Skip to contents

Fit mortality model and derive adjusted relative risk of death

Source: R/daly_rr_and_los.R
daly_fit_mortality_rr.Rd

Fits a pathogen-class specific mixed-effects logistic regression model:

Usage

daly_fit_mortality_rr(
  data,
  patient_id_col = "PatientInformation_id",
  facility_col = "center_name",
  organism_col = "organism_name",
  syndrome_col = "syndrome",
  infection_type_col = "type_of_infection",
  antibiotic_class_col = "antibiotic_class",
  antibiotic_name_col = "antibiotic_name",
  antibiotic_value_col = "antibiotic_value",
  unit_type_col = "unit_type",
  date_admission_col = "date_of_admission",
  date_culture_col = "date_of_first_positive_culture",
  final_outcome_col = "final_outcome",
  final_outcome_date_col = "final_outcome_date",
  age_col = "Age",
  sex_col = "Gender",
  comorbidity_col = NULL,
  death_value = "Death",
  syndrome_name = NULL,
  organism_name = NULL,
  hai_threshold_hours = 48,
  icu_values = c("ICU", "Intensive Care", "Critical Care", "PICU", "NICU"),
  phi_threshold = 0.7,
  min_n = 20L,
  min_deaths = 10L,
  use_random_intercept = TRUE
)

Arguments

data

Data frame.

patient_id_col

Character. Patient identifier column.

facility_col

Character or NULL. Facility / centre column. When NULL, no hospital random effect is used and a single pooled RR is returned using standard logistic regression (hospital column will be NA). When provided, a separate logistic model is fit per hospital and the result contains one row per pathogen-class-hospital combination (hospital-specific RR).

organism_col

Character. Pathogen column.

syndrome_col

Character. Syndrome column.

infection_type_col

Character. Raw infection-type column.

antibiotic_class_col

Character. Antibiotic class column.

antibiotic_name_col

Character. Antibiotic name column.

antibiotic_value_col

Character. Antibiotic susceptibility column.

unit_type_col

Character or NULL. ICU / ward location column.

date_admission_col

Character. Admission date column.

date_culture_col

Character. First positive culture date column.

final_outcome_col

Character. Final outcome column.

final_outcome_date_col

Character. Final outcome date column.

age_col

Character. Age column.

sex_col

Character. Sex column.

comorbidity_col

Character or NULL. Comorbidity column.

death_value

Character. Value indicating death.

syndrome_name

Character or NULL. Restrict to one syndrome.

organism_name

Character vector or NULL. Restrict to these pathogens.

hai_threshold_hours

Numeric. HAI derivation threshold.

icu_values

Character vector. Values treated as ICU.

phi_threshold

Numeric. HAI / ICU collinearity warning threshold.

min_n

Integer. Minimum patients per fitted model.

min_deaths

Integer. Minimum deaths per fitted model.

use_random_intercept

Logical. Retained for backward compatibility; ignored when facility_col is provided (per-hospital fitting is used) or NULL (no facility information available).

Value

Data frame with hospital (NA when facility_col is NULL), adjusted OR and adjusted RR of death.

Details

Death_i ~ Bernoulli(pi_i) logit(pi_i) = beta0 + beta_kd * Resistant_id + beta_age * Age_i + beta_sex * Sex_i + beta_hai * HAI_i + beta_icu * ICU_i + beta_comorb * Comorbidity_i + beta_syn * Syndrome_i + u_facility

where u_facility is a random intercept when more than one facility is present.

The function returns both: - OR_death = exp(beta_kd) - RR_death = mean(p_resistant_cf) / mean(p_susceptible_cf)

RR_death is derived from model-based predicted probabilities under resistant and susceptible counterfactual scenarios, as required by the burden methodology.