Skip to contents

CanEpiRisk

CanEpiRisk.Rmd

1. Overview

Quantifying cancer risks associated with ionizing radiation is central to radiation epidemiology and public health, especially when estimating lifetime risks across populations and exposure scenarios. CanEpiRisk provides an integrated R framework to: (i) import exposure and demographic inputs, (ii) specify risk models—such as excess relative rate (ERR) and excess absolute rate (EAR)—and (iii) compute key measures including cumulative excess risk (CER) and years of life lost (YLL). The package streamlines end-to-end risk assessment, improves reproducibility, and enables transparent comparisons across models, populations, and exposure profiles.

2. Installation 

# install.packages("devtools")
devtools::install_github("KyojiFurukawa/CanEpiRisk")
library(CanEpiRisk)

3. Data & Model Components

  • Reference data: Baseline cause-specific rates (incidence/mortality) plus all-cause mortality. Built-in objects Mortality and Incidence provide reference data for the five WHO global regions. Examples (i=Region ID, 1, 2, …, or 5):
    • Mortality[[i]]$allsolid
    • Mortality[[i]]$allcause
    • Incidence[[i]]$leukaemia (etc.)
  • Risk models: Risk model lists per site/outcome. Built-in objects LSS_mortality and LSS_incidence provide information about the models derived from mortality and incidence data in the Life Span Study (LSS) cohort for various cancer sites.
    • LSS_mortality$allsolid$L$err (linear ERR)
    • LSS_incidence$leukaemia$LQ$ear (linear-quadratic EAR)

See companion articles:

4. Typical Workflow

  1. Define exposure scenario (ages at exposure, doses, sex).
  2. Select reference data (site-specific baseline + all-cause mortality).
  3. Choose a risk model (model appropriate to outcome/site).
  4. Set options (e.g., follow-up max age, weighting between ERR vs EAR, Monte Carlo sampling, … etc.).
  5. Compute CER / YLL and summarize.

5. Quick Examples 

set.seed(100)     # for reproducibility

## Example 1: CER for all solid cancer mortality
## Region 1, female (sex = 2), 0.1 Gy at age 15, follow to age 100; LSS linear ERR
exp1 <- list(agex = 15, doseGy = 0.1, sex = 2)   # exposure scenario
ref1 <- list(
  baseline  = Mortality[[1]]$allsolid,          # site-specific baseline
  mortality = Mortality[[1]]$allcause           # all-cause mortality
)
mod1 <- LSS_mortality$allsolid$L                # risk model (linear ERR)
opt1 <- list(maxage = 100, err_wgt = 1, n_mcsamp = 10000)

cer1 <- CER(exposure = exp1, reference = ref1, riskmodel = mod1, option = opt1)
cer1_per10k <- cer1 * 10000
cer1_per10k
## Example 2: CER for leukaemia incidence
## Region 4, male (sex = 1), 100 mGy total delivered evenly from age 30–45,
## follow to age 60; LSS LQ EAR
exp2 <- list(agex = 30:44 + 0.5, doseGy = rep(0.1/15, 15), sex = 1)
ref2 <- list(
  baseline  = Incidence[[4]]$leukaemia,         # site-specific baseline
  mortality = Mortality[[4]]$allcause           # all-cause mortality rates
)
mod2 <- LSS_incidence$leukaemia$LQ              # risk model (LQ EAR)
opt2 <- list(maxage = 60, err_wgt = 0, n_mcsamp = 10000)

cer2 <- CER(exposure = exp2, reference = ref2, riskmodel = mod2, option = opt2)
cer2_per10k <- cer2 * 10000
cer2_per10k

Notes.

  • sex: commonly 1 = male, 2 = female.
  • Region indices follow the package’s WHO regional ordering.
  • maxage: Upper bound of attained age for lifetime accumulation.
  • err_wgt: a value between 0 and 1 (inclusive; 1 = pure ERR; 0 = pure EAR).
  • n_mcsamp: Monte Carlo samples (default=10,000).

Comparing Scenarios

scenario_grid <- data.frame(
  doseGy = c(0.05, 0.1, 0.2),
  agex   = c(15, 30, 45)
)

compareCER <- function(doseGy, agex) {
  exp <- list(agex = agex, doseGy = doseGy, sex = 2)
  CER(exposure = exp, reference = ref1, riskmodel = mod1, option = opt1) * 10000
}

CER_per10k <- t( mapply(compareCER, scenario_grid$doseGy, scenario_grid$agex) )
cbind( scenario_grid, CER_per10k )

6. See Also

Reference data article

Model specification article

Risk calculation article