Skip to contents

Symmetric Vaccine Efficacy from Fitted Model

sve_from_model.Rd

Computes the symmetric vaccine efficacy (SVE) and associated confidence intervals from a fitted regression model (Cox, Poisson, etc.) using the exact likelihood from the model.

Usage

sve_from_model(
  model,
  effect_name,
  data = NULL,
  level = 0.95,
  method = c("profile", "tanh-wald", "wald")
)

Arguments

model

A fitted model object. Currently supports:

  • coxph objects from the survival package

  • glm objects with family poisson or binomial

effect_name

Character. Name of the effect variable in the model.

data

A data frame containing the original data used to fit the model. Required for profile likelihood method.

level

Confidence level for the interval (default is 0.95).

method

Method used to construct the confidence interval.

One of:

  • "profile" (default): Profile likelihood confidence interval using the exact likelihood from the fitted model. Requires data argument.

  • "tanh-wald": Applies a hyperbolic arctangent transform before forming the Wald interval, then transforms back. Uses the variance extracted from the model.

  • "wald": Standard Wald interval on the untransformed scale.

Value

A data frame with the following columns:

estimate

The symmetric vaccine efficacy estimate.

lower

Lower bound of the confidence interval.

upper

Upper bound of the confidence interval.

level

Confidence interval level.

method

Indicates the method used for the confidence interval.

Details

This function extracts the coefficient and variance for the effect variable from a fitted model and computes SVE. For the profile likelihood method, it refits the model with the effect coefficient constrained to values corresponding to different SVE values, using the exact likelihood from the original model.

The symmetric vaccine efficacy (SVE) from a relative effect measure is: $$ \text{SVE} = \frac{1 - \theta}{\max(1, \theta)} $$ where \(\theta = \exp(\beta)\) is the hazard ratio, relative risk, or odds ratio from the model.

Examples

if (FALSE) { # \dontrun{
library(survival)

# Cox proportional hazards model
fit_cox <- coxph(Surv(time, status) ~ vaccination + age,
                 data = sim_trial_data)
sve_from_model(fit_cox, "vaccination", data = sim_trial_data)

# Poisson regression
fit_poisson <- glm(status ~ vaccination + age,
                   family = poisson(), data = sim_trial_data)
sve_from_model(fit_poisson, "vaccination", data = sim_trial_data)

# Wald methods don't require data
sve_from_model(fit_cox, "vaccination", method = "wald")
} # }