Calculate Medicare Hierarchical Condition Categories HCC

DESCRIPTION

@@ -8,7 +8,7 @@ Description: Calculate comorbidities, Charlson scores, perform fast and accurate
     from Quan (Deyo and Elixhauser versions), Elixhauser and AHRQ included. This
     package replaces 'icd9', which can now be uninstalled.
 Classification/ACM-2012: Social and professional topics~Medical records, Applied computing~Health care information systems, Applied computing~Health informatics, Applied computing~Bioinformatics
-Version: 2.0.1
+Version: 2.0.1.9000
 Date: 2016-04-10
 Authors@R: c(
     person(given = "Jack O.", family = "Wasey", role = c("aut", "cre", "cph"), email = "jack@jackwasey.com"),

NAMESPACE

@@ -101,6 +101,7 @@ export(as.icd_wide_data)
 export(icd10_comorbid)
 export(icd10_comorbid_ahrq)
 export(icd10_comorbid_elix)
+export(icd10_comorbid_hcc)
 export(icd10_comorbid_quan_deyo)
 export(icd10_comorbid_quan_elix)
 export(icd10_filter_invalid)
@@ -208,6 +209,7 @@ export(icd9WideToLong)
 export(icd9_comorbid)
 export(icd9_comorbid_ahrq)
 export(icd9_comorbid_elix)
+export(icd9_comorbid_hcc)
 export(icd9_comorbid_quan_deyo)
 export(icd9_comorbid_quan_elix)
 export(icd9_filter_invalid)
@@ -222,6 +224,7 @@ export(icd_comorbid)
 export(icd_comorbid_ahrq)
 export(icd_comorbid_df_to_mat)
 export(icd_comorbid_elix)
+export(icd_comorbid_hcc)
 export(icd_comorbid_mat_to_df)
 export(icd_comorbid_quan_deyo)
 export(icd_comorbid_quan_elix)

R/comorbid.R

@@ -89,6 +89,13 @@ icd9PoaChoices <- icd_poa_choices
 #'   pts <- icd_long_data(visit_name = c("2", "1", "2", "3", "3"),
 #'                    icd9 = c("39891", "40110", "09322", "41514", "39891"))
 #'   icd_comorbid(pts, icd9_map_ahrq, short_code = TRUE) # visit_name is now sorted
+#'   pts <- icd_long_data(
+#'              visit_name = c("1", "2", "3", "4", "4"),
+#'              icd_name = c("20084", "1742", "30410", "41514", "95893"),
+#'              date = as.Date(c("2011-01-01", "2011-01-02", "2011-01-03",
+#'                "2011-01-04", "2011-01-04")))
+#'   pt_hccs <- icd_comorbid_hcc(pts, date_name = "date")
+
 #' @export
 icd_comorbid <- function(x, map, ...) {
   ver <- icd_guess_version.character(map[[1]])
@@ -588,6 +595,152 @@ icd10_comorbid_quan_deyo <- function(x, ..., abbrev_names = TRUE, hierarchy = TR
   apply_hier_quan_deyo(cbd, abbrev_names = abbrev_names, hierarchy = hierarchy)
 }
 
+#' @rdname icd_comorbid
+#' @export
+icd9_comorbid_hcc <- function(x,
+                              date_name = "date",
+                              visit_name = NULL,
+                              icd_name = NULL
+                              ) {
+  assert_data_frame(x, min.cols = 3, col.names = "unique")
+  assert(checkString(visit_name), checkNull(visit_name))
+  assert(checkString(icd_name), checkNull(icd_name))
+  visit_name <- get_visit_name(x, visit_name)
+  icd_name <- get_icd_name(x, icd_name)
+  assert_string(date_name)
+  assert_string(visit_name)
+  assert_string(icd_name)
+
+  icd_map <- icd::icd9_map_cc
+  # Add column for year
+  x$year <- as.numeric(format(x[[date_name]], "%Y"))
+
+  # merge CCs to patient data based on ICD and year drop ICD info
+  x <- merge(x, icd::icd9_map_cc, all.x = TRUE)
+
+  # Drop missing CC and convert to numeric
+  # Not all ICDs resolve to a CC by definition
+  x <- x[!is.na(x$cc), ]
+  x$cc <- as.numeric(x$cc)
+
+  # keep id, date, and cc columns only, reorder
+  x <- x[, c(visit_name, date_name, "year", "cc")]
+
+  # Keep only unique records
+  # Multiple ICDs for a patient can resolve to same CC
+  x <- unique(x)
+
+  # Import hierarchy mappings, and duplicate the ifcc column
+  # needed for future matching
+  hierarchy <- icd::icd_map_cc_hcc
+  hierarchy$cc <- hierarchy$ifcc
+
+  # Merge hierarchy rules with patient data
+  x <- merge(x, hierarchy, all.x = TRUE)
+
+  # Create a list of dataframes that contain the CCs that will be zeroed out
+  todrop <- list()
+  for (i in 1:6) {
+    todrop[[i]] <- x[!is.na(x$ifcc), c(3, 4, 5 + i)]
+  }
+
+  # Rename all dataframes in list to same column names
+  # rbind into a single dataframe
+  todrop <- lapply(1:length(todrop), function(x) {
+    names(todrop[[x]]) <- c(visit_name, date_name, "cc")
+    return(todrop[[x]])
+    }
+  )
+  todrop <- do.call(rbind, todrop)
+
+  # Remove all NAs from CC field
+  todrop <- todrop[!is.na(todrop$cc), ]
+
+  # Set flag for all of the CCs to be dropped
+  todrop$todrop <- TRUE
+
+  # Merge drop flags with patient data
+  x <- merge(x, todrop, all.x = TRUE)
+
+  # Drop flagged patients and keep columns of interest
+  x <- x[is.na(x$todrop), ]
+  x <- x[, c(visit_name, date_name, "cc")]
+  names(x) <- c(visit_name, date_name, "hcc")
+  x
+}
+
+#' @rdname icd_comorbid
+#' @export
+icd10_comorbid_hcc <- function(x,
+                              date_name = "date",
+                              visit_name = NULL,
+                              icd_name = NULL) {
+  assert_data_frame(x, min.cols = 3, col.names = "unique")
+  assert(checkString(visit_name), checkNull(visit_name))
+  assert(checkString(icd_name), checkNull(icd_name))
+  visit_name <- get_visit_name(x, visit_name)
+  icd_name <- get_icd_name(x, icd_name)
+  assert_string(date_name)
+  assert_string(visit_name)
+  assert_string(icd_name)
+
+  # Add column for year
+  x$year <- as.numeric(format(x[[date_name]], "%Y"))
+
+  # merge CCs to patient data based on ICD and year drop ICD info
+  x <- merge(x, icd::icd10_map_cc, all.x = TRUE)
+
+  # Drop missing CC and convert to numeric
+  # Not all ICDs resolve to a CC by definition
+  x <- x[!is.na(x$cc), ]
+  x$cc <- as.numeric(x$cc)
+
+  # keep id, date, and cc columns only, reorder
+  x <- x[, c(visit_name, date_name, "year", "cc")]
+
+  # Keep only unique records
+  # Multiple ICDs for a patient can resolve to same CC
+  x <- unique(x)
+
+  # Import hierarchy mappings, and duplicate the ifcc column
+  # needed for future matching
+  hierarchy <- icd::icd_map_cc_hcc
+  hierarchy$cc <- icd::icd_map_cc_hcc$ifcc
+
+  # Merge hierarchy rules with patient data
+  x <- merge(x, hierarchy, all.x = TRUE)
+
+  # Create a list of dataframes that contain the CCs that will be zeroed out
+  todrop <- list()
+  for (i in 1:6) {
+    todrop[[i]] <- x[!is.na(x$ifcc), c(3, 4, 5 + i)]
+  }
+
+  # Rename all dataframes in list to same column names
+  # rbind into a single dataframe
+  todrop <- lapply(1:length(todrop), function(x) {
+    names(todrop[[x]]) <- c(visit_name, date_name, "cc")
+    return(todrop[[x]])
+    }
+  )
+  todrop <- do.call(rbind, todrop)
+
+  # Remove all NAs from CC field
+  todrop <- todrop[!is.na(todrop$cc), ]
+
+  # Set flag for all of the CCs to be dropped
+  todrop$todrop <- TRUE
+
+  # Merge drop flags with patient data
+  x <- merge(x, todrop, all.x = TRUE)
+
+  # Drop flagged patients and keep columns of interest
+  x <- x[is.na(x$todrop), ]
+  x <- x[, c(visit_name, date_name, "cc")]
+  names(x) <- c(visit_name, date_name, "hcc")
+  x
+}
+
 #' @rdname icd_comorbid
 #' @export
 icd_comorbid_ahrq <- function(x, icd_name = get_icd_name(x), ...) {
@@ -636,6 +789,29 @@ icd_comorbid_quan_deyo <- function(x, icd_name = get_icd_name(x), ...) {
     stop("could not guess the ICD version using icd_name = ", icd_name)
 }
 
+#' @details Applying CMS Hierarchical Condition Categories
+#'   \code{icd_comorbid_hcc} functions differently from the rest of the
+#'   comorbidity assignment functions. This is because CMS publishes a specific
+#'   ICD to Condition Category mapping including all child ICDs. In addition,
+#'   while these mappings were the same for 2007-2012, after 2013 there are
+#'   annual versions. In addition, there is a many:many linkage between ICD and
+#'   Condition Categories (CC). Once CCs are assigned, a series of hierarchy
+#'   rules (which can also change annually) are applied to create HCCs.
+#' @rdname icd_comorbid
+#' @param date column representing,  the date each record took place, as in each
+#'   year there is a different ICD9/10 to CC mapping). This is only necessary
+#'   for HCC mappings.
+#' @export
+icd_comorbid_hcc <- function(x, icd_name = get_icd_name(x), ...) {
+  ver <- icd_guess_version.data.frame(x, icd_name = icd_name)
+  if (ver == "icd9")
+    icd9_comorbid_hcc(x, icd_name = icd_name, ...)
+  else if (ver == "icd10")
+    icd10_comorbid_hcc(x, icd_name = icd_name, ...)
+  else
+    stop("could not guess the ICD version using icd_name = ", icd_name)
+}
+
 #' Apply hierarchy and choose naming for each comorbidity map
 #'
 #' Re-used by ICD-9 and ICD-10 versions which have the same rules.

R/datadocs.R

@@ -195,8 +195,8 @@ NULL
 #' @keywords datasets
 #' @format list of character vectors, each named by co-morbidity
 #' @references Sharabiani, Mansour T. A., Paul Aylin, and Alex Bottle.
-#'   "Systematic Review of Comorbidity Indices for Administrative Data." Medical
-#'   Care December 2012 50, no. 12 (2012): 1109-18.
+#'   "Systematic Review of Comorbidity Indices for Administrative Data."
+#' 	 Medical Care December 2012 50, no. 12 (2012): 1109-18.
 #'   doi:10.1097/MLR.0b013e31825f64d0.
 #'   \url{http://www.ncbi.nlm.nih.gov/pubmed/22929993}
 #'
@@ -207,6 +207,31 @@ NULL
 #' @aliases elixComorbid icd10_map_elix
 NULL
 
+#' Medicare Hierarchical Condition Categories
+#' 
+#' Medicare HCC model was developed to use current year diagnoses and demographics
+#' predict current year healthcare expenditure. This classification has been used
+#' for additional risk adjustment models. ICD codes are first assigned to numeric
+#' Condition Categories (CCs). A hierarchy rule is then applied so that each patient
+#' is coded for only the most severe of the Condition Categories in a group. For example,
+#' if a patient has metastatic lung cancer, they will only be assigned the CC
+#' for "Metastatic Cancer and Acute Leukemia", and will not be assigned the 
+#' CC for "Lung and other Severe Cancers". Once the hierarchy rules are applied, the codes
+#' are referred to as HCCs. This mapping can change over time. It remained the same from
+#' 2007-10
+#' @docType data
+#' @keywords datasets
+#' @format \code{dataframe} with 3 columns (\code{icd_code}, \code{cc}, and \code{year})
+#' @references Pope, Gregory C., et al. 
+#'   "Diagnostic cost group hierarchical condition category models for Medicare risk adjustment." 
+#'   Health Economics Research, Inc. Waltham, MA (2000).
+#'   \url{https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Reports/Downloads/Pope_2000_2.pdf}
+#' 
+#'   Risk Adjustment, Centers for Medicare and Medicaid Services
+#'   \url{https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Risk-Adjustors.html}
+#' @name icd9_map_hcc
+#' @aliases icd10_map_hcc icd_nap_cc_hcc
+
 #' Comorbidity names
 #'
 #' These lists provide correctly sorted names of the comorbidities and their
@@ -231,7 +256,7 @@ NULL
 #'   icd_names_quan_elix icd_names_quan_elix_abbrev icd_names_quan_elix_htn
 #'   icd_names_quan_elix_htn_abbrev icd_names_ahrq icd_names_ahrq_abbrev
 #'   icd_names_ahrq_htn icd_names_ahrq_htn_abbrev icd_names_charlson
-#'   icd_names_charlson_abbrev elixComorbidNames elixComorbidNamesAbbrev
+#'   icd_names_charlson_abbrev icd_names_cc elixComorbidNames elixComorbidNamesAbbrev
 #'   elixComorbidNamesHtn elixComorbidNamesHtnAbbrev quanElixComorbidNames
 #'   quanElixComorbidNamesAbbrev quanElixComorbidNamesHtn
 #'   quanElixComorbidNamesHtnAbbrev ahrqComorbidNames ahrqComorbidNamesAbbrev