## ----setup, include = FALSE---------------------------------------------------
knitr::opts_chunk$set(
  collapse = TRUE,
  comment  = "#>",
  eval     = FALSE
)

## ----load-data----------------------------------------------------------------
# library(ukbflow)
# 
# df <- ops_toy(n = 500)

## ----derive-missing-----------------------------------------------------------
# df <- derive_missing(df)

## ----derive-missing-unknown---------------------------------------------------
# df <- derive_missing(df, action = "unknown")

## ----derive-missing-extra-----------------------------------------------------
# df <- derive_missing(df, extra_labels = "Not applicable")

## ----derive-covariate---------------------------------------------------------
# df <- derive_covariate(
#   df,
#   as_factor = c(
#     "p31",        # sex
#     "p20116_i0",  # smoking_status_i0
#     "p1558_i0"    # alcohol_intake_frequency_i0
#   ),
#   factor_levels = list(
#     p20116_i0 = c("Never", "Previous", "Current")
#   )
# )

## ----derive-cut---------------------------------------------------------------
# df <- derive_cut(
#   df,
#   col    = "p21001_i0",                              # body_mass_index_bmi_i0
#   n      = 4,
#   breaks = c(18.5, 25, 30),
#   labels = c("Underweight", "Normal", "Overweight", "Obese"),
#   name   = "bmi_cat"
# )
# 
# df <- derive_cut(
#   df,
#   col    = "p22189",                                 # townsend_deprivation_index_at_recruitment
#   n      = 4,
#   labels = c("Q1 (least deprived)", "Q2", "Q3", "Q4 (most deprived)"),
#   name   = "tdi_cat"
# )

## ----derive-selfreport--------------------------------------------------------
# # Non-cancer: type 2 diabetes (field 20002)
# df <- derive_selfreport(df,
#   name  = "dm",
#   regex = "type 2 diabetes"
# )

## ----derive-selfreport-cancer-------------------------------------------------
# # Cancer: lung cancer (field 20001)
# df <- derive_selfreport(df,
#   name  = "lung_cancer",
#   regex = "lung cancer",
#   field = "cancer"
# )

## ----derive-hes---------------------------------------------------------------
# # Prefix match: codes starting with "I10" (hypertension)
# df <- derive_hes(df, name = "htn", icd10 = "I10")
# 
# # Exact match
# df <- derive_hes(df, name = "dm_hes", icd10 = "E11", match = "exact")
# 
# # Regex: E10 and E11 simultaneously
# df <- derive_hes(df, name = "dm_broad", icd10 = "^E1[01]", match = "regex")

## ----derive-fo----------------------------------------------------------------
# # ops_toy includes p131742 as a representative First Occurrence column
# df <- derive_first_occurrence(df, name = "htn", field = 131742L, col = "p131742")

## ----derive-cancer------------------------------------------------------------
# # ICD-10 only
# df <- derive_cancer_registry(df,
#   name  = "skin_cancer",
#   icd10 = "^C44"
# )
# 
# # With histology and behaviour filters
# df <- derive_cancer_registry(df,
#   name      = "scc",
#   icd10     = "^C44",
#   histology = c(8070L, 8071L, 8072L),
#   behaviour = 3L                        # 3 = malignant
# )

## ----derive-death-------------------------------------------------------------
# df <- derive_death_registry(df, name = "mi",   icd10 = "I21")
# df <- derive_death_registry(df, name = "dm",   icd10 = "E11")
# df <- derive_death_registry(df, name = "lung", icd10 = "C34")

## ----derive-icd10-------------------------------------------------------------
# # Non-cancer disease: HES + death + First Occurrence
# df <- derive_icd10(df,
#   name   = "dm",
#   icd10  = "E11",
#   source = c("hes", "death", "first_occurrence"),
#   fo_col = "p131742"
# )
# 
# # Cancer outcome: cancer registry
# df <- derive_icd10(df,
#   name      = "lung",
#   icd10     = "^C3[34]",
#   match     = "regex",
#   source    = "cancer_registry",
#   behaviour = 3L
# )

## ----derive-case--------------------------------------------------------------
# df <- derive_case(df, name = "dm")