D(456[0-2]|572[2-8])', icd9),
mets = grepl('^D19[6-9]', icd9),
aids = grepl('^D04[2-4]', icd9))
}
#' Create Romano map of ICD-9-CM to Charlson comorbidities
#'
#' Function that creates a dataframe which links ICD-9-CM codes to the Charlson
#' comorbidity categories using the Romano mapping.
#'
#' NOTE: The input vector of ICD-9-CM codes must be unique, because the output dataframe
#' uses the ICD-9-CM code as row.name.
#'
#' Uses regular expressions created from the paper by Romano in 1993.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{charlson_list}}
#' @references 1. Romano PS, Roos LL, Jollis JG: Adapting a clinical comorbidity index for
#' use with ICD-9-CM administrative data: differing perspectives. Journal of
#' clinical epidemiology 1993; 46:1075-9; discussion 1081-90
#' \url{http://www.ncbi.nlm.nih.gov/pubmed/8410092}
#
#' @seealso \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_charlson_deyo}},
#' \code{\link{icd9cm_elixhauser_quan}}, \code{\link{icd9cm_elixhauser_ahrq37}},
#' \code{\link{charlson_weights}},
#' @examples
#' # Identify Charlson categories in ICD-9-CM listing
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_charlson_romano(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_charlson_romano <- function(icd9) {
data.frame(
row.names = icd9,
mi = grepl('^D(410|412)', icd9),
chf = grepl('^D(402[01]1|40291|425|428|4293)', icd9),
perivasc = grepl('^(D(44[012]|443[1-9]|4471|7854)|P(38(1[3468]|3[3468]|4[3468])|392[2-69]))', icd9),
cvd = grepl('^(D(36234|43[0-6]|437[019]|438|7814|7843|9970)|P38[14]2)', icd9),
dementia = grepl('^D(290[0-9]|331[0-2])', icd9),
chrnlung = grepl('^D(4150|416[89]|49[1-46])', icd9),
rheum = F,
ulcer = grepl('^D53[1-4]', icd9),
liver = grepl('^D571[25689]', icd9),
dm = grepl('^D250[0-3]', icd9),
dmcx = grepl('^D250[4-9]', icd9),
para = grepl('^D(34[2,4])', icd9),
renal = grepl('^(D(58[56]|V420|V451|V56)|P(39(27|42|9[3-5])|5498))', icd9),
tumor = grepl('^(D(1[4-6]|17[0-14-9]|18|19[0-5]|20[0-8]|273[03]|V1046)|P(605|6241))', icd9),
modliver = grepl('^(D(456[0-2]|572[2-4])|P(391|4291))', icd9),
mets = grepl('^D19[6-9]', icd9),
aids = grepl('^D04[2-4]', icd9))
}
#' Create Quan map of ICD-9-CM to Charlson comorbidities
#'
#' Function that creates a dataframe that links ICD-9-CM codes to the Charlson comorbidity
#' categories using Quan's method.
#'
#' NOTE: The input vector of ICD-9-CM codes must be unique, because the output dataframe
#' uses the ICD-9-CM code as row.name.
#'
#' Uses regular expressions created from the paper by Quan in 2005.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{charlson_list}}
#' @references 1. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi
#' J-C, Saunders LD, Beck CA, Feasby TE, Ghali WA: Coding algorithms for
#' defining comorbidities in ICD-9-CM and ICD-10 administrative data. Medical
#' care 2005; 43:1130-9
#' \url{http://www.ncbi.nlm.nih.gov/pubmed/16224307}
#
#' @seealso \code{\link{icd9cm_charlson_deyo}}, \code{\link{icd9cm_charlson_romano}},
#' \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_elixhauser_quan}}
#' @examples
#' # Identify Charlson categories in ICD-9-CM listing
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_charlson_quan(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_charlson_quan <- function(icd9) {
data.frame(
row.names = icd9,
mi = grepl('^D(410|412)', icd9),
chf = grepl('^D(39891|402[019]1|404[019][13]|425[4-9]|428)', icd9),
perivasc = grepl('^D(0930|4373|44[01]|443[1-9]|4471|557[19]|V434)', icd9),
cvd = grepl('^D(36234|43[0-8])', icd9),
dementia = grepl('^D(290|2941|3312)', icd9),
chrnlung = grepl('^D(416[89]|49|50[0-5]|5064|508[18])', icd9),
rheum = grepl('^D(4465|710[0-4]|714[0-2]|7148|725)', icd9),
ulcer = grepl('^D53[1-4]', icd9),
liver = grepl('^D(070[23][23]|070[45]4|070[69]|57[01]|573[3489]|V427)', icd9),
dm = grepl('^D250[0-389]', icd9),
dmcx = grepl('^D250[4-7]', icd9),
para = grepl('^D(3341|34[23]|344[0-69])', icd9),
renal = grepl('^D(403[019]1|404[019][23]|582|583[0-7]|58[56]|5880|V420|V451|V56)', icd9),
tumor = grepl('^D(1[4-6]|17[0-24-9]|18|19[0-5]|20[0-8]|2386)', icd9),
modliver = grepl('^D(456[0-2]|572[2-8])', icd9),
mets = grepl('^D19[6-9]', icd9),
aids = grepl('^D04[2-4]', icd9))
}
#' Create Quan map of ICD-9-CM to Elixhauser comorbidities
#'
#' Function to make a dataframe that links ICD-9-CM codes to the Elixhauser comorbidity
#' categories using the Quan mapping.
#'
#' Uses regular expressions created from the Quan paper from 2005.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' Some ICD-9-CM codes will correspond to more than one category. For example,
#' \code{404.03} (Hypertensive heart and chronic kidney disease ... stage V) is
#' in both \code{chf} and \code{renlfail} categories.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{elixhauser_list}}
#' @references 1. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi
#' J-C, Saunders LD, Beck CA, Feasby TE, Ghali WA: Coding algorithms for
#' defining comorbidities in ICD-9-CM and ICD-10 administrative data. Medical
#' care 2005; 43:1130-9
#' \url{http://www.ncbi.nlm.nih.gov/pubmed/16224307}
#
#' @seealso \code{\link{icd9cm_charlson_deyo}}, \code{\link{icd9cm_charlson_romano}},
#' \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_elixhauser_ahrq37}}
#' @examples
#' # Identify Elixhauser categories
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_elixhauser_quan(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_elixhauser_quan <- function(icd9) {
data.frame(
row.names = icd9,
chf = grepl('^D(39891|402[019]1|404[019][13]|425[4-9]|428)', icd9),
arrhythmia = grepl('^D(426([079]|13)|426(10|12)|427[0-46-9]|7850|9960[14]|V450|V533)', icd9),
valve = grepl('^D(0932|39[4567]|424|746[3-6]|V422|V433)', icd9),
pulmcirc = grepl('^D(415[01]|416|417[089])', icd9),
perivasc = grepl('^D(0930|4373|44[01]|443[1-9]|4471|557[19]|V434)', icd9),
htn = grepl('^D401', icd9),
htncx = grepl('^D40[2-5]', icd9),
para = grepl('^D(3341|34[23]|344[0-69])', icd9),
neuro = grepl('^D(3319|332[01]|333[45]|33392|33[45]|3362|34[015]|348[13]|7803|7843)', icd9),
chrnlung = grepl('^D(416[89]|49|50[0-5]|5064|508[18])', icd9),
dm = grepl('^D250[0-3]', icd9),
dmcx = grepl('^D250[4-9]', icd9),
hypothy = grepl('^D(2409|24[34]|246[18])', icd9),
renlfail = grepl('^D(403[019]1|404[019][23]|58[56]|5880|V420|V451|V56)', icd9),
liver = grepl('^D(070[23][23]|070[45]4|070[69]|456[0-2]|57[01]|572[2-8]|573[3489]|V427)', icd9),
ulcer = grepl('^D53[1-4][79]', icd9),
aids = grepl('^D04[2-4]', icd9),
lymph = grepl('^D(20[0-2]|2030|2386)', icd9),
mets = grepl('^D19[6-9]', icd9),
tumor = grepl('^D(1[4-6]|17[0-24-9]|18|19[0-5])', icd9),
rheum = grepl('^D(446|7010|710[0-489]|7112|714|7193|72[05]|7285|72889|72930)', icd9),
coag = grepl('^D(286|287[13-5])', icd9),
obese = grepl('^D2780', icd9),
wghtloss = grepl('^D(26[0-3]|7832|7994)', icd9),
lytes = grepl('^D(2536|276)', icd9),
bldloss = grepl('^D2800', icd9),
anemdef = grepl('^D(280[1-9]|281)', icd9),
alcohol = grepl('^D(2652|291[1-35-9]|303[09]|3050|3575|4255|5353|571[0-3]|980|V113)', icd9),
drug = grepl('^D(292|304|305[2-9]|V6542)', icd9),
psych = grepl('^D(2938|295|296[0145]4|29[78])', icd9),
depress = grepl('^D(296[235]|3004|309|311)', icd9))
}
#' Create AHRQ v3.7 map of ICD-9-CM to Elixhauser comorbidities
#'
#' Function makes a dataframe that links ICD-9-CM codes to the Elixhauser comorbidity
#' categories using the AHRQ v3.7 mapping.
#'
#' Uses regular expressions based on the file "comformat2012-2013.txt" from AHRQ.
#'
#' The Agency for Healthcare Research and Quality (AHRQ) has developed
#' Comorbidity Software as part of the Healthcare Cost and Utilization Project
#' (HCUP). The software was developed to report on the comorbidity measures
#' reported by Elixhauser (1998).
#'
#' The AHRQ software has two parts, one that classifies ICD-9-CM codes by
#' comorbidity, and another that performs heuristics to eliminate duplicate
#' comorbidities and ignore comorbidities which are the primary reason for the
#' hospital visit, as per the DRG.
#'
#' This table is a translation of the first part of the software, the
#' classifier, as implemented in the SAS file \code{Comformat2012-2013.txt}.
#'
#' ICD-9-CM codes must have periods removed. Diagnostic codes are prefixed with
#' 'D' while procedure codes are prefixed with 'P'. So, diagnostic code
#' \code{404.03} should be \code{"D40403"}.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and one logical column for each
#' comorbidity in \code{\link{elixhauser_list}}
#' @references 1. \url{http://www.hcup-us.ahrq.gov/toolssoftware/comorbidity/comorbidity.jsp}
#'
#' @seealso \code{\link{icd9cm_charlson_deyo}}, \code{\link{icd9cm_charlson_romano}},
#' \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_elixhauser_quan}}
#' @examples
#' # Identify Elixhauser categories
#' cases <- data.frame(id=c(1,1,1,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403"),
#' stringsAsFactors=TRUE)
#' cases_with_cm <- merge(cases, icd9cm_elixhauser_ahrq37(levels(cases$icd9cm)),
#' by.x="icd9cm", by.y="row.names", all.x=TRUE)
#'
#' # generate crude comorbidity summary for each patient
#' library(plyr)
#' ddply(cases_with_cm, .(id),
#' function(x) { data.frame(lapply(x[,3:ncol(x)], any)) })
#' @export
icd9cm_elixhauser_ahrq37 <- function(icd9) {
# the below is taken from comformat2012-2013.txt
# htncx: htncx, htnpreg, htnwochf, hrenworf, hhrwohrf
# htncx & chf: htnwchf, hhrwchf
# htncx & renlfail: hrenwrf, hhrwrf
# htncx & chf & renlfail: hhrwhrf
htnpreg <- grepl('^D6422', icd9)
htnwochf <- grepl('^D(402[019]0|405[019]9)', icd9)
htnwchf <- grepl('^D402[019]1', icd9)
hrenworf <- grepl('^D(403[019]0|405[019]1|6421)', icd9)
hrenwrf <- grepl('^D403[019]1', icd9)
hhrwohrf <- grepl('^D404[019]0', icd9)
hhrwchf <- grepl('^D404[019]1', icd9)
hhrwrf <- grepl('^D404[019]2', icd9)
hhrwhrf <- grepl('^D404[019]3', icd9)
ohtnpreg <- grepl('^D642[79]', icd9)
data.frame(
row.names = icd9,
chf = grepl('^D(39891|428)', icd9) | htnwchf | hhrwchf | hhrwhrf,
arrhythmia = F,
valve = grepl('^D(0932|39[456]|397[019]|424|746[3-6]|V422|V433)', icd9),
pulmcirc = grepl('^D(4151|416|4179)', icd9),
perivasc = grepl('^D(44[012]|443[1-9]|4442|4471|449|557[19]|V434)', icd9),
htn = grepl('^D(401[19]|6420)', icd9),
htncx = grepl('^D(4010|4372)', icd9) |
htnpreg | htnwochf | hrenworf | hhrwohrf |
htnwchf | hhrwchf | hrenwrf | hhrwrf | hhrwhrf,
para = grepl('^D(34[234]|438[2-5]|78072)', icd9),
neuro = grepl('^D(33[01]|3320|333[457]|33385|33394|33[45]|3380|34[0157]|6494|7687|7803|78097|7843)', icd9),
chrnlung = grepl('^D(49|50[0-5]|5064)', icd9),
dm = grepl('^D(250[0-3]|6480|249[0-3])', icd9),
dmcx = grepl('^D(250[4-9]|7751|249[4-9])', icd9),
hypothy = grepl('^D(243|244[01289])', icd9),
renlfail = grepl('^D(585[3-9]|586|V420|V451|V56)', icd9) | hrenwrf | hhrwrf,
liver = grepl('^D(070[23][23]|070[45]4|456[0-2]|571[02-9]|572[38]|5735|V427)', icd9),
ulcer = grepl('^D(53[1234][456]1|53[1234]7[01]|53[1234]91)', icd9),
aids = grepl('^D04[2-4]', icd9),
lymph = grepl('^D(20[01]|202[0-35-9]|203|2386|2733)', icd9),
mets = grepl('^D(19[6-8]|199[01]|2097|78951)', icd9),
tumor = grepl('^D(1[4-6]|17[012459]|18|19[0-5]|209[0-3]|2580)', icd9),
rheum = grepl('^D(7010|71[04]|720|725)', icd9),
coag = grepl('^D(286|287[13-5]|6493|28984)', icd9),
obese = grepl('^D(2780|6491|V85[34]|V8554|79391)', icd9),
wghtloss = grepl('^D(26[0-3]|7832)', icd9),
lytes = grepl('^D276', icd9),
bldloss = grepl('^D(2800|6482)', icd9),
anemdef = grepl('^D(280[1-9]|281|285[29])', icd9),
alcohol = grepl('^D(291[0-35-9]|303[09]|3050)', icd9),
drug = grepl('^D(292|304|305[2-9]|6483)', icd9),
psych = grepl('^D(29[5678]|2991)', icd9),
depress = grepl('^D(3004|30112|309[01]|311)', icd9))
}
#' Create Map of ICD-9-CM to Revised Cardiac Risk Index classes
#'
#' Function to generate data frame that links ICD-9-CM codes to the RCRI comorbidity categories.
#'
#' Lee et al in 1999 published a "Revised Cardiac Risk Index" based on the work
#' on Goldman in 1997. The RCRI is used to determine the major cardiac
#' complication risk for a patient about to undergo major noncardiac surgery.
#' The six predictors that make up the RCRI are: 1. high-risk surgery 2. history
#' of ischemic heart disease 3. history of congestive heart failure 4. history
#' of cerebrovascular disease 5. preoperative treatment with insulin 6.
#' preoperative serum creatinine with Cr > 2 mg/dL.
#'
#' In 2005 Boersma et al demonstrated that the Lee indexed can be adapted to use
#' administrative data to predict cardiovascular mortality. They used the
#' following for each point above: 1. retroperitoneal, intrathoracic, or
#' suprainguinal vascular procedure; 2. Ischemia: ICD-9 codes 410.*, 411.*,
#' 412.*, 413.*, 414.*; 3. CVA: ICD-9 428.*; 4. CHF: ICD-9 943.0; 5. DM: ICD-9
#' 425.0; 6. Renal: ICD-9 958.0.
#'
#' This function merges the ICD-9 guidelines
#' used by Boersma with some of the other ICD-9 classifiers in this package.
#' This data set uses the following for each aspect of the RCRI:
#' 1. procedure is left to you
#' 2. 'ischemia' as defined in Boersma
#' 3. 'cvd' as defined by Quan in \code{\link{icd9cm_charlson_quan}}
#' 4. 'chf' as defined by AHRQ in \code{\link{icd9cm_elixhauser_ahrq37}}
#' 5. 'dm' as defined by AHRQ (both 'dm' and 'dmcx')
#' 6. renlfail' as defined by AHRQ.
#'
#' @param icd9 a unique character vector of ICD-9-CM codes
#' @return A data frame, with ICD9 codes as row names and logical columns for
#' \code{chf}, \code{cvd}, \code{dm}, \code{ischemia}, and \code{renlfail}.
#' @references 1. Lee TH, Marcantonio ER, Mangione CM, Thomas EJ, Polanczyk CA,
#' Cook EF, Sugarbaker DJ, Donaldson MC, Poss R, Ho KK, Ludwig LE, Pedan A,
#' Goldman L: Derivation and prospective validation of a simple index for
#' prediction of cardiac risk of major noncardiac surgery. Circulation 1999;
#' 100:1043-9 \url{http://www.ncbi.nlm.nih.gov/pubmed/10477528}
#'
#' 2. Boersma E, Kertai MD, Schouten O, Bax JJ, Noordzij P, Steyerberg EW,
#' Schinkel AFL, Santen M van, Simoons ML, Thomson IR, Klein J, Urk H van,
#' Poldermans D: Perioperative cardiovascular mortality in noncardiac surgery:
#' validation of the Lee cardiac risk index. The American journal of medicine
#' 2005; 118:1134-41 \url{http://www.ncbi.nlm.nih.gov/pubmed/16194645}
#'
#' @seealso \code{\link{icd9cm_charlson_quan}}, \code{\link{icd9cm_elixhauser_quan}},
#' \code{\link{icd9cm_elixhauser_ahrq37}}
#' @export
icd9cm_rcri <- function(icd9) {
ahrq <- icd9cm_elixhauser_ahrq37(icd9)
quan <- icd9cm_charlson_quan(icd9)
data.frame(
row.names = icd9,
chf = ahrq$chf,
cvd = quan$cvd,
dm = ahrq$dm | ahrq$dmcx,
ischemia = grepl('^D41[01234]', icd9),
renlfail = ahrq$renlfail)
}
#' Convert ICD-9-CM code list to dataframe
#'
#' \code{melt_icd9list} uses \code{\link{ddply}} to melt a column of comma-separated ICD-9-CM
#' codes into a series of rows, one for each code.
#'
#' @param df a data frame with at least two columns, specified as \code{idvar}
#' and \code{icd9var}.
#' @param idvar string with name of ID variable within \code{df} (defaults to "id")
#' @param icd9var string with name of ICD code variable within \code{df} (defaults to "icd9cm")
#' @param .progress passed to \code{\link{ddply}}
#' @param .parallel passed to \code{\link{ddply}}
#' @param .paropts passed to \code{\link{ddply}}
#' @return a dataframe with two columns, \code{idvar} and \code{"icd9cm"}
#' @importFrom plyr ddply
#' @examples
#' cases <- data.frame(id=c(1,2),
#' icd9list=c('162.4,070.30,155.0,401.9','996.52,E878.8,V45.86'),
#' stringsAsFactors=TRUE)
#' melt_icd9list(cases, "id", "icd9list")
#' @export
melt_icd9list <- function(df, idvar="id", icd9var="icd9cm",
.progress="none", .parallel=FALSE, .paropts=NULL) {
ddply(df, idvar, function(x) {
data.frame(
icd9cm=unlist(strsplit(
paste(gsub('.','',x[,icd9var],fixed=TRUE),collapse=','),',',fixed=TRUE)))
}, .progress=.progress, .parallel=.parallel, .paropts=.paropts)
}
#' Merge ICD-9-CM diagnostic and procedure codes
#'
#' Merges a dataframe containing ICD-9-CM diagostic codes with a dataframe containing ICD-9 procedure codes
#' Diagnostic codes are prefixed with 'D', while procedure codes are prefixed with 'P'
#'
#' @param dx_df a data frame with at least two columns, specified as \code{idvar}
#' and \code{icd9dxvar}, where the values are ICD-9 diagnostic codes
#' @param proc_df a data frame with at least two columns, specified as \code{idvar}
#' and \code{icd9pvar}, where the values are ICD-9 procedure codes
#' @param icd9dxvar name of icd9 diagnostic code column, default "icd9cm"
#' @param icd9pvar name of icd9 procedure code column, default "icd9cm"
#' @return a merged dataframe with common columns and \code{"icd9cm"}
#' @examples
#' cases <- data.frame(id=c(1,2),
#' icd9dxlist=c('162.4,070.30,155.0,401.9','996.52,E878.8,V45.86'),
#' icd9plist=c('38.16','38.42'),
#' stringsAsFactors=TRUE)
#' dx_df <- melt_icd9list(cases, "id", "icd9dxlist")
#' proc_df <- melt_icd9list(cases, "id", "icd9plist")
#' merge_icd9_dx_and_procs(dx_df, proc_df)
#' @export
merge_icd9_dx_and_procs <- function(dx_df, proc_df, icd9dxvar="icd9cm", icd9pvar="icd9cm") {
if (!is.null(dx_df)) {
dx_df$icd9cm <- paste('D',dx_df[,icd9dxvar],sep='')
}
if (!is.null(proc_df)) {
proc_df$icd9cm <- paste('P',proc_df[,icd9pvar],sep='')
}
if (!is.null(proc_df) & !is.null(dx_df)) {
commonCols <- intersect(names(dx_df),names(proc_df))
df <- rbind(dx_df[,commonCols],proc_df[,commonCols])
} else if (is.null(proc_df)) {
df <- dx_df
} else if (is.null(dx_df)) {
df <- proc_df
}
df$icd9cm <- factor(df$icd9cm)
df
}
#' Generate a comorbidity dataframe
#'
#' Merges a given DF of IDs and ICD-9-CM codes to one of the ICD9CM maps,
#' removes redundant comorbidities, and returns a dataframe.
#'
#' Redundancy 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 returned.
#'
#' Van Walraven has a modification adopted here where the following "dmcx" codes
#' are downgraded to "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 PVD
#'
#' Cases without any comorbidities will not appear in the returned data
#' frame.
#'
#' @param df a data frame with at least two columns, specified as \code{idvar}
#' and \code{icd9var}.
#' @param idvar string with name of ID variable within \code{df} (defaults to
#' "id")
#' @param icd9var string with name of ICD code variable within \code{df}
#' (defaults to \code{icd9cm})
#' @param icd9mapfn Function to generate comorbidity data frame from ICD-9 codes
#' (defaults to \code{\link{icd9cm_charlson_quan}})
#' @param .progress passed to \code{\link{ddply}}
#' @param .parallel passed to \code{\link{ddply}}
#' @param .paropts passed to \code{\link{ddply}}
#' @return a dataframe with column \code{idvar} and a logical column for each comorbidity
#' @importFrom plyr ddply
#' @examples
#' cases <- data.frame(id=c(1,1,1,2,2,2,2,2),
#' icd9cm=c("D20206","D24220","D4439","D5064","DE8788","D40403","D1960","D1958"),
#' stringsAsFactors=TRUE)
#' generate_comorbidity_df(cases)
#' # generate categories for patients in the \code{\link{vt_inp_sample}}
#' generate_comorbidity_df(vt_inp_sample)
#' # 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 <- data.frame(id=c(1,1,1,1,2,2,2,2),
#' icd9cm=c("D1970","D20206","D25071","D4439","D25001","D25040","D45621","D570"),
#' stringsAsFactors=TRUE)
#' generate_comorbidity_df(cases)
#' @export
generate_comorbidity_df <- function(df, idvar="id", icd9var="icd9cm",
icd9mapfn=icd9cm_charlson_quan,
.progress="none", .parallel=FALSE, .paropts=NULL) {
# create a unique vector for icd9 codes
uniq_icd9map <-
if (is.factor(df[,icd9var])) levels(df[,icd9var]) else unique(as.character(df[,icd9var]))
icd9map <- icd9mapfn(uniq_icd9map)
cases_merged <- merge(df[,c(idvar,icd9var)], icd9map, by.x=icd9var, by.y="row.names", all.x=T)
ddply(cases_merged, c(idvar),
function(df) {
rows <- c()
# preprocess df with van Walraven step
if ((all(c("renal","dm","dmcx") %in% names(df)) & any(df$renal)) |
(all(c("renlfail","dm","dmcx") %in% names(df)) & any(df$renlfail))) {
rows <- grep('^D2(49|50)4', df[,icd9var])
if (length(rows))
df[rows, c("dm", "dmcx")] <- matrix(rep(c(T,F), length(rows)), byrow = TRUE, ncol = 2)
}
if ((all(c("neuro","dm","dmcx") %in% names(df)) & any(df$neuro)) |
(all(c("cvd","dm","dmcx") %in% names(df)) & any(df$cvd))) {
rows <- grep('^D2(49|50)6', df[,icd9var])
if (length(rows))
df[rows, c("dm", "dmcx")] <- matrix(rep(c(T,F), length(rows)), byrow = TRUE, ncol = 2)
}
if (all(c("perivasc","dm","dmcx") %in% names(df)) & any(df$perivasc)) {
rows <- grep('^D2(49|50)7', df[,icd9var])
if (length(rows))
df[rows, c("dm", "dmcx")] <- matrix(rep(c(T,F), length(rows)), byrow = TRUE, ncol = 2)
}
# now merge rows together with any
out <- data.frame(lapply(df[,names(icd9map)], any))
# postprocess: eliminate redundancy
if (all(c("mets","tumor") %in% names(out))) {
rows <- which(out$mets & out$tumor)
if (length(rows))
out$tumor[rows] <- F
}
if (all(c("htn","htncx") %in% names(out))) {
rows <- which(out$htn & out$htncx)
if (length(rows))
out$htn[rows] <- F
}
if (all(c("dm","dmcx") %in% names(out))) {
rows <- which(out$dm & out$dmcx)
if (length(rows))
out$dm[rows] <- F
}
if (all(c("liver","modliver") %in% names(out))) {
rows <- which(out$liver & out$modliver)
if (length(rows))
out$liver[rows] <- F
}
out
},
.progress=.progress, .parallel=.parallel, .paropts=.paropts)
}
#' Calculate the Charlson Comorbidity Index
#'
#' \code{generate_charlson_index_df} merges a data frame of Charlson
#' comorbidities with \code{\link{charlson_weights}} and sums the results per
#' patient.
#'
#' @param df a data frame with ID column \code{idvar} and logical columns for each
#' comorbidity, such as that generated by \code{\link{generate_comorbidity_df}}
#' @param idvar string with name of ID variable within \code{df}
#' @param weights defaults to \code{\link{charlson_weights}}
#' @return a dataframe with two columns, \code{idvar} and \code{"index"}
#' @seealso \code{\link{generate_comorbidity_df}},
#' \code{\link{charlson_weights}}, \code{\link{charlson_weights_orig}}
#' @importFrom plyr empty
#' @examples
#' # calculate Charlson Comorbidity Index for all patients in the \code{\link{vt_inp_sample}}
#' data(vt_inp_sample)
#' generate_charlson_index_df(generate_comorbidity_df(vt_inp_sample))
#' @export
generate_charlson_index_df <- function(df, idvar="id", weights=medicalrisk::charlson_weights) {
if (empty(df)) {
return(df)
}
if (!exists(idvar,where=df)) {
stop("Data frame must have variable specified in idvar")
}
missing_cols <- setdiff(names(weights), names(df))
if (length(missing_cols)) {
stop(sprintf("Missing Charlson columns: %s", paste(missing_cols, collapse=",")))
}
# to generate index score, multiply weights by logical columns, then take row sum of that
rv <- data.frame(id=df[,idvar],
index=rowSums(df[,names(weights)] * weights))
names(rv)[1] <- idvar
rv
}
Try the medicalrisk package in your browser
Any scripts or data that you put into this service are public.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.