Convert a list of icd9 codes associated with poor prognosis cancers to icd10 codes using the 2016 General Equivalency Mappings from CMS

convert_icd9to10_cancer_dx.r

# Convert ICD 9 to ICD 10 codes based on the CMS General Equivalency Mappings 2016
# https://www.cms.gov/Medicare/Coding/ICD10/2016-ICD-10-CM-and-GEMs.html

library(data.table)
library(icd)

gem <- fread('~/Desktop/SCIP - Study of Critical Illness Pathways/2018-redo/final_code/2016-General-Equivalence-Mappings/2016_I9gem.txt',
            colClasses = rep('character',3))
setnames(gem, c('icd9', 'icd10', 'flags_composite'))

# these flags defined in the enclosed documentation
gem[, flag_approximate := substr(flags_composite, 1, 1) == '1'] # 0/1
gem[, flag_nomap := substr(flags_composite, 2, 2) == '1'] # 0/1
gem[, flag_combination := substr(flags_composite, 3, 3) == '1'] # 0/1
gem[, scenario_code := substr(flags_composite, 4, 4)] # 0-9
gem[, choice_list_code := substr(flags_composite, 5, 5)] # 0-9

#
#Combination flag—attribute in a GEM that when turned on indicates that more than one code in the target system is required to satisfy the full equivalent meaning of a code in the source system
#Combination entry—an entry in a GEM for which a code in the source system must be linked to more than one code option in the target system to be a valid entry

# See Obermeyer Z, Powers BW, Makar M, Keating NL, Cutler DM. Physician characteristics strongly predict patient enrollment in hospice. Health Aff (Millwood). 2015;34(6), Supplemental Table 1:

icd9_codes <- c("1500", "1501", "1502", "1503", "1504", "1505", "1508",
  "1509", "151", "1510", "1511", "1512", "1513", "1514",
  "1515", "1516", "1518", "1519", "155", "1550", "1551",
  "1552", "1570", "1571", "1572", "1573", "1574", "1578",
  "1579", "1580", "1588", "1589", "159", "1590", "1591",
  "1598", "1599", "1620", "1622", "1623", "1624", "1625",
  "1628", "1629", "1630", "1631", "1638", "1639", "1641",
  "1642", "1643", "1648", "1649", "1650", "1658", "1659",
  "191", "1910", "1911", "1912", "1913", "1914", "1915",
  "1916", "1917", "1918", "1919", "195", "1951", "1952",
  "1953", "1958", "1973", "1974", "1975", "1976", "1977",
  "1978", "1980", "1981", "1982", "1983", "1984", "1985",
  "1986", "1987", "1988", "19881", "19882", "19889", "1990",
  "1991", "1992", "20500", "20502", "20520", "20522", "20530",
  "20532", "20580", "20582", "20590", "20592", "20780", "20800",
  "20880", "20890", "20892", "2375", "23873", "2391", "2396",
  "78951")

# Run checks:
cat('There are', length(icd9_codes), 'codes available for lookup.')
cat('There are', sum(icd9_codes %in% gem$icd9), ' codes found in the GEM file.')
cat('Of the',sum(icd9_codes %in% gem$icd9),'matches,',gem[icd9 %in% icd9_codes, sum(!flag_nomap),by=icd9][, sum(V1>=1)],'have mappings.')
cat('Of the',sum(icd9_codes %in% gem$icd9),'matches,',gem[icd9 %in% icd9_codes, sum(!flag_nomap),by=icd9][, sum(V1>1)],'map to multiple codes.')
cat('Of the', sum(icd9_codes %in% gem$icd9), 'matches,', gem[icd9 %in% icd9_codes, sum(flag_combination),by=icd9][,sum(V1>0)],'require a combination of ICD10 codes to match.')

# The final mapping to ICD10 includes:
final_list_icd10 <- gem[icd9 %in% icd9_codes]$icd10

# and a manual review:
knitr::kable(data.table(icd10 = final_list_icd10,
           description = sapply(final_list_icd10, explain_code)))