Add ICD-9 / ICD-10 crosswalk

NAMESPACE

@@ -98,6 +98,11 @@ S3method(icd_conflicts,icd9)
 S3method(icd_conflicts,icd9cm)
 S3method(icd_conflicts,icd9cm_pc)
 S3method(icd_conflicts,icd9who)
+S3method(icd_gem,default)
+S3method(icd_gem,icd10)
+S3method(icd_gem,icd10cm_pc)
+S3method(icd_gem,icd9)
+S3method(icd_gem,icd9cm_pc)
 S3method(is_billable,default)
 S3method(is_billable,icd10)
 S3method(is_billable,icd10cm)
@@ -131,6 +136,7 @@ S3method(short_to_decimal,default)
 S3method(short_to_decimal,icd10)
 S3method(short_to_decimal,icd10cm)
 S3method(short_to_decimal,icd9)
+S3method(short_to_decimal,icd9cm_pc)
 S3method(short_to_parts,character)
 S3method(short_to_parts,icd10)
 S3method(short_to_parts,icd10cm)
@@ -271,6 +277,7 @@ export(icd9_comorbid_quan_deyo)
 export(icd9_comorbid_quan_elix)
 export(icd9_filter_invalid)
 export(icd9_filter_valid)
+export(icd_gem)
 export(icd_long_data)
 export(icd_wide_data)
 export(is.comorbidity_map)

R/cim-10-fr.R

@@ -8,7 +8,7 @@
       "https://www.atih.sante.fr",
       "plateformes-de-transmission-et-logiciels",
       "logiciels-espace-de-telechargement",
-      "telecharger/gratuit/11616/456"
+      "telecharger/gratuit/12713/456"
     ),
     file_name = "LIBCIM10MULTI.TXT",
     dl_msg = "Working on ICD-10-FR (CIM-10-FR)",

R/convert.R

@@ -293,6 +293,14 @@ short_to_decimal.icd9 <- function(x) {
   icd9(as.decimal_diag(icd9_short_to_decimal_rcpp(x)))
 }
 
+#' @describeIn short_to_decimal convert ICD-9 codes from short to decimal
+#'   format
+#' @export
+#' @keywords internal manip
+short_to_decimal.icd9cm_pc <- function(x) {
+  icd10cm_pc(as.decimal_diag(paste0(substr(x, 1L, 2L), ".", substr(x, 3L, 4L))))
+}
+
 #' @describeIn short_to_decimal convert ICD-10 codes from short to decimal
 #'   format
 #' @export

R/datadocs.R

@@ -542,3 +542,114 @@ NULL
 #' @docType data
 #' @keywords datasets
 NULL
+
+#' ICD-9 / ICD-10 GEMs
+#'
+#' A representation of the ICD-9 / ICD-10 General Equivalence Mappings (GEMs).
+#' The \code{icd_gem_9_to_10} dataset provides the mapping from ICD-9 codes to
+#' ICD-10 codes, while the \code{icd_gem_10_to_9} dataset provides the mapping
+#' from ICD-10 codes to ICD-9 codes.
+#'
+#' The GEMs denote \emph{source} codes as the ICD codes that you are mapping
+#' from, and \emph{target} codes as the codes that you are mapping to.  The GEMs
+#' also have a concept of what they call an \emph{entry}, which is a mapping
+#' between a single source code and one or more target codes.  In the raw form
+#' of the GEMs, an entry is a collection of one or more rows in the table that
+#' all have the same value for \code{source}.  It is worth noting that for some
+#' codes, there is no reasonable representation of the given code in the target
+#' code set.
+#'
+#' The relationship between the ICD-9 codes and ICD-10 codes is often not
+#' straightforward (in general, ICD-10 codes tend to be more specific than ICD-9
+#' codes).  In the simplest setting, a given source code can be exactly or
+#' approximately represented by one or more target codes.  In the more
+#' complicated (but fortunately less common) setting, the source code is what is
+#' denoted in the GEMs as a \emph{combination} code, which means that you need
+#' to observe two or more target codes in conjunction to represent the source
+#' event.  For example, some codes represent bilateral events (i.e. an event
+#' that occurs to both sides of the body), so that if the source code represents
+#' an event that occurred to both feet, then you could represent this by two
+#' target codes, where one code represents that event occurring to the left
+#' foot, and the other code represents that event occurring to the right foot.
+#' So in in terms of codes, suppose we have a collection of codes that we call
+#' set A and a collection of codes that we call set B, then a combination code
+#' might require both a code from a set A \strong{and} a code from set B.  In
+#' some cases, the combinations might require three or more codes in
+#' conjunction, so that you would require a code from a set A \strong{and} a
+#' code from set B \strong{and} a code from set C, and so on.  These different
+#' sets are referred to in the GEMs' terminology as \emph{choices}.
+#'
+#' The GEMs also have a concept of what they call a \emph{scenario}.  These are
+#' typically used when the source event has multiple ways of being satisfied.
+#' For example, a particular symptom occurring as the result of a chemical or
+#' biological exposure could be represented with codes for the symptom and the
+#' chemical exposure, \emph{or} codes for the symptom and the biological
+#' exposure (so this example has two scenarios, each of which is a combination).
+#' In terms of logic, scenarios represent OR statements, while combinations
+#' represent AND statements.
+#'
+#' The representation of the GEMs provided by \code{icd_gem_9_to_10} and
+#' \code{icd_gem_10_to_9} is the following.  Each row in the data corresponds to
+#' a single scenario for a given source code.  The columns in the data are
+#' listed as follows.
+#'
+#' \itemize{
+#'
+#'   \item \code{source_short}: The source ICD codes in short form.
+#'
+#'   \item \code{source_decimal}: The source ICD codes in decimal form.
+#'
+#'   \item \code{scenario}: The scenario.
+#'
+#'   \item \code{type}: The scenario type.
+#'
+#'   \item \code{approx}: Whether the mapping is an exact or an approximate
+#'     mapping.
+#'
+#'   \item \code{codes_short}: A list with 0 or more code sets in short form
+#'     with the semantics that there must be at least one code from the target
+#'     set in every one of the code sets for the scenario to be satisfied.
+#'
+#'   \item \code{codes_decimal}: A list with the same form as
+#'   \code{codes_short}, but in decimal rather than short form.
+#'
+#' }
+#'
+#' @section Example ICD-9 to ICD-10 mapping:
+#'
+#' For a concrete example, suppose that we want to map ICD-9 codes to ICD-10
+#' codes.  Further suppose that for a given row that the value of
+#' \code{source_short} (i.e. the short form ICD-9 code) is \code{"24951"}, the
+#' value of \code{scenario} is \code{"0"}, and the contents of the value of
+#' \code{codes_short} is a single vector \code{c("E0839", "E0939")}.  Then one
+#' of the ICD-10 codes in our dataset must have a value of \code{E0839} or
+#' \code{E0939} for our data to have an event corresponding to the ICD-9 code
+#' 24951 by way of scenario 0.
+#'
+#' Continuing this example, as it happens there are two scenarios for source
+#' ICD-9 code 24951, and thus there is another row of the data, and for which
+#' the value of \code{source_short} is \code{"24951"}, the value of
+#' \code{scenario} is \code{"1"}, and the contents of the value of
+#' \code{codes_short} is the vectors \code{c("E08311", "E08319", "E0836",
+#' "E09311", "E09319", "E0936")} and \code{"E0865"}.  Then one of the ICD-10
+#' codes in our dataset must have a value of one of \code{c("E08311", "E08319",
+#' "E0836", "E09311", "E09319", "E0936")}, \emph{and} another of the ICD-10
+#' codes in our dataset must have value of \code{E0939} for our data to have an
+#' event corresponding to the ICD-9 code 24951 by way of scenario 1.
+#'
+#' Since these are the only two scenarios corresponding to ICD-9 code 24951, we
+#' must have ICD-10 codes in our data that satisfy either scenario 0 or scenario
+#' 1 for us to conclude that an event corresponding to ICD-9 code 24951
+#' occurred.
+#'
+#' Some input source codes don't have a representation in the target code set or
+#' may not exist in the corresponding GEM at all.  When either of these events
+#' is the case then the corresponding value of \code{codes} is an empty list.
+#'
+#' @docType data
+#' @source \url{https://www.cms.gov/Medicare/Coding/ICD10/2018-ICD-10-CM-and-GEMs.html}
+#' @name icd_gem_9_to_10
+#' @aliases icd_gem_10_to_9
+#' @seealso \code{\link{icd_gem}}
+#' @keywords datasets
+NULL