Skip to contents

Plot HAI vs CAI Infection Distribution by Facility

Source: R/prep_eda_plots.R
plot_hai_cai_by_facility.Rd

Derives Healthcare-Associated Infection (HAI) vs Community-Associated Infection (CAI) status from dates, then produces a stacked bar chart showing unique patient counts per centre. Each bar carries count and percentage labels inside each segment, and an n= total label above.

Usage

plot_hai_cai_by_facility(
  data,
  patient_col = "PatientInformation_id",
  center_col = "center_name",
  admission_col = "admission_date",
  culture_col = "culture_date",
  hai_threshold_hours = 48,
  hai_label = "HAI",
  cai_label = "CAI",
  colours = NULL,
  base_size = 14,
  title = NULL,
  syndrome_col = NULL,
  syndrome_name = NULL
)

Arguments

data

Data frame. Long-format AMR dataset.

patient_col

Character. Patient ID column. Default "PatientInformation_id".

center_col

Character. Centre/facility column. Default "center_name".

admission_col

Character. Admission date column. Default "admission_date".

culture_col

Character. Culture/specimen date column. Default "culture_date".

hai_threshold_hours

Numeric. Hours from admission to first culture at or above which a case is classified as HAI. Default 48.

hai_label

Character. Label for HAI bars. Default "HAI".

cai_label

Character. Label for CAI bars. Default "CAI".

colours

Named character vector. Fill colours keyed by hai_label and cai_label. Default red/green.

base_size

Numeric. Base font size. Default 14.

title

Character. Custom title. Auto-generated if NULL.

syndrome_col

Character or NULL. Column name containing syndrome/infection category labels. NULL = no filter.

syndrome_name

Character or NULL. Syndrome value to retain. Requires syndrome_col.

Value

A ggplot object.

Details

Classification rule: for each patient x centre, the earliest culture date is used. If the time from admission to that culture is \(\geq\) hai_threshold_hours (default 48 h), the patient is classified as HAI; otherwise CAI. The existing type_of_infection column in the dataset is ignored – classification is derived entirely from dates.

Deduplication: one classification per patient per centre (earliest culture date).