Skip to contents

Calculate incidence rates

Source: R/calculate_incidence.R
calculate_incidence.Rd

The calculate_incidence() function calculates incidence rates based on the given diagnostic and demographic information. Incidence represents the number of new cases of a given diagnosis that exist in a population of interest at a specified point or period in time.

Usage

calculate_incidence(
  linked_data,
  type = c("cumulative", "rate"),
  id_col = "id",
  date_col = "date",
  pop_data = NULL,
  pop_col = "pop_count",
  person_time_data = NULL,
  person_time_col = NULL,
  time_p = NULL,
  grouping_vars = NULL,
  only_counts = FALSE,
  suppression = TRUE,
  suppression_threshold = 5,
  CI = TRUE,
  CI_level = 0.99,
  log_path = NULL
)

Arguments

linked_data

A data frame containing linked relevant diagnostic and demographic information. Should include only first time diagnosis, see 'curate_diag'

type

Character string. Valid options are "cumulative" or "rate".

id_col

A character string. Name of ID (unique personal identifier) column in linked_data. Default is "id".

date_col

A character string. Name of the date column in linked_data. Default is "date".

pop_data

A data frame containing corresponding population at risk information.

pop_col

A character string. Name of the column containing population counts in pop_data.

person_time_data

A data frame containing corresponding person-time information.

person_time_col

A character string. Name of the column containing person-time counts in person_time_data.

time_p

A numeric value or numeric vector. Time point or time period used to calculate the incidence.

  • For time period, specify as a range. The first value of the vector is the period's lower bound, and the second element is the period's upper bound. Example: time_p = c(2010,2015)

  • For time point, single numeric value. Example: time_p = 2010

grouping_vars

Character vector (optional). Grouping variables for the aggregation of diagnostic counts (e.g. sex, education).

only_counts

Logical. Only want diagnostic counts? Default is FALSE.

  • If TRUE, return only counts.

suppression

Logical. Suppress results (counts and rates) in order to maintain statistical confidentiality? Default is TRUE.

  • If TRUE, applies primary suppression (NA) to any value under the threshold defined by suppression_threshold

suppression_threshold

Integer. Threshold used for suppression, default is set to 5 (NPR standard).

CI

Logical. Want to compute binomial confidence intervals? Default is TRUE.

  • If TRUE, add two new columns with the upper and lower CI bound with significance level defined by CI_level. Uses the Pearson-Klopper method.

CI_level

A numerical value between 0 and 1. Level for confidence intervals, default is set to 0.99

log_path

A character string. Path to the log file to append function logs. Default is NULL.

  • If NULL, a new directory /log and file is created in the current working directory.

Value

Incidence table

Examples

log_file <- tempfile()
cat("Example log file", file = log_file)

pop_df <- tibble::tibble(year = "2012-2013", population = 4500)
linked_df <- linked_df |> dplyr::rename("year"= "y_diagnosis_first")

incidence_df <- calculate_incidence(linked_df,
  type = "cumulative",
  id_col = "id",
  date_col = "year",
  pop_data = pop_df,
  pop_col = "population",
  time_p = c(2012,2013),
  only_counts = FALSE,
  suppression = TRUE,
  suppression_threshold = 10,
  log_path = log_file)
#> 
#> ! To correctly calculate incidence rates, the provided dataset should only contain new/first time diagnoses.
#> 
#> 
#> ✔ Suppressed counts using 10 threshold
#> ℹ Removed 0 cells out of 1
#> 
#> Joining with `by = join_by(year)`
#> 
#> ✔ Cumulative incidence ready