Skip to contents

Handle Duplicate AST Results

Source: R/prep_ast.R
prep_deduplicate_ast.Rd

Deduplicates AST records in two sequential steps:

  1. Exact-duplicate removal: drops rows that are fully identical across all five key columns (patient + organism + antibiotic + date + value). These are true redundant records with no ambiguity.

  2. Conflict resolution: handles the remaining cases where the same patient + organism + antibiotic + date group has different values.

Usage

prep_deduplicate_ast(
  data,
  mode = c("detect", "remove"),
  strategy = c("resistant_wins", "susceptible_wins", "first"),
  patient_col = "patient_id",
  organism_col = "organism_normalized",
  antibiotic_col = "antibiotic_normalized",
  date_col = "culture_date",
  ast_col = "ast_value_harmonized"
)

Arguments

data

Data frame with AST data in long format.

mode

Character. "detect" (flag only) or "remove" (flag then resolve). Default "detect".

strategy

Character. Resolution strategy used only when mode = "remove". One of "resistant_wins" (default), "susceptible_wins", or "first".

patient_col

Character. Patient ID column. Default "patient_id".

organism_col

Character. Organism column. Default "organism_normalized".

antibiotic_col

Character. Antibiotic column. Default "antibiotic_normalized".

date_col

Character. Culture date column. Default "culture_date".

ast_col

Character. Harmonized AST value column. Default "ast_value_harmonized".

Value

  • mode = "detect": original data with is_ast_duplicate logical column added.

  • mode = "remove": data with conflicts resolved (one row per key) and no flag column.

Details

"detect"

After exact-dedup, flags conflicting rows with is_ast_duplicate = TRUE and prints a QC summary. Returns the data with the flag column so you can inspect before deciding how to resolve.

"remove"

Runs both steps: removes exact duplicates, then applies strategy to keep one row per key combination for any remaining conflicts.