Skip to contents

Encode Comorbidity Column for Mortality Model

Source: R/burden.R, R/yll.R
dot-encode_comorbidity_mortality.Rd

Standardises a free-text or numeric comorbidity column to a consistent coding for use as a covariate in fit_mortality_rr_logistic().

Standardises a comorbidity column to numeric, binary (0/1), or ordinal factor encoding for use in fit_mortality_rr_logistic().

Usage

.encode_comorbidity_mortality(
  data,
  comorbidity_col = "comorbidities",
  patient_id_col = "PatientInformation_id"
)

.encode_comorbidity_mortality(
  data,
  comorbidity_col = "comorbidities",
  patient_id_col = "PatientInformation_id"
)

Arguments

data

Patient-level data frame.

comorbidity_col

Character. Default "comorbidities".

patient_id_col

Character. Default "PatientInformation_id".

Value

data with column comorbidity_encoded added. Attribute "comorbidity_encoding" records the strategy used: "numeric", "binary", "ordinal", or "absent".

data with comorbidity_encoded added. Attribute "comorbidity_encoding" records the strategy used.

Details

Three encoding strategies are applied in order:

  1. Numeric (Charlson / Elixhauser index already present): used as-is.

  2. Binary text ("present" / "none", etc.): recoded to integer 0 / 1.

  3. Ordinal text ("none" / "mild" / "moderate" / "severe"): recoded to an ordered factor.

Missing / unknown values are set to NA in all cases.