charlson_score function

Charlson Score Constructor

Charlson comorbidity score for Danish ICD-10 and ICD-8 data. This is a SAS-macro ASO translated to R in March of 2022

charlson_score(
  data,
  Person_ID,
  diagnosis_variable,
  time_variable = NULL,
  end_date = NULL,
  days_before_end_date = NULL,
  amount_output = "total"
)

Arguments

• data: A data.frame with at least an id variable and a variable with all diagnosis codes. The data should be in the long format (only one variable with diagnoses, but several lines per person is OK).

• Person_ID: <data-masking> An unquoted expression naming the id variable in data. This variable must always be specified.

• diagnosis_variable: <data-masking> An unquoted expression naming the diagnosis variable in data. This variable must always be specified.

• time_variable: <data-masking> An unquoted expression naming the diagnosis time variable in data if needed. The time_variable must be in a date format.

  When time_variable is specified, end_date must also be specified.

• end_date: <data-masking> An unquoted expression naming the end of time-period to search for relevant diagnoses or a single date specifying the end date. If end_date names a variable, this variable must be in a date format.

• days_before_end_date: A numeric specifying the number of days look-back from end_date to search for relevant diagnoses.

• amount_output: A character specifying whether all created index variables should be returned. When amount_output is "total" (the default) only the resulting Charlson scores are returned, otherwise all disease- specific index variables are returned.

Returns

If Person_ID and diagnosis_variable are the only specifications, the function will calculate the different versions of the Charlson score on all data available for each person, regardless of timing etc. This is OK if only relevant records are included.

Details

The charlson_score() function calculates the Charlson Charlson Comorbidity Index for each person. Three different variations on the score has been implemented:

• cc: Article from Quan et al. (Coding Algorithms for Defining Comorbidities in ICD-9 and ICD-10 Administrative Data, Med Care 2005:43: 1130-1139), the same HTR and others have used - ICD10 only
• ch: Article from Christensen et al. (Comparison of Charlson comorbidity index with SAPS and APACHE sources for prediction of mortality following intensive care, Clinical Epidemiology 2011:3 203-211), include ICD8 and ICD10 but the included diagnoses are not the same as in Quan
• cd: Article from Sundarajan et al. (New ICD-10 version of Charlson Comorbidity Index predicted in-hospital mortality, Journal of clinical Epidemiology 57 (2004) 1288-1294, include ICD10 = Charlson-Deyo including cancer

NOTE

The diagnoses to use in this function at the current state should be either ICD-8, but preferably ICD-10. The ICD-10 codes should start with two letters, where the first one is "D". Furthermore, the code should only have letters and digits (i.e. the form "DA000" not "DA00.0")

Examples

# An example dataset

test_data <- data.frame(
  IDs = c(
    1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,
    17, 18, 19, 20, 21, 22, 22, 23, 23, 24, 24, 24, 24, 24
  ),
  Diags = c(
    "DZ36", "DZ38", "DZ40", "DZ42", "DC20", "DI252",
    "DP290", "DI71", "DH340", "DG30", "DJ40", "DM353",
    "DK26", "DK700", "DK711", "DE106", "DE112", "DG82",
    "DZ940", "DC80", "DB20", "DK74", "DK704", "DE101",
    "DE102", "DB20", "DK74", "DK704", "DE101", "DE102"
 ),
 time = as.Date(c(
   "2001-01-30", "2004-05-20", "2007-01-02", "2013-12-01",
   "2017-04-30", "2001-01-30", "2004-05-20", "2007-01-02",
   "2013-12-01", "2017-04-30", "2001-01-30", "2004-05-20",
   "2007-01-02", "2013-12-01", "2017-04-30", "2001-01-30",
   "2004-05-20", "2007-01-02", "2013-12-01", "2017-04-30",
   "2001-01-30", "2004-05-20", "2007-01-02", "2013-12-01",
   "2017-04-30", "2001-01-30", "2004-05-20", "2007-01-02",
   "2013-12-01", "2017-04-30"
 )),
 match_date = as.Date(c(
   "2001-10-15", "2005-10-15", "2011-10-15", "2021-10-15",
   "2021-10-15", "2001-10-15", "2005-10-15", "2011-10-15",
   "2021-10-15", "2021-10-15", "2001-10-15", "2005-10-15",
   "2011-10-15", "2021-10-15", "2021-10-15", "2001-10-15",
   "2005-10-15", "2011-10-15", "2021-10-15", "2021-10-15",
   "2001-10-15", "2005-10-15", "2011-10-15", "2021-10-15",
   "2021-10-15", "2001-10-15", "2005-10-15", "2011-10-15",
   "2021-10-15", "2021-10-15"
 ))
)

# Minimal example
charlson_score(
  data = test_data,
  Person_ID = IDs,
  diagnosis_variable = Diags
)

# Minimal example with all index diagnosis variables
charlson_score(
  data = test_data,
  Person_ID = IDs,
  diagnosis_variable = Diags,
  amount_output = "all"
)

# Imposing uniform date restrictions to diagnoses
charlson_score(
  data = test_data,
  Person_ID = IDs,
  diagnosis_variable = Diags,
  time_variable = time,
  end_date = as.Date("2012-01-01")
)

# Imposing differing date restriction to diagnoses
charlson_score(
  data = test_data,
  Person_ID = IDs,
  diagnosis_variable = Diags,
  time_variable = time,
  end_date = match_date
)

# Imposing both a start and end to the lookup period for
# relevant diagnoses
charlson_score(
  data = test_data,
  Person_ID = IDs,
  diagnosis_variable = Diags,
  time_variable = time,
  end_date = match_date,
  days_before_end_date = 365.25
)

Author(s)

ASO & ADLS