Skip to contents

Compute Polymicrobial Weights

Source: R/weight.R
compute_polymicrobial_weight.Rd

Calculates proportional weights for polymicrobial infections. Monomicrobial patients always receive weight = 1.0. For polymicrobial patients, weight is computed per organism within each episode using one of three methods.

Usage

compute_polymicrobial_weight(
  data,
  episode_col = "episode_id",
  organism_col = "organism_normalized",
  polymicrobial_col = "is_polymicrobial",
  method = "monomicrobial_proportion",
  weight_map = NULL,
  facility_col = NULL,
  facility_name = NULL,
  syndrome_col = NULL,
  syndrome_name = NULL
)

Arguments

data

Data frame with episode_id, is_polymicrobial (0/1), and organism columns (output of flag_polymicrobial()).

episode_col

Character. Episode ID column. Default "episode_id".

organism_col

Character. Organism column. Default "organism_normalized".

polymicrobial_col

Character. Polymicrobial flag column (0/1). Default "is_polymicrobial".

method

Character. Weighting method: "monomicrobial_proportion" (default), "equal", or "manual".

weight_map

Named numeric vector. Custom organism weights when method = "manual".

facility_col

Character or NULL. Facility column. When supplied, monomicrobial proportions are computed per-facility so each facility's local organism distribution is used as reference.

facility_name

Character or NULL. When supplied together with facility_col, data are first filtered to that facility.

syndrome_col

Character or NULL. Syndrome column. When supplied, monomicrobial proportions are computed per-syndrome.

syndrome_name

Character or NULL. When supplied together with syndrome_col, data are first filtered to that syndrome.

Value

Data frame with polymicrobial_weight column (range 0-1), plus weight_method and weight_confidence audit columns. episode_id is removed (internal use only).

Details

When facility_col or syndrome_col are supplied, the monomicrobial reference pool (method "monomicrobial_proportion") is computed within each stratum, so the reference distribution is local to each facility / syndrome rather than global.

Examples

if (FALSE) { # \dontrun{
# Global monomicrobial proportion weights
data_weighted <- compute_polymicrobial_weight(data_flagged)

# Per-facility reference distribution
data_weighted <- compute_polymicrobial_weight(
  data_flagged,
  facility_col = "center_name"
)

# Filter to one facility + one syndrome, then weight
data_weighted <- compute_polymicrobial_weight(
  data_flagged,
  facility_col  = "center_name",
  facility_name = "PGIMER",
  syndrome_col  = "infectious_syndrome",
  syndrome_name = "Bloodstream infection"
)
} # }