CLRtools (version 0.1.0)

coeff.OR: Compute Odds Ratios for Logistic and Conditional Logistic Regression Models

Description

Calculates odds ratios and confidence intervals for one or more coefficients in a fitted logistic regression model (from glm) or conditional logistic regression model (from clogit). Also supports scaling coefficients to meaningful units (e.g., per 5 years, 5 kg, or 5 cm) for clearer interpretation.

Usage

coeff.OR(model, variable = NULL, c = 1, confidence.level = 0.95)

Value

A named list containing:

odds.ratio

Exponentiated coefficients (odds ratios).

lower.limit

Lower bounds of the confidence intervals.

upper.limit

Upper bounds of the confidence intervals.

Arguments

model

A fitted model object from glm() (with family = binomial) or survival::clogit().

variable

Optional character vector specifying the variable(s) for which to compute odds ratios. If NULL (default), checks all coefficients

c

A numeric value specifying the increment(s) for the variables. Either a single (unnamed) numeric value applied to all selected variables, or a named numeric vector specifying a separate increment for each. The names of c must match the coefficient names in the model. Defaults to 1.

confidence.level

Confidence level for the confidence interval. Defaults to 0.95.

Details

Supports both logistic regression models fit with glm() and conditional logistic regression models fit with clogit() from the survival package. Confidence intervals are calculated on the log-odds scale and transformed back using exp().

Examples

# Example from Hosmer et al., 2013
# Applied Logistic Regression (3rd ed.), Chapter 5, Table 5.18

# Recode 'raterisk' into a binary categorical variable 'raterisk_cat'
glow500<-dplyr::mutate(
  glow500,
  raterisk_cat = dplyr::case_when(
    raterisk %in% c('Less', 'Same') ~ 'C1',
    raterisk == 'Greater' ~ 'C2'))

 model.int <- glm(
   fracture ~ age + height + priorfrac + momfrac +
     armassist + raterisk_cat + age*priorfrac + momfrac*armassist,
  family = binomial, data = glow500)

 # Specify variables and interpretation units for OR computation
 var.or<-c('raterisk_catC2','height')
 units.var<-c('raterisk_catC2'=1,'height'=5)

 # Calculate and interpret adjusted odds ratios
 coeff.OR(model.int,variable = var.or, c = units.var)