Skip to contents

Competing risks survival analysis between exposure and response diagnoses

Source: R/survival_advanced.R
survival_analysis.Rd

Performs a competing risks survival analysis using registry-based diagnosis dates. The function supports different follow-up starting points (exposure, response, or age 50) and multiple strategies for handling diagnoses that occur before cohort entry. Results include cumulative incidence estimates and ready-made ggplot visualizations.

Usage

survival_analysis(
  exposure_diagnoses,
  response_diagnoses,
  dpop,
  start = c("DATE_EXPOSURE", "DATE_RESPONSE", "DATE_50"),
  censoring_date = as.Date("2024-12-21"),
  pre_entry_handling = c("initialize", "skip", "asis")
)

Arguments

exposure_diagnoses

A data frame containing exposure diagnoses. Must include columns ID and DATE.

response_diagnoses

A data frame containing response diagnoses. Must include columns ID and DATE.

dpop

Population-level data frame with at least the columns ID, DATE_BIRTH, DATE_DEATH, and DATE_MIGRATION.

start

Character string defining the start of follow-up. One of "DATE_EXPOSURE", "DATE_RESPONSE", or "DATE_50".

censoring_date

Date defining administrative censoring. This affects to DATE_DEATH, DATE_MIGRATION and DATE (diagnoses). (default: 2024-12-21).

pre_entry_handling

Strategy for handling diagnoses occurring before cohort entry (age 50):

initialize

Diagnosis date is set to entry date.

skip

Diagnoses before entry are ignored; first post-entry diagnosis is used.

asis

Diagnosis date is used as recorded.

Value

A named list with the following elements:

plot_days

ggplot object of cumulative incidence (days).

plot_years

ggplot object of cumulative incidence (years).

plot_mortality

ggplot object of cumulative incidence of death (years).

CR_days

cuminc object with time in days.

CR_years

cuminc object with time in years.

dmodel

Final individual-level survival data used in the main analysis.

dmortality

Final individual-level survival data used in the death analysis.

Details

The analysis workflow consists of:

  1. Constructing individual-level date data (birth, death, migration, diagnoses).

  2. Defining follow-up start based on start.

  3. Handling diagnoses occurring before cohort entry using pre_entry_handling.

  4. Computing event times (exposure, response, death, censoring).

  5. Selecting the first occurring event per individual.

  6. Estimating cumulative incidence functions using cuminc.

Time is internally calculated in days and additionally expressed in years (365.25 days). Individuals who die before follow-up start are excluded.

Censoring affects inviduals date of death, immigration date and diagnosis dates.

See also

cuminc, ggcompetingrisks